Stafford’s Conjecture 3.8
Paper & Lean proof
Manuscript updated 13 November 2026. Section 2 statement numbers follow this version; Lean names retain their release numbering. No review finding is open.
Proof map
Dependencies between the statements of the paper. Click a statement to open its card; the colour gives how its Lean statement relates to the printed one.
- Lean statement:
- exact
1. Introduction
Abstract
We prove that over a Weyl algebra over a field of characteristic zero , every nonzero admits such that , a conjecture Stafford proposed in 1978. As a consequence, every finitely generated torsion right -module is cyclic. The proof maps , by a symplectic change of variables, to a normal form and shows that the cyclic module vanishes. We first show that right multiplication by is surjective on . A comparison of lengths of localized modules over the associated graded ring then gives , so that the characteristic variety of is disjoint from the hyperplane . Finally, Gabber's involutivity theorem together with an argument about conormal directions at infinity from algebraic geometry shows that such a characteristic variety must be empty and therefore, .
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
/-- **Corollary 6.3** (Stafford 1978). Every finitely generated torsion right `A_n(k)`-module is cyclic. Proved as printed by `Section6.corollary_6_3_of` from the cited [Sta78, Thm. 3.7] (`Section6.stafford_1978_theorem_3_7`), Theorem 1.1 and the right Ore condition (`Section6.weyl_rightOreCondition`). -/ theorem corollary_6_3 (k : Type u) [Field k] [CharZero k] (n : ℕ) (T : Type v) [AddCommGroup T] [Module (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] [Module.Finite (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] (hT : Section2.IsTorsion (Stafford38Challenge.WeylAlg k n) T) : Section2.IsCyclic (Stafford38Challenge.WeylAlg k n) T := (Section6.corollary_6_3_of (Section6.weyl_rightOreCondition (k := k) n) (fun d hd => theorem_1_1 k n d hd) hT (Section6.stafford_1978_theorem_3_7 k n T hT)).1
More Lean proofs and definitions
/-- **Definition 2.1 (cyclic).** `M = mA` for some `m ∈ M`. -/ def IsCyclic : Prop := ∃ m : M, ∀ x : M, ∃ a : A, x = MulOpposite.op a • m
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
Comparison notes
The abstract states the identity and the cyclicity of torsion modules, proved as Stafford38.Paper.theorem_1_1 and Stafford38.Paper.corollary_6_3; it does not mention the strengthening of Theorem 1.1 with F a power of a linear coordinate, which needs n>=1. The universal identity also holds at rank zero. Compare both proved endpoints and their fully expanded challenge definitions.
Review notes
Theorem 1.1 — Stafford 3.8 with fixed source F = Φ_g⁻¹(x₁)^{deg_B d}
Let be a field of characteristic zero, and let
Stafford's Conjecture 3.8 in [Stafford 1978] asks whether every nonzero admits with [Stafford 1978, p. 438]. Stafford shows that this formulation implies stable cyclicity for all finitely generated torsion right Weyl-algebra modules [Stafford 1978, pp. 437–438]. We prove Stafford's conjecture:
Lean: Paper.theorem_1_1 proves Stafford38Challenge.UniversalStatement;
Paper.theorem_1_1_fixedSource proves Stafford38FixedSourceChallenge.UniversalFixedSourceStatement
with Stafford38FixedSourceChallenge.IsLinearWeylCoordinate and Stafford38FixedSourceChallenge.bernsteinDegree;
Mathlib-only statements compared by Comparator: Stafford38Challenge.PaperTheorem_1_1,
Stafford38Challenge.UniversalStatement,
Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.
/-- **Theorem 1.1.** For every field of characteristic zero `k`, every `n ≥ 0` and every nonzero `d ∈ A_n(k)`, there exists an `F ∈ A_n(k)` such that `A_n(k) = dA_n(k) + FdA_n(k)`, or equivalently, there exist `F, S, R ∈ A_n(k)` such that `1 = dR + FdS`. Moreover, if `n ≥ 1`, one may choose `F = Φ_g^{-1}(x₁)^{deg_B d}` for a linear `ω`-preserving automorphism `g`. -/ def PaperTheorem_1_1 : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (d : WeylAlg k n), d ≠ 0 → (∃ F : WeylAlg k n, (⊤ : Submodule (WeylAlg k n)ᵐᵒᵖ (WeylAlg k n)) = Submodule.span (WeylAlg k n)ᵐᵒᵖ {d} ⊔ Submodule.span (WeylAlg k n)ᵐᵒᵖ {F * d}) ∧ (∃ F S R : WeylAlg k n, (1 : WeylAlg k n) = d * R + F * d * S) ∧ (∀ hn : 1 ≤ n, ∃ g : Matrix.symplecticGroup (Fin n) k, ∃ R S : WeylAlg k n, (1 : WeylAlg k n) = d * R + phiG k n g⁻¹ (x₁ k hn) ^ Stafford38FixedSourceChallenge.bernsteinDegree k d * d * S)
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
/-- **Proof of Theorem 1.1.** `n = 0` is `step_rank_zero`. For `n ≥ 1`, Section 3 (`section_3_reduction`) either finds `deg_B d = 0` and a unit, or the normal form `d̃ = c⁻¹Φ_g(d)` with `N = deg_B d ≥ 1`, for which `Q = 0` (`theorem_6_Q_eq_zero`) gives `1 = d̃R + x₁^N d̃S` and, transported by `Φ_g^{-1}` (Remark 3.3), `1 = d(c⁻¹Φ_g^{-1}(R)) + F d(c⁻¹Φ_g^{-1}(S))` with `F = Φ_g^{-1}(x₁)^N`. -/ theorem theorem_1_1_paper : Stafford38Challenge.UniversalStatement.{u} := by intro k _ _ n d hd cases n with | zero => exact step_rank_zero d hd | succ n => rcases section_3_reduction k hd with ⟨-, h⟩ | ⟨hN, g, c, -, hform, hdeg, himp⟩ · exact h · obtain ⟨ell, R, S, -, h⟩ := himp (theorem_6_Q_eq_zero hN hform hdeg) exact ⟨ell ^ bernsteinDegree k d, R, S, h⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
More Lean proofs and definitions
/-- **Theorem 1.1, moreover.** If `n ≥ 1`, one may choose `F = ℓ^{deg_B d}` for a linear symplectic coordinate `ℓ = Φ_g^{-1}(x₁)`. -/ theorem theorem_1_1_fixedSource : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u} := Section6.theorem_1_1_fixedSource_paper
theorem theorem_1_1 : PaperTheorem_1_1 := paperTheorem_1_1_of Stafford38.Paper.theorem_1_1 Stafford38.Paper.theorem_1_1_fixedSource
/-- **Theorem 1.1, "moreover".** For `n ≥ 1` one may take `F = Φ_g^{-1}(x₁)^{deg_B d}` with a linear `ω`-preserving `g`; for `deg_B d = 0` any linear coordinate works, since then `F = 1`. -/ theorem theorem_1_1_fixedSource_paper : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u} := by intro k _ _ n d hd rw [← bernsteinDegree_eq] rcases section_3_reduction k hd with ⟨hN0, -⟩ | ⟨hN, g, c, -, hform, hdeg, himp⟩ · obtain ⟨c, -, -, hcert⟩ := step_scalar_case k hd hN0 refine ⟨_, algebraMap k _ c⁻¹, 0, phiG_symm_x₁_isLinearWeylCoordinate k 1, ?_⟩ rw [hN0, pow_zero] exact hcert · exact himp (theorem_6_Q_eq_zero hN hform hdeg)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- Theorem 1.1 as printed follows from the identity form and the fixed-source form. -/ theorem paperTheorem_1_1_of (h1 : UniversalStatement.{u}) (h2 : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u}) : PaperTheorem_1_1.{u} := by intro k _ _ n d hd obtain ⟨F, R, S, h⟩ := h1 k n d hd refine ⟨⟨F, (top_eq_sup_span_iff d F).2 ⟨R, S, h⟩⟩, ⟨F, S, R, h⟩, fun hn => ?_⟩ obtain ⟨m, rfl⟩ : ∃ m, n = m + 1 := ⟨n - 1, by omega⟩ obtain ⟨ell, R', S', hell, h'⟩ := h2 k m d hd obtain ⟨g, rfl⟩ := (isLinearWeylCoordinate_iff k m ell).1 hell exact ⟨g, R', S', h'⟩
Comparison notes
Stafford38Challenge.PaperTheorem_1_1 (Challenge.lean, Mathlib-only) states Theorem 1.1 as printed: for every field k of characteristic zero, every n ≥ 0 and every nonzero d ∈ A_n(k), A_n(k) = dA_n(k) + FdA_n(k) for some F, equivalently 1 = dR + FdS, and for n ≥ 1 one may take F = Φ_g⁻¹(x₁)^{deg_B d} with g ∈ Sp(2n, k). Solution.lean proves it as Stafford38Challenge.theorem_1_1 from Stafford38.Paper.theorem_1_1 and theorem_1_1_fixedSource, which are proved along Sections 3–6 (Section6.theorem_1_1_paper; see the card of the proof in Section 6). AlternativeSolution.lean proves the same statement by the earlier route.
Review notes
Challenge comparison — Unchanged Stafford38Challenge.UniversalStatement.
Lean: Paper.theorem_1_1 proves Stafford38Challenge.UniversalStatement;
Paper.theorem_1_1_fixedSource proves Stafford38FixedSourceChallenge.UniversalFixedSourceStatement
with Stafford38FixedSourceChallenge.IsLinearWeylCoordinate and Stafford38FixedSourceChallenge.bernsteinDegree;
Mathlib-only statements compared by Comparator: Stafford38Challenge.PaperTheorem_1_1,
Stafford38Challenge.UniversalStatement,
Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.
/-- Stafford's Conjecture 3.8 in its intended nonzero form. -/ def UniversalStatement : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (d : WeylAlg k n), d ≠ 0 → Stafford.Reduction.Stafford38 d
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
More Lean proofs and definitions
theorem universalStatement : UniversalStatement := Stafford38.Paper.theorem_1_1
theorem universalStatement : UniversalStatement := by intro k _ _ n d hd exact Stafford38.universalStatement (k := k) n d hd
Comparison notes
The compared Palomar statement Stafford38Challenge.UniversalStatement is the identity form 1 = dR + FdS of Theorem 1.1. Solution.lean proves it as Stafford38Challenge.universalStatement := Stafford38.Paper.theorem_1_1, the proof along Sections 3–6; AlternativeSolution.lean proves the same unchanged statement by the earlier route. The comparison checks the statement only; it makes no claim that the two proofs are the same.
Review notes
Challenge comparison — Unchanged Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.
Lean: Paper.theorem_1_1 proves Stafford38Challenge.UniversalStatement;
Paper.theorem_1_1_fixedSource proves Stafford38FixedSourceChallenge.UniversalFixedSourceStatement
with Stafford38FixedSourceChallenge.IsLinearWeylCoordinate and Stafford38FixedSourceChallenge.bernsteinDegree;
Mathlib-only statements compared by Comparator: Stafford38Challenge.PaperTheorem_1_1,
Stafford38Challenge.UniversalStatement,
Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.
/-- The exact fixed-source challenge, using the intrinsic degree. -/ abbrev UniversalFixedSourceStatement : Prop := UniversalFixedSourceStatementWithDegree.{u} bernsteinDegree
/-- **Theorem 1.1, moreover.** If `n ≥ 1`, one may choose `F = ℓ^{deg_B d}` for a linear symplectic coordinate `ℓ = Φ_g^{-1}(x₁)`. -/ theorem theorem_1_1_fixedSource : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u} := Section6.theorem_1_1_fixedSource_paper
More Lean proofs and definitions
theorem universalFixedSourceStatement : UniversalFixedSourceStatement := Stafford38.Paper.theorem_1_1_fixedSource
theorem universalFixedSourceStatement : UniversalFixedSourceStatement := by intro k _ _ n d hd obtain ⟨ell, R, S, hell, hcert⟩ := Stafford38.universalFixedSourceStatement (k := k) n d hd have hdeg := Stafford38FixedSourceChallengeTransport.bernsteinDegree_eq k d refine ⟨ell, R, S, (Stafford38FixedSourceChallengeTransport.isLinearWeylCoordinate_iff k n ell).mpr hell, ?_⟩ rw [hdeg] exact hcert
Comparison notes
The compared Palomar statement Stafford38FixedSourceChallenge.UniversalFixedSourceStatement is the “moreover” clause of Theorem 1.1. FixedSourceSolution.lean proves it as universalFixedSourceStatement := Stafford38.Paper.theorem_1_1_fixedSource, the proof along Sections 3–6; AlternativeFixedSourceSolution.lean proves the same unchanged statement by the earlier route. The comparison checks the statement only.
Review notes
Proof sketch §1 — Proof sketch: quotient-vanishing architecture
Proof sketch: We state some preliminaries in section 2, then start in section 3 by defining a map sending any chosen of Bernstein degree to a monic normal form with order symbol using Bernstein and order filtrations of . For , we prove in sections 4-6 that the module vanishes, and find the final form of Theorem 1.1 by applying the inverse map to The main components of the proof are the following: in section 4, we show that using the extended euclidean algorithm, and through two contradictory length estimations of localized modules. In section 5, using mainly results from algebraic geometry, we show that a set fulfilling certain criteria (Zariski closed over an algebraically closed field of characteristic zero, fibre-conical, involutive, there exists a polynomial with vanishing on , and does not contain points with ) must always be empty (Theorem 5.8). Finally, we show in section 6 that by defining to be the characteristic variety of the closed-field pendant to , the results of section 4 as well as the definition of in section 3 ensure that fulfills all criteria and must thus be empty. We conclude that therefore, also itself must be zero, from which Theorem 1.1 follows directly. Lastly, we provide the proof by Stafford [Stafford 1978] that hence, every finitely generated torsion module is stably cyclic (Corollary 6.3).
No corresponding declaration.
Comparison notes
The selected review sketch and Lean proof share the quotient-vanishing route: Euler surjectivity, monic-symbol support control, filtered length vanishing, Gabber involutivity, conormal exclusion, then algebraic-closure descent. The sketch is a high-level account; it does not claim line-by-line proof correspondence.
Review notes
1.1. Related literature
Context — Related literature
Related literature
Dixmier showed that over the first Weyl algebra with having characteristic zero, any right ideal can be generated by at most two elements [Dixmier 1968]. In a seminal paper in 1978, Stafford extended this result to all and moreover showed, that any finitely generated left torsion -module is a homomorphic image of a projective left ideal, and as such can also be (stably) generated by at most two elements [Stafford 1978, Theorem 3.7]. Hillebrand and Schmale [Hillebrand 2001], as well as Leykin [Leykin 2004] give an algorithm to find the two generators of left ideals, also valid for rational coefficient algebras . Caro and Levcovitz [Caro 2010] generalize the result for both left and right ideals to formal and to convergent Laurent series, and Quadrat and Robertz [Quadrat 2014] provide a constructive proof method to calculate the two generators of torsion modules over the Weyl algebras, as well as rings with rational coefficients, formal or convergent Laurent series, and, for ordinary differential operators, with formal or locally convergent power series coefficients.In the same paper [Stafford 1978], Stafford states the Conjecture 3.8 (our Theorem 1.1), and shows that this conjecture implies that any finitely generated right torsion module can be stably generated by one element, i.e. is stably cyclic (Corollary 6.3). He shows that cyclicity is true if (since has the restricted minimum condition which makes torsion modules Artinian and he proves that Artinian modules are cyclic) and also states that the conjecture is true if is monomial, but provides, as far as the authors are aware, no proof of this. This case is also contained in our proof. For holonomic modules over “simple rings such that they do not have finite length over themselves as left modules” (which all Weyl algebras are), Leykin [Leykin 2004] proved cyclicity constructively, and attributed the proof method to Stafford. However, torsion modules need not be holonomic: for , Stafford constructed quotients of Gelfand–Kirillov dimension as counterexamples [Stafford 1985], so the question about cyclicity remained open for non-holonomic torsion modules. A paper by Bellamy in 2026 states this explicitly ([Bellamy 2026] Conjecture 3.4).
We base our proof on the Weyl algebra and on filtrations and characteristic varieties of -modules as treated in the textbooks [Coutinho 1995, Bj 1979, Hotta 2008, McConnell 2001, Schnell 2019], on the involutivity of the characteristic variety due to Gabber [Gabber 1981, Gabber 1982, Singh 2014], on commutative algebra, in particular localization, length, valuation rings, Kähler differentials and completions [Atiyah 1969, Matsumura 1986, Authors 2026], field theory and bilinear forms [Lang 2012] and spectral sequences [Weibel 1994], and on algebraic geometry, mostly based on [hartshorne].
No corresponding declaration.
Comparison notes
Review notes
1.2. Formal verification with Lean
Subsection 1.2 — Formal verification with Lean
Formal verification with Lean
Theorem 1.1 and Corollary 6.3 are proved in Lean 4 [Moura 2021] with Mathlib [Community 2020] and AlgebraicAnalysis v0.3.3 [Albert 2026]. The proofs use only Lean's three standard axioms: propositional extensionality (propext), the axiom
of choice (Classical.choice) and soundness of quotient types
(Quot.sound). Every intermediate result is proved in Lean as well,
including Gabber's involutivity theorem for cyclic -modules. The
proof in Sections 3–6 follows the formal proof
step by step. The main theorem is registered
in Palomar [Albert 2026], and the Lean sources are
archived [Albert 2026]. Tooling and the role of AI assistance
are described in Appendix A. Section 7 states further consequences proved in Lean and two generalizations, to rings of differential operators on smooth affine varieties and to simple filtered algebras.
The supplementary
material [Albert 2026] shows each formalized statement of
this paper next to its Lean version and proof. It can be browsed at
https://itpplasma.github.io/stafford38-supplementary/.
theorem theorem_1_1 : PaperTheorem_1_1 := paperTheorem_1_1_of Stafford38.Paper.theorem_1_1 Stafford38.Paper.theorem_1_1_fixedSource
theorem corollary_6_3 : PaperCorollary_6_3 := by intro k _ _ n M _ _ _ hM exact Stafford38.Paper.corollary_6_3 k n M hM
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
More Lean proofs and definitions
/-- **Corollary 6.3** (Stafford 1978). Every finitely generated torsion right `A_n(k)`-module is cyclic. Proved as printed by `Section6.corollary_6_3_of` from the cited [Sta78, Thm. 3.7] (`Section6.stafford_1978_theorem_3_7`), Theorem 1.1 and the right Ore condition (`Section6.weyl_rightOreCondition`). -/ theorem corollary_6_3 (k : Type u) [Field k] [CharZero k] (n : ℕ) (T : Type v) [AddCommGroup T] [Module (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] [Module.Finite (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] (hT : Section2.IsTorsion (Stafford38Challenge.WeylAlg k n) T) : Section2.IsCyclic (Stafford38Challenge.WeylAlg k n) T := (Section6.corollary_6_3_of (Section6.weyl_rightOreCondition (k := k) n) (fun d hd => theorem_1_1 k n d hd) hT (Section6.stafford_1978_theorem_3_7 k n T hT)).1
/-- **Definition 2.1 (cyclic).** `M = mA` for some `m ∈ M`. -/ def IsCyclic : Prop := ∃ m : M, ∀ x : M, ∃ a : A, x = MulOpposite.op a • m
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
Comparison notes
The subsection states that Theorem 1.1 and Corollary 6.3 are proved in Lean from the three standard axioms, that every intermediate result is proved as well, and that the proof in Sections 3–6 follows the formal proof step by step. Every numbered statement of Sections 2–6 has its own declaration in Stafford38.Paper (Stafford38/Paper/Section2–Section6, Main.lean), and each printed proof is a chain of step declarations, one per paragraph or claim, in the printed order; the cards list them. Solution.lean and FixedSourceSolution.lean prove the compared statements through these declarations. Some paragraphs are proved by results of the earlier formal route, which the step declarations call; the cards say where. The earlier route remains as AlternativeSolution.lean.
Review notes
2. Preliminaries
Definition 2.1 — Torsion and cyclic modules
/-- **Definition 2.1 (torsion).** Every `m ∈ M` has `md = 0` for some `0 ≠ d ∈ A`. This is the repository's `TorsionCyclicity.IsRightTorsion`. -/ abbrev IsTorsion : Prop := Stafford38.TorsionCyclicity.IsRightTorsion (A := A) (M := M)
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
/-- **Definition 2.1 (cyclic).** `M = mA` for some `m ∈ M`. -/ def IsCyclic : Prop := ∃ m : M, ∀ x : M, ∃ a : A, x = MulOpposite.op a • m
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
More Lean proofs and definitions
/-- Cyclicity is the Mathlib form `span {m} = ⊤` used in the corollary. -/ theorem isCyclic_iff_span : IsCyclic A M ↔ ∃ m : M, Submodule.span Aᵐᵒᵖ {m} = ⊤ := by constructor · rintro ⟨m, hm⟩ refine ⟨m, eq_top_iff.mpr fun x _ => ?_⟩ obtain ⟨a, rfl⟩ := hm x exact Submodule.smul_mem _ _ (Submodule.mem_span_singleton_self m) · rintro ⟨m, hm⟩ refine ⟨m, fun x => ?_⟩ have hx : x ∈ Submodule.span Aᵐᵒᵖ {m} := hm ▸ Submodule.mem_top obtain ⟨a, rfl⟩ := Submodule.mem_span_singleton.mp hx exact ⟨a.unop, rfl⟩
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
Comparison notes
Section2.IsTorsion (every m has md = 0 for some d ≠ 0; the repository's TorsionCyclicity.IsRightTorsion) and Section2.IsCyclic (M = mA) are stated for right modules over any ring A; with A = A_n(k) they are the printed definitions. isCyclic_iff_span gives the form span{m} = ⊤ used in Corollary 6.3.
Review notes
Notation 2.2 — Distinguished Weyl pair and Ore presentation
Section2.transverse,
Section2.C,
Section2.convention_2_2_relations,
Section2.convention_2_2_exists,
Section2.convention_2_2_unique/-- `C` is the coefficient ring of the last Ore stage of the iterated Ore presentation `A_n(k) = C[p; δ]`, embedded by `coordinateStageToPresented`. -/ theorem C_eq_range (n : ℕ) : C k n = (coordinateStageToPresented k n).range := by apply le_antisymm · apply Algebra.adjoin_le rintro _ (⟨w, rfl⟩ | hx) · refine ⟨normalCoefficient zeroDerivation (presentedToIterated k n w), ?_⟩ change iteratedToPresented k (n + 1) (stageEmbedding k n (presentedToIterated k n w)) = _ rw [iteratedToPresented_coefficient] exact congrArg (previousWeylEmbedding k n) (DFunLike.congr_fun (iteratedToPresented_comp_presentedToIterated k n) w) · rw [Set.mem_singleton_iff.mp hx] exact ⟨normalVariable zeroDerivation, iteratedToPresented_coordinate k n⟩ · rintro _ ⟨z, rfl⟩ obtain ⟨q, rfl⟩ := normalForm_surjective (zeroDerivation : OreDivisionDerivation (IteratedPairStage k n)) z induction q using Polynomial.induction_on' with | add q₁ q₂ h₁ h₂ => rw [normalForm_add] change coordinateStageToPresented k n (normalForm _ q₁ + normalForm _ q₂) ∈ C k n rw [map_add] exact add_mem h₁ h₂ | monomial i b => rw [normalForm_monomial] change coordinateStageToPresented k n (normalCoefficient zeroDerivation b * normalVariable zeroDerivation ^ i) ∈ C k n rw [map_mul, map_pow] have hx : coordinateStageToPresented k n (normalVariable zeroDerivation) = x₁ k n := iteratedToPresented_coordinate k n rw [hx] refine mul_mem (Algebra.subset_adjoin (Set.mem_union_left _ ?_)) (pow_mem (Algebra.subset_adjoin (Set.mem_union_right _ (Set.mem_singleton _))) i) refine ⟨iteratedToPresented k n b, ?_⟩ change _ = iteratedToPresented k (n + 1) (stageEmbedding k n b) rw [iteratedToPresented_coefficient] rfl
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2, relations.** `x₁` is central in `C`, `[p, x₁] = 1`, and the transverse algebra commutes with `x₁` and `p`. -/ theorem convention_2_2_relations (n : ℕ) : p k n * x₁ k n = x₁ k n * p k n + 1 ∧ (∀ c ∈ C k n, x₁ k n * c = c * x₁ k n) ∧ ∀ t ∈ transverse k n, t * x₁ k n = x₁ k n * t ∧ t * p k n = p k n * t := by refine ⟨presentedMomentum_mul_coordinate k n, ?_, ?_⟩ · intro c hc induction hc using Algebra.adjoin_induction with | mem x hx => rcases hx with ⟨w, rfl⟩ | hx · exact presentedCoordinate_commutes_previous k n w · rw [Set.mem_singleton_iff.mp hx] | algebraMap r => exact (Algebra.commutes r _).symm | add x y _ _ hx hy => rw [mul_add, add_mul, hx, hy] | mul x y _ _ hx hy => rw [← mul_assoc, hx, mul_assoc, hy, mul_assoc] · rintro _ ⟨w, rfl⟩ exact ⟨(presentedCoordinate_commutes_previous k n w).symm, (presentedMomentum_commutes_previous k n w).symm⟩
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
More Lean proofs and definitions
/-- **Convention 2.2.** The coordinate `x₁` of the first Weyl pair. -/ abbrev x₁ (n : ℕ) : PresentedWeyl k (n + 1) := presentedCoordinate k n
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2.** The momentum `p = ∂₁` of the first Weyl pair. -/ abbrev p (n : ℕ) : PresentedWeyl k (n + 1) := presentedMomentum k n
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2.** The transverse algebra `A_{n-1}(k) = k⟨x_i, ∂_i : i ≥ 2⟩`. -/ abbrev transverse (n : ℕ) : Subalgebra k (PresentedWeyl k (n + 1)) := (previousWeylEmbedding k n).range
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2.** `C = A_{n-1}(k)[x₁]`, the subalgebra generated by the transverse algebra and `x₁`. -/ def C (n : ℕ) : Subalgebra k (PresentedWeyl k (n + 1)) := Algebra.adjoin k ((transverse k n : Set (PresentedWeyl k (n + 1))) ∪ {x₁ k n})
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2, normal form (existence).** Every element of `A_n(k)` can be written as `∑_j p^j c_j` with `c_j ∈ C`. -/ theorem convention_2_2_exists (n : ℕ) (d : PresentedWeyl k (n + 1)) : ∃ (L : ℕ) (a : ℕ → PresentedWeyl k (n + 1)), (∀ j, a j ∈ C k n) ∧ d = ∑ j ∈ Finset.range L, p k n ^ j * a j := by let b := pairStage_rightOrePBWBasis (B := IteratedPairStage k n) let c := b.repr (presentedToIterated k (n + 1) d) let L := c.support.sup id + 1 have hsub : c.support ⊆ Finset.range L := by intro j hj exact Finset.mem_range.mpr (Nat.lt_succ_of_le (Finset.le_sup (f := id) hj)) refine ⟨L, fun j => coordinateStageToPresented k n (c j).unop, fun j => (C_eq_range k n).symm ▸ ⟨_, rfl⟩, ?_⟩ have hd : presentedToIterated k (n + 1) d = ∑ j ∈ Finset.range L, c j • rightPBWMonomial (coordinateDerivation : OreDivisionDerivation (CoordinateStage (B := IteratedPairStage k n))) j := by conv_lhs => rw [← b.linearCombination_repr (presentedToIterated k (n + 1) d)] rw [Finsupp.linearCombination_apply, Finsupp.sum_of_support_subset _ hsub _ (by simp)] apply Finset.sum_congr rfl intro j _ simp [b, pairStage_rightOrePBWBasis] rfl have hback : d = stageToPresented k n (presentedToIterated k (n + 1) d) := (DFunLike.congr_fun (iteratedToPresented_comp_presentedToIterated k (n + 1)) d).symm rw [hback, hd, map_sum] apply Finset.sum_congr rfl intro j _ exact stageToPresented_rightTerm k n j (c j).unop
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2, normal form (uniqueness).** If `∑_{j<L} p^j c_j = 0` with `c_j ∈ C`, then every `c_j = 0`. -/ theorem convention_2_2_unique (n L : ℕ) (a : ℕ → PresentedWeyl k (n + 1)) (ha : ∀ j, a j ∈ C k n) (h0 : ∑ j ∈ Finset.range L, p k n ^ j * a j = 0) : ∀ j < L, a j = 0 := by have hpre : ∀ j, ∃ z, coordinateStageToPresented k n z = a j := fun j => by have := ha j rw [C_eq_range] at this exact this choose b hb using hpre let D : OreDivisionDerivation (CoordinateStage (B := IteratedPairStage k n)) := coordinateDerivation have hsum : rightPBWCombination D (∑ j ∈ Finset.range L, Finsupp.single j (MulOpposite.op (b j))) = 0 := by rw [map_sum] simp only [rightPBWCombination_single] have himg : stageToPresented k n (∑ j ∈ Finset.range L, MulOpposite.op (b j) • rightPBWMonomial D j) = 0 := by rw [map_sum, ← h0] apply Finset.sum_congr rfl intro j _ rw [stageToPresented_rightTerm, hb] have := congrArg (presentedToIterated k (n + 1)) himg rwa [presentedToIterated_stageToPresented, map_zero] at this have hzero := (rightPBWCombination_injective D) (hsum.trans (map_zero _).symm) intro j hj have hj' := congrArg (fun f : ℕ →₀ (CoordinateStage (B := IteratedPairStage k n))ᵐᵒᵖ => f j) hzero simp only [Finsupp.coe_finsetSum, Finset.sum_apply, Finsupp.single_apply, Finset.sum_ite_eq', Finset.mem_range, ite_eq_left hj, Finsupp.coe_zero, Pi.zero_apply] at hj' rw [← hb, MulOpposite.op_injective (hj'.trans MulOpposite.op_zero.symm), map_zero]
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
Comparison notes
Lean states the convention for A_{n+1}(k), the first pair being index 0, so the printed n ≥ 1 is built in. x₁, p, the transverse algebra A_{n-1}(k) and C = A_{n-1}(k)[x₁] are defined as printed; C_eq_range identifies C with the coefficient ring of the Ore presentation A_n(k) = C[p; δ]. convention_2_2_relations gives x₁ central in C, [p, x₁] = 1 and the transverse algebra commuting with x₁ and p. The normal form Σ_j p^j c_j, c_j ∈ C, which the paper cites [Coutinho, Prop. 1.2.1], is proved: existence (convention_2_2_exists) and uniqueness (convention_2_2_unique).
Review notes
Definition 2.3 — Filtrations; Bernstein and order filtrations
In the proof, we use two filtrations of (for more details, see e.g. [Coutinho 1995, Chapter 7])
- The Bernstein filtration assigns degree to every and every . The Bernstein degree of is written .
- The order filtration assigns degree to every and degree to every ; is spanned by the monomials with .
Section2.IsFiltration,
Section2.gr,
Section2.filtrationDegree,
Section2.symbol,
Section2.bernsteinFiltration,
Section2.orderFiltration,
Section2.bernsteinDegree,
Section2.grEquivSymbolRing
For both the order filtration and the Bernstein filtration, the associated graded algebra is simply the polynomial ring in variables with being the symbol of [Coutinho 1995, Th. 7.3.1]. In particular, is commutative [Schnell 2019, Prop. 2.10].
Good filtrations are treated systematically for example in [Bj 1979, Hotta 2008].
/-- **`gr A_n(k) = k[x, ξ]`** for the Bernstein and the order filtration (any weight): `⊕_j F_j/F_{j-1}` is identified with the polynomial ring by summing the homogeneous symbols. Multiplication of classes is the product of symbols, `σ_{i+j}(ab) = σ_i(a)σ_j(b)` (`WeylLeadingSymbol.presentedPrincipalComponent_mul_bernstein`, `presentedPrincipalComponent_mul_order`), so `gr A_n(k)` is commutative. -/ def grEquivSymbolRing {n : ℕ} (w : PhaseVar n → ℕ) : gr (presentedWeightPiece k w) ≃ₗ[k] SymbolRing k n := by letI := MvPolynomial.weightedDecomposition k w exact (DFinsupp.mapRange.linearEquiv fun j => grPieceEquiv k w j).trans (DirectSum.decomposeLinearEquiv (MvPolynomial.weightedHomogeneousSubmodule k w)).symm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- **Definition 2.3 (filtration).** `F_0 ⊆ F_1 ⊆ … ⊆ A` linear subspaces with `F_i F_j ⊆ F_{i+j}` and `A = ⋃_j F_j`. -/ structure IsFiltration (F : ℕ → Submodule k A) : Prop where mono : Monotone F mul_mem : ∀ {i j : ℕ} {a b : A}, a ∈ F i → b ∈ F j → a * b ∈ F (i + j) exhaustive : ∀ a : A, ∃ j, a ∈ F j
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable {k : Type u} [Field k] {A : Type v} [Ring A] [Algebra k A]
/-- **Definition 2.3 (associated graded).** `gr^F = ⊕_{j ≥ 0} F_j/F_{j-1}` as a `k`-module, for a filtration of an algebra or (Definition 2.4) of a module. -/ abbrev gr (F : ℕ → Submodule k M) := DirectSum ℕ (grPiece F)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable {k : Type u} [Field k] {M : Type v} [AddCommGroup M] [Module k M]
/-- **Definition 2.3 (filtration degree).** `deg_F a` is the smallest `j` with `a ∈ F_j`. -/ def filtrationDegree (F : ℕ → Submodule k A) (a : A) : ℕ := sInf {j | a ∈ F j}
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable {k : Type u} [Field k] {A : Type v} [Ring A] [Algebra k A]
/-- **Definition 2.3 (symbol).** `σ_j(a)` is the class of `a` in `F_j/F_{j-1}` for `j = deg_F a`. -/ def symbol {F : ℕ → Submodule k A} (hF : IsFiltration F) (a : A) : grPiece F (filtrationDegree F a) := Submodule.Quotient.mk ⟨a, mem_filtrationDegree hF a⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable {k : Type u} [Field k] {A : Type v} [Ring A] [Algebra k A]
/-- **The Bernstein filtration `B_•`**: degree `1` on every `x_i` and `∂_i`. -/ abbrev bernsteinFiltration (n : ℕ) : ℕ → Submodule k (PresentedWeyl k n) := bernsteinPiece k n
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **The order filtration `F_•`**: degree `0` on `x_i`, `1` on `∂_i`; `F_m` is spanned by the `x^α ∂^β` with `|β| ≤ m`. -/ abbrev orderFiltration (n : ℕ) : ℕ → Submodule k (PresentedWeyl k n) := orderPiece k n
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **The Bernstein degree** `deg_B d`, the filtration degree for `B_•`. -/ def bernsteinDegree {n : ℕ} (d : PresentedWeyl k n) : ℕ := filtrationDegree (bernsteinFiltration k n) d
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
theorem isFiltration_bernstein (n : ℕ) : IsFiltration (bernsteinFiltration k n) := ⟨fun _ _ h => presentedWeightPiece_mono k _ h, mul_mem_bernsteinPiece k, exists_mem_bernsteinPiece k⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
theorem isFiltration_order (n : ℕ) : IsFiltration (orderFiltration k n) := ⟨fun _ _ h => presentedWeightPiece_mono k _ h, mul_mem_orderPiece k, exists_mem_orderPiece k⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- The printed Bernstein degree is the repository's intrinsic `Stafford38FixedSourceChallenge.bernsteinDegree`. -/ theorem bernsteinDegree_eq {n : ℕ} (d : PresentedWeyl k n) : bernsteinDegree k d = Stafford38FixedSourceChallenge.bernsteinDegree k d := by simp only [bernsteinDegree, filtrationDegree, Stafford38FixedSourceChallenge.bernsteinDegree, Stafford38FixedSourceChallengeTransport.bernsteinPiece_eq]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
IsFiltration, gr, filtrationDegree and symbol are the printed notions with filtrations indexed by ℕ: F_{-1} = 0 is lowerPiece F 0 = ⊥, and gr = ⊕_j F_j/F_{j-1} is the direct sum of k-modules. filtrationDegree is defined for every a (the least j with a ∈ F_j); the paper defines it for a ≠ 0. The Bernstein and order filtrations are spans of ordered PBW words of bounded weight (isFiltration_bernstein, isFiltration_order), and bernsteinDegree is the filtration degree for B_• (bernsteinDegree_eq: the degree of the Palomar statement). The cited identification gr A_n(k) = k[x, ξ] [Coutinho, Th. 7.3.1] is proved for both filtrations (grEquivSymbolRing), with σ_{i+j}(ab) = σ_i(a)σ_j(b), so gr A_n(k) is commutative.
Review notes
Definition 2.4 — Good filtrations
For a finitely generated right -module with a good order filtration , the graded module is a finitely generated -module.
/-- **Definition 2.4, example.** For `Q = A_n(k)/I`, the filtration `Q_m = qF_m` induced by the generator `q = [1]` is good. -/ theorem qFilt_isGoodFiltration (I : RightIdeal (PresentedWeyl k n)) : IsGoodFiltration k n (qFilt k I) := ⟨cyclicFiltration_order_isCompatible k I, qFilt_good k I⟩
More Lean proofs and definitions
/-- **Definition 2.4 (good filtration).** A filtration `M_•` of a right `A_n(k)`-module `M`, compatible with the order filtration `F_•`, is good if `gr M = ⊕_i M_i/M_{i-1}` is finitely generated over `gr^F A_n = k[x, ξ]`. -/ structure IsGoodFiltration (Γ : ℕ → Submodule k M) : Prop where compatible : IsCompatibleFiltration (orderFiltration k n) Γ finite : Module.Finite (SymbolRing k n) (GrModule Γ)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Paper.Support.GabberSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.EulerSurjectivitySource ↗universe u wSource ↗
variable (k : Type u) [Field k] (n : ℕ)Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl k n)ᵐᵒᵖ M] [Module k M] [IsScalarTower k (PresentedWeyl k n)ᵐᵒᵖ M]
/-- **Definition 2.4 (compatible filtration).** An exhaustive increasing chain `M_0 ⊆ M_1 ⊆ …` of `k`-subspaces (`M_i = 0` for `i < 0`) with `M_i F_j ⊆ M_{i+j}`. -/ structure IsCompatibleFiltration (F : ℕ → Submodule k A) (Mf : ℕ → Submodule k M) : Prop where mono : Monotone Mf smul_mem : ∀ {i j : ℕ} {m : M} {a : A}, m ∈ Mf i → a ∈ F j → MulOpposite.op a • m ∈ Mf (i + j) exhaustive : ∀ m : M, ∃ i, m ∈ Mf i
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]Source ↗
variable {k}Source ↗variable {A : Type v} [Ring A] [Algebra k A]Source ↗
variable {M : Type w} [AddCommGroup M] [Module Aᵐᵒᵖ M] [Module k M] [IsScalarTower k Aᵐᵒᵖ M]
/-- `gr M = ⊕_i M_i/M_{i-1}` with its `K[x, ξ]`-module structure (below). -/ def GrModule (Γ : ℕ → Submodule K M) : Type w := gr Γ
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.EulerSurjectivitySource ↗universe u v w w' w₁Source ↗
variable {K : Type u} [Field K] {n : ℕ}Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (PresentedWeyl K n)ᵐᵒᵖ M]
/-- The filtration `Q_m = qF_m` of `Q = A_n/I`. -/ abbrev qFilt : ℕ → Submodule K (RightQuotient I) := cyclicFiltration (orderFiltration K n) (qClass K I)
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicFilteredQuotientGradedSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.EulerSurjectivitySource ↗universe u w w' w₁Source ↗
variable (K : Type u) [Field K] {n : ℕ} (I : RightIdeal (PresentedWeyl K n))
/-- The filtration `M_i = qF_i` induced by an element `q`. -/ def cyclicFiltration (F : ℕ → Submodule k A) (q : M) (i : ℕ) : Submodule k M := (F i).map (orbitMap q)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]Source ↗
variable {k}Source ↗variable {A : Type v} [Ring A] [Algebra k A]Source ↗
variable {M : Type w} [AddCommGroup M] [Module Aᵐᵒᵖ M] [Module k M] [IsScalarTower k Aᵐᵒᵖ M]
/-- **Definition 2.4, example.** For a cyclic module `M = qA` and a filtration `F_•` of `A`, `M_i = qF_i` is a compatible filtration. -/ theorem cyclicFiltration_isCompatible {F : ℕ → Submodule k A} (hF : IsFiltration F) (q : M) (hq : ∀ x : M, ∃ a : A, x = MulOpposite.op a • q) : IsCompatibleFiltration F (cyclicFiltration F q) := by refine ⟨fun i j hij => Submodule.map_mono (f := orbitMap q) (hF.mono hij), ?_, ?_⟩ · rintro i j _ a ⟨b, hb, rfl⟩ ha refine ⟨b * a, hF.mul_mem hb ha, ?_⟩ change MulOpposite.op (b * a) • q = MulOpposite.op a • MulOpposite.op b • q rw [smul_smul, MulOpposite.op_mul] · intro x obtain ⟨a, rfl⟩ := hq x obtain ⟨i, hi⟩ := hF.exhaustive a exact ⟨i, a, hi, rfl⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]Source ↗
variable {k}Source ↗variable {A : Type v} [Ring A] [Algebra k A]Source ↗
variable {M : Type w} [AddCommGroup M] [Module Aᵐᵒᵖ M] [Module k M] [IsScalarTower k Aᵐᵒᵖ M]
/-- **Definition 2.4 for `Q = A_n/I`.** `M_i = qF_i` is a compatible filtration of `Q` (`cyclicFiltration_isCompatible`). -/ theorem cyclicFiltration_order_isCompatible {n : ℕ} (I : RightIdeal (PresentedWeyl k n)) : IsCompatibleFiltration (orderFiltration k n) (cyclicFiltration (orderFiltration k n) (qClass k I)) := cyclicFiltration_isCompatible (isFiltration_order k n) _ (qClass_generates k I)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- `Q_m = qF_m` is good: `gr Q` is generated by `q̄` (Definition 2.4). -/ theorem qFilt_good : Module.Finite (SymbolRing K n) (GrModule (qFilt K I)) := by have := grCyclic_good K I exact Module.Finite.equiv (grCyclicEquivS K I)
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicFilteredQuotientGradedSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.EulerSurjectivitySource ↗universe u w w' w₁Source ↗
variable (K : Type u) [Field K] {n : ℕ} (I : RightIdeal (PresentedWeyl K n))
/-- **Definition 2.4 for `Q = A_n/I`.** The order filtration `Q_m = qF_m` induced by `q = [1]` is the image of `F_m`; its associated graded module `gr Q = ⊕_m Q_m/Q_{m-1}` with its `k[x, ξ]`-module structure is the repository's `OrderAssociatedGradedModule`. -/ abbrev grCyclic {n : ℕ} (I : RightIdeal (PresentedWeyl k n)) := OrderAssociatedGradedModule k I
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.4, `gr Q`.** The associated graded module `⊕_m Q_m/Q_{m-1}` of the filtration `Q_m = qF_m` (Definition 2.3, `gr`) is `grCyclic`, the module on which `k[x, ξ]` acts. -/ def grCyclicEquiv {n : ℕ} (I : RightIdeal (PresentedWeyl k n)) : grCyclic k I ≃ₗ[k] gr (cyclicFiltration (orderFiltration k n) (qClass k I)) := DFinsupp.mapRange.linearEquiv fun m => (quotientMapEquiv (filteredRightQuotientEquivRightQuotient k I) _ _).trans (Submodule.Quotient.equiv _ _ (LinearEquiv.ofEq _ _ (cyclicFiltration_order_eq k I m).symm) (by rw [lowerPiece_cyclicFiltration_order_eq] ext ⟨x, hx⟩ simp [Submodule.mem_comap]))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- `gr Q` is generated over `k[x, ξ]` by `q̄`. -/ theorem grCyclic_generated {n : ℕ} (I : RightIdeal (PresentedWeyl k n)) (m : grCyclic k I) : ∃ f : SymbolRing k n, f • qBar k I = m := by rw [qBar, ← orderAssociatedGradedGenerator_eq_oneClass] exact exists_smul_orderAssociatedGradedGenerator k _ m
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.4, example: goodness.** `gr Q` is finitely generated over `gr A_n = k[x, ξ]`, being generated by `q̄`. -/ theorem grCyclic_good {n : ℕ} (I : RightIdeal (PresentedWeyl k n)) : Module.Finite (SymbolRing k n) (grCyclic k I) := by refine ⟨⟨{qBar k I}, eq_top_iff.mpr fun m _ => ?_⟩⟩ obtain ⟨f, rfl⟩ := grCyclic_generated k I m exact Submodule.smul_mem _ f (Submodule.subset_span (by simp))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
IsGoodFiltration is Definition 2.4 for the order filtration: a filtration M_• of a right A_n(k)-module M that is compatible with F_• (IsCompatibleFiltration: increasing, exhaustive, M_iF_j ⊆ M_{i+j}, and M_i = 0 for i < 0 through indexing by ℕ) and whose graded module gr M = ⊕_i M_i/M_{i−1} (Support.Gabber.GrModule) is finitely generated over gr A_n = k[x, ξ]. IsCompatibleFiltration is stated for a right module over any filtered algebra, but the gr^F A_n-module structure of gr M, and with it goodness, is formalized only for the order filtration, the only filtration the paper uses for modules. The example is qFilt_isGoodFiltration: for Q = A_n/I the filtration Q_m = qF_m (Support.Gabber.qFilt) is compatible with the order filtration (cyclicFiltration_order_isCompatible) and gr Q is generated by q̄ (Support.Gabber.qFilt_good; gr Q is grCyclic by grCyclicEquiv, generated by q̄ by grCyclic_generated and grCyclic_good).
Review notes
Definition 2.5 — Spectrum, vanishing loci and vanishing ideals
For an ideal of , its radical is . is called radical if . Since a prime ideal contains iff it contains , we have .
/-- **Definition 2.6, last sentence.** `a ∈ V_K(J)` iff `𝔪_a ∈ V(J)`. -/ theorem mem_VK_iff_closedPoint_mem_V (J : Ideal (MvPolynomial σ K)) (a : σ → K) : a ∈ VK K J ↔ closedPoint K a ∈ V J := by rw [mem_V_iff] change (∀ f ∈ J, MvPolynomial.eval a f = 0) ↔ J ≤ maxIdealAt K a exact ⟨fun h f hf => (mem_maxIdealAt_iff K a f).mpr (h f hf), fun h f hf => (mem_maxIdealAt_iff K a f).mp (h hf)⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
More Lean proofs and definitions
/-- **Definition 2.6 (vanishing locus).** `V(J) = {𝔭 ∈ Spec S : 𝔭 ⊇ J}`, the Mathlib `PrimeSpectrum.zeroLocus`. -/ abbrev V (J : Ideal S) : Set (PrimeSpectrum S) := PrimeSpectrum.zeroLocus (J : Set S)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable {S : Type u} [CommRing S]
/-- **Definition 2.6 (classical zero set).** `V_K(J) = {a ∈ K^N : f(a) = 0 ∀ f ∈ J}`. -/ def VK (J : Ideal (MvPolynomial σ K)) : Set (σ → K) := {a | ∀ f ∈ J, MvPolynomial.eval a f = 0}
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
/-- **Definition 2.6 (vanishing ideal).** `I(S) = {f : f(a) = 0 ∀ a ∈ S}`. -/ def vanishingIdealOf (T : Set (σ → K)) : Ideal (MvPolynomial σ K) := MvPolynomial.vanishingIdeal K T
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
/-- **Definition 2.6.** `𝔪_a = (z_1 - a_1, …, z_N - a_N)`. -/ def maxIdealAt (a : σ → K) : Ideal (MvPolynomial σ K) := Ideal.span (Set.range fun i => MvPolynomial.X i - MvPolynomial.C (a i))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
/-- `f ∈ 𝔪_a` iff `f(a) = 0`. -/ theorem mem_maxIdealAt_iff (a : σ → K) (f : MvPolynomial σ K) : f ∈ maxIdealAt K a ↔ MvPolynomial.eval a f = 0 := by constructor · intro hf refine Submodule.span_induction ?_ (by simp) (fun x y _ _ hx hy => by simp [hx, hy]) (fun r x _ hx => by simp [hx]) hf rintro _ ⟨i, rfl⟩ simp · intro hf let q := Ideal.Quotient.mk (maxIdealAt K a) have hred : ∀ g : MvPolynomial σ K, q g = q (MvPolynomial.C (MvPolynomial.eval a g)) := by intro g induction g using MvPolynomial.induction_on with | C c => simp | add g h hg hh => rw [map_add, hg, hh, map_add, map_add, map_add] | mul_X g i hg => have hX : q (MvPolynomial.X i) = q (MvPolynomial.C (a i)) := by rw [Ideal.Quotient.eq] exact Ideal.subset_span ⟨i, rfl⟩ rw [map_mul, hg, hX, ← map_mul, ← map_mul] simp have h0 := hred f rw [hf, map_zero, map_zero] at h0 exact Ideal.Quotient.eq_zero_iff_mem.mp h0
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
Comparison notes
V(J) is Mathlib's zero locus in the prime spectrum, V_K(J) the classical zero set, vanishingIdealOf the vanishing ideal I(S), and maxIdealAt the ideal 𝔪_a. mem_maxIdealAt_iff is f ∈ 𝔪_a ⇔ f(a) = 0; mem_VK_iff_closedPoint_mem_V is a ∈ V_K(J) ⇔ 𝔪_a ∈ V(J). The added radical notation is Mathlib's Ideal.radical. PrimeSpectrum.zeroLocus_radical gives V(J) = V(√J); the proof of definition_2_7_independent already uses it.
Review notes
Definition 2.6 — Cotangent bundle of affine space
Lean: Section2.CotangentBundle,
Section2.proj,
Section2.fibre;
Characteristic.SymbolRing
(the ring ; the first summand indexes base coordinates , and the second indexes fibre coordinates )
/-- **Definition 2.5 (cotangent bundle).** `T^*X = Spec k[x, ξ]` for `X = 𝔸^n_k`. -/ abbrev CotangentBundle (n : ℕ) := PrimeSpectrum (SymbolRing k n)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.5 (projection on closed points).** `π(y, ξ) = y`. -/ def proj {K : Type v} {n : ℕ} (z : PhaseVar n → K) : Fin n → K := fun i => z (Sum.inl i)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.5 (fibre).** `T^*_y X = π^{-1}(y) ≅ (K^n)^*`. -/ def fibre {K : Type v} {n : ℕ} (y : Fin n → K) : Set (PhaseVar n → K) := proj ⁻¹' {y}
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (k : Type u) [Field k]
Comparison notes
CotangentBundle is Spec k[x, ξ], the spectrum in Definition 2.6. Section 5.1 identifies its closed points over algebraically closed K with pairs (y, ξ), by Theorem 2.11(a). On these points, proj is π(y, ξ) = y and fibre y = π⁻¹(y), as defined in Section 5.1. Lean represents a pair by a function PhaseVar n → K, with Sum.inl i the base coordinate x_{i+1} and Sum.inr i the fibre coordinate ξ_{i+1}.
Review notes
Definition 2.7 — Characteristic variety
is independent of the choice of the good filtration. [Hotta 2008, Lemma D.3.1]
Section2.CharOf,
Section2.Char,
Section2.definition_2_7_independent,
Section2.Char_eq/-- **Definition 2.7, last sentence.** `Char(M)` is independent of the choice of the good filtration [HTT, Lemma D.3.1]. -/ theorem definition_2_7_independent {Γ Γ' : ℕ → Submodule k M} (hΓ : IsGoodFiltration k n Γ) (hΓ' : IsGoodFiltration k n Γ') : CharOf k n Γ = CharOf k n Γ' := by rw [CharOf, CharOf, V, V, ← PrimeSpectrum.zeroLocus_radical, radical_annihilator_eq_of_good hΓ.compatible hΓ'.compatible hΓ.finite hΓ'.finite, PrimeSpectrum.zeroLocus_radical]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Paper.Support.GabberSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.EulerSurjectivitySource ↗universe u wSource ↗
variable (k : Type u) [Field k] (n : ℕ)Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl k n)ᵐᵒᵖ M] [Module k M] [IsScalarTower k (PresentedWeyl k n)ᵐᵒᵖ M]Source ↗
variable {k n}More Lean proofs and definitions
/-- **Definition 2.7 (characteristic variety).** For a finitely generated `A_n(k)`-module `M` with a good order filtration `M_•`, `Char(M) = V(Ann_{k[x,ξ]} gr M) ⊆ T^*X = Spec k[x, ξ]`. -/ def CharOf (Γ : ℕ → Submodule k M) : Set (CotangentBundle k n) := V (Module.annihilator (SymbolRing k n) (GrModule Γ))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Paper.Support.GabberSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.EulerSurjectivitySource ↗universe u wSource ↗
variable (k : Type u) [Field k] (n : ℕ)Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl k n)ᵐᵒᵖ M] [Module k M] [IsScalarTower k (PresentedWeyl k n)ᵐᵒᵖ M]
/-- **`√(Ann gr M)` is independent of the good filtration** [HTT, Lemma D.3.1]. -/ theorem radical_annihilator_eq_of_good (hΓ : IsCompatibleFiltration (orderFiltration K n) Γ) (hΓ' : IsCompatibleFiltration (orderFiltration K n) Γ') (hgood : Module.Finite (SymbolRing K n) (GrModule Γ)) (hgood' : Module.Finite (SymbolRing K n) (GrModule Γ')) : (Module.annihilator (SymbolRing K n) (GrModule Γ)).radical = (Module.annihilator (SymbolRing K n) (GrModule Γ')).radical := by obtain ⟨c, hc⟩ := exists_shift_le hΓ' hgood' hΓ obtain ⟨c', hc'⟩ := exists_shift_le hΓ hgood hΓ' exact radical_annihilator_eq_of_shift hΓ hΓ' hc hc'
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗universe u w w' w₁Source ↗
variable {K : Type u} [Field K] {n : ℕ}Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (PresentedWeyl K n)ᵐᵒᵖ M]Source ↗
variable {Γ Γ' : ℕ → Submodule K M}
/-- A good filtration is dominated, up to a shift, by every compatible exhaustive filtration: `M_i ⊆ M'_{i+c}`. -/ theorem exists_shift_le (hΓ : IsCompatibleFiltration (orderFiltration K n) Γ) (hgood : Module.Finite (SymbolRing K n) (GrModule Γ)) (hΓ' : IsCompatibleFiltration (orderFiltration K n) Γ') : ∃ c, ∀ i, Γ i ≤ Γ' (i + c) := by classical obtain ⟨s, hs⟩ := hgood.fg_top choose cz hcz using exists_mem_boundedSpan (Γ := Γ) hΓ' let C := s.sup cz let W : Submodule (SymbolRing K n) (GrModule Γ) := { boundedSpan Γ Γ' C with smul_mem' := fun f _ hz => symbol_smul_mem_boundedSpan hΓ hΓ' C f hz } have htop : W = ⊤ := by rw [eq_top_iff, ← hs, Submodule.span_le] intro z hz exact boundedSpan_mono hΓ' (Finset.le_sup (f := cz) hz) (hcz z) refine ⟨C, fun i => ?_⟩ induction i with | zero => intro x hx have hmem : cls Γ 0 x ∈ W := htop ▸ Submodule.mem_top obtain ⟨y, hy, hev⟩ := ev_of_mem_boundedSpan hmem 0 rw [ev_cls 0 hx, ite_eq_left rfl, Submodule.Quotient.eq] at hev have : x = y := sub_eq_zero.mp ((Submodule.mem_bot K).mp hev) rwa [this] | succ i ih => intro x hx have hmem : cls Γ (i + 1) x ∈ W := htop ▸ Submodule.mem_top obtain ⟨y, hy, hev⟩ := ev_of_mem_boundedSpan hmem (i + 1) rw [ev_cls _ hx, ite_eq_left rfl, Submodule.Quotient.eq] at hev have hlow : x - y ∈ Γ' (i + C) := ih hev have : x = (x - y) + y := by abel rw [this] exact add_mem (hΓ'.mono (by omega) hlow) hy
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗universe u w w' w₁Source ↗
variable {K : Type u} [Field K] {n : ℕ}Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (PresentedWeyl K n)ᵐᵒᵖ M]Source ↗
variable {Γ Γ' : ℕ → Submodule K M}
/-- Shift-comparable compatible filtrations have the same `√(Ann gr M)`. -/ theorem radical_annihilator_eq_of_shift (hΓ : IsCompatibleFiltration (orderFiltration K n) Γ) (hΓ' : IsCompatibleFiltration (orderFiltration K n) Γ') {c c' : ℕ} (h₁ : ∀ i, Γ' i ≤ Γ (i + c)) (h₂ : ∀ i, Γ i ≤ Γ' (i + c')) : (Module.annihilator (SymbolRing K n) (GrModule Γ)).radical = (Module.annihilator (SymbolRing K n) (GrModule Γ')).radical := le_antisymm (Ideal.radical_le_radical_iff.mpr (annihilator_le_radical hΓ hΓ' h₁ h₂)) (Ideal.radical_le_radical_iff.mpr (annihilator_le_radical hΓ' hΓ h₂ h₁))
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗universe u w w' w₁Source ↗
variable {K : Type u} [Field K] {n : ℕ}Source ↗
variable {M : Type w} [AddCommGroup M] [Module (PresentedWeyl K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (PresentedWeyl K n)ᵐᵒᵖ M]Source ↗
variable {Γ Γ' : ℕ → Submodule K M}
/-- `Char(A_n(k)/I)`, Definition 2.7 for `M = A_n(k)/I` with its good filtration `Q_m = qF_m`; by `definition_2_7_independent` any good filtration gives the same set (`Char_eq_charOf`). -/ def Char (I : RightIdeal (PresentedWeyl k n)) : Set (CotangentBundle k n) := CharOf k n (qFilt k I)
/-- `Char(A_n/I) = V(Ann_{k[x,ξ]} gr Q)` with `gr Q` the repository's `grCyclic`. -/ theorem Char_eq (I : RightIdeal (PresentedWeyl k n)) : Char k I = V (Module.annihilator (SymbolRing k n) (grCyclic k I)) := by rw [Char, CharOf, annihilator_qFilt]
/-- `Char(A_n/I)` computed from any good filtration of `A_n/I`. -/ theorem Char_eq_charOf (I : RightIdeal (PresentedWeyl k n)) {Γ : ℕ → Submodule k (RightQuotient I)} (hΓ : IsGoodFiltration k n Γ) : Char k I = CharOf k n Γ := definition_2_7_independent (qFilt_isGoodFiltration k I) hΓ
Comparison notes
CharOf is the printed definition: for a right A_n(k)-module M with a filtration M_•, CharOf M_• = V(Ann_{k[x,ξ]} gr M) ⊆ Spec k[x, ξ], meant for good order filtrations (IsGoodFiltration, Definition 2.4). The last sentence is definition_2_7_independent: two good order filtrations of M give the same set. The paper cites it [HTT, Lemma D.3.1]; Lean proves it along that reference: two good filtrations satisfy F_j ⊆ G_{j+c} and G_j ⊆ F_{j+c'} (Support.Gabber.exists_shift_le), so the radicals of the two annihilators agree (radical_annihilator_eq_of_shift, radical_annihilator_eq_of_good), and V(J) = V(√J). Neither the definition nor the independence needs M finitely generated. Char is the case M = A_n(k)/I with Q_m = qF_m used in Sections 3–6; Char_eq writes it as V(Ann gr Q) with gr Q = grCyclic, and Char_eq_charOf computes it from any good filtration of A_n(k)/I. The displayed equality V(√Ann gr M) = V(Ann gr M) is PrimeSpectrum.zeroLocus_radical; the existing independence proof uses that equality.
Review notes
Definition 2.8 — Symplectic forms and the standard form on k^{2n}
/-- **Definition 2.8 (symplectic form).** `ω` is antisymmetric and nondegenerate (`ω(u, ·) = 0` forces `u = 0`). -/ def IsSymplectic (ω : BilinForm k V) : Prop := (∀ x y, ω x y = -ω y x) ∧ ω.Nondegenerate
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Definition 2.8 (Gram matrix).** `J = (ω(b_i, b_j))_{ij}`. -/ def gramMatrix {ι : Type*} (ω : BilinForm k V) (b : ι → V) : Matrix ι ι k := fun i j => ω (b i) (b j)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
More Lean proofs and definitions
/-- **Definition 2.8 (symplectic basis).** `e_1, …, e_n, f_1, …, f_n` with `ω(e_i, f_j) = δ_ij` and `ω(e_i, e_j) = ω(f_i, f_j) = 0`. -/ def IsSymplecticBasis {n : ℕ} (ω : BilinForm k V) (b : Basis (Fin n ⊕ Fin n) k V) : Prop := (∀ i j, ω (b (Sum.inl i)) (b (Sum.inr j)) = if i = j then 1 else 0) ∧ (∀ i j, ω (b (Sum.inl i)) (b (Sum.inl j)) = 0) ∧ (∀ i j, ω (b (Sum.inr i)) (b (Sum.inr j)) = 0)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Definition 2.8 (standard symplectic form).** On `𝕍 = k^{2n}` with basis `x_1, …, x_n, ∂_1, …, ∂_n`, `ω(u, v) = ∑_i (u_{x_i} v_{∂_i} - u_{∂_i} v_{x_i})`, so `ω(x_i, ∂_j) = δ_ij` and `ω(x_i, x_j) = ω(∂_i, ∂_j) = 0`. It is the negative of the repository's `phasePairing`, which uses the opposite sign convention. -/ def omega (n : ℕ) : BilinForm k (PhaseVar n → k) := LinearMap.mk₂ k (fun u v => -phasePairing k u v) (fun u v w => by rw [phasePairing_add_left, neg_add]) (fun c u v => by rw [phasePairing_smul_left, smul_eq_mul, mul_neg]) (fun u v w => by rw [phasePairing_add_right, neg_add]) (fun c u v => by rw [phasePairing_smul_right, smul_eq_mul, mul_neg])
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
theorem omega_isSymplectic (n : ℕ) : IsSymplectic (omega k n) := ⟨omega_antisymm k, omega_nondegenerate k n⟩
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- The Gram matrix of `ω` in the generator basis `x_1, …, ∂_n` is `-J`, where `J` is the matrix of Weyl commutators `[u_i, u_j]` (`standardForm`). -/ theorem gramMatrix_omega (n : ℕ) : gramMatrix (omega k n) (Pi.basisFun k (PhaseVar n)) = -standardForm k n := by rw [← omega_toMatrix] ext i j rw [LinearMap.BilinForm.toMatrix'_apply] simp [gramMatrix]
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
IsSymplectic (antisymmetric and nondegenerate), gramMatrix, IsSymplecticBasis and the standard form omega with ω(x_i, ∂_j) = δ_ij, ω(x_i, x_j) = ω(∂_i, ∂_j) = 0 are as printed; omega_isSymplectic. omega is the negative of the repository's phasePairing, and its Gram matrix in the generator basis is −J for J the matrix of Weyl commutators (gramMatrix_omega).
Review notes
Definition 2.9 — Poisson bracket and involutive ideals
Lean ():
Section2.poisson,
Section2.poisson_eq
Lean: Section2.IsInvolutive,
Section2.IsInvolutiveSet,
Section2.isInvolutive_iff;
negative control GeneralCoisotropicSetsTest.zeroSectionIdealOne_not_isPoisson
/-- **Definition 2.9 (Poisson bracket).** `{f, g} = ∑_i (∂f/∂x_i ∂g/∂ξ_i - ∂f/∂ξ_i ∂g/∂x_i)` on `k[x, ξ]`. -/ def poisson {n : ℕ} (f g : SymbolRing k n) : SymbolRing k n := ∑ i : Fin n, (MvPolynomial.pderiv (Sum.inl i) f * MvPolynomial.pderiv (Sum.inr i) g - MvPolynomial.pderiv (Sum.inr i) f * MvPolynomial.pderiv (Sum.inl i) g)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.10 (involutive ideal).** `{J, J} ⊆ J`. -/ def IsInvolutive {n : ℕ} (J : Ideal (SymbolRing k n)) : Prop := ∀ f ∈ J, ∀ g ∈ J, poisson k f g ∈ J
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- The printed bracket is the repository's `Characteristic.poissonBracket`, with the same sign. -/ theorem poisson_eq {n : ℕ} (f g : SymbolRing k n) : poisson k f g = poissonBracket f g := rfl
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- The printed notion is the repository's `PostScalarExtensionPoisson.IsInvolutive`. -/ theorem isInvolutive_iff {n : ℕ} (J : Ideal (SymbolRing k n)) : IsInvolutive k J ↔ Stafford38.Characteristic.PostScalarExtensionPoisson.IsInvolutive J := Iff.rfl
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
poisson f g = Σ_i (∂f/∂x_i ∂g/∂ξ_i − ∂f/∂ξ_i ∂g/∂x_i), the printed bracket; poisson_eq identifies it with the repository's Characteristic.poissonBracket, same sign. IsInvolutive J is the printed condition {J, J} ⊆ J; isInvolutive_iff identifies it with the condition used by the proofs. The set-level definition is in Definition 5.6.
Review notes
Theorem 2.10 — Gabber’s involutivity theorem
Section2.theorem_2_11,
Section2.theorem_2_11_cyclic/-- **Theorem 2.11 (Gabber).** Let `K` be a field of characteristic zero and `M` a finitely generated right `A_n(K)`-module with a good order filtration `M_•` (Definition 2.4, `IsGoodFiltration`: compatible and exhaustive, `gr M` finitely generated over `K[x, ξ]`). Then `J = √(Ann_{K[x,ξ]} gr M)` is involutive. The cited result [Gab81] is proved in `Stafford38.Paper.Support.Gabber`, by reduction to cyclic modules and the repository's proof of the cyclic case. -/ theorem theorem_2_11 (K : Type u) [Field K] [CharZero K] (n : ℕ) {M : Type w} [AddCommGroup M] [Module (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] [Module.Finite (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] (Γ : ℕ → Submodule K M) (hgood : IsGoodFiltration K n Γ) : IsInvolutive K (Module.annihilator (SymbolRing K n) (Stafford38.Paper.Support.Gabber.GrModule Γ)).radical := (isInvolutive_iff K _).2 (Stafford38.Paper.Support.Gabber.theorem_2_11_general K n Γ hgood.compatible hgood.finite)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- Theorem 2.11 for the cyclic module `M = A_n(K)/I` with `M_m = qF_m`, the case used in the proof of Theorem 1.1 (for `Q_K`): for this filtration `gr M` is `grCyclic` as a `K[x, ξ]`-module. -/ theorem theorem_2_11_cyclic (K : Type u) [Field K] [CharZero K] (n : ℕ) (I : RightIdeal (PresentedWeyl K n)) : IsInvolutive K (Module.annihilator (SymbolRing K n) (grCyclic K I)).radical := by have : Module.Finite (PresentedWeyl K n)ᵐᵒᵖ (RightQuotient I) := Stafford38.Paper.Support.Gabber.rightQuotient_finite I rw [← Stafford38.Paper.Support.Gabber.annihilator_qFilt] exact theorem_2_11 K n (Stafford38.Paper.Support.Gabber.qFilt K I) (qFilt_isGoodFiltration K I)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Theorem 2.11 (Gabber).** Let `K` be a field of characteristic zero and `M` a finitely generated right `A_n(K)`-module with a good order filtration `M_•` (Definition 2.4: compatible, exhaustive, `gr M` finitely generated over `K[x, ξ]`). Then `J = √(Ann_{K[x,ξ]} gr M)` is involutive, `{J, J} ⊆ J`. The proof reduces to the cyclic case proved in the repository (`GabberGlobalAssembly.associatedGraded_radical_isInvolutive`); see the module docstring. -/ theorem theorem_2_11_general (K : Type u) [Field K] [CharZero K] (n : ℕ) {M : Type w} [AddCommGroup M] [Module (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] [Module K M] [IsScalarTower K (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] [Module.Finite (Stafford38Challenge.WeylAlg K n)ᵐᵒᵖ M] (Γ : ℕ → Submodule K M) (hΓ : IsCompatibleFiltration (orderFiltration K n) Γ) (hgood : Module.Finite (SymbolRing K n) (GrModule Γ)) : IsInvolutive (Module.annihilator (SymbolRing K n) (GrModule Γ)).radical := by obtain ⟨s, hs⟩ := Module.Finite.fg_top (R := (PresentedWeyl K n)ᵐᵒᵖ) (M := M) exact isInvolutive_of_card s.card s le_rfl hs Γ hΓ hgood
Variables and notation
Source ↗open Stafford38.Paper.Section2Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicFilteredQuotientGradedSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.EulerSurjectivitySource ↗universe u w w' w₁
More Lean proofs and definitions
theorem associatedGraded_radical_isInvolutive : IsInvolutive (Module.annihilator (SymbolRing k n) (OrderAssociatedGradedModule k I)).radical := by let J := Module.annihilator (SymbolRing k n) (OrderAssociatedGradedModule k I) intro f hf g hg rw [← J.sInf_minimalPrimes] apply Ideal.mem_sInf.mpr intro P hP letI : P.IsPrime := hP.1.1 have hle : J.radical ≤ P := hP.1.1.radical_le_iff.mpr hP.1.2 exact minimalPrime_isInvolutive k I P hP f (hle hf) g (hle hg)
Variables and notation
Source ↗open Stafford38.Characteristic.BGab001CoefficientFieldTraceSource ↗open Stafford38.Characteristic.ConcreteInducedZActionSource ↗open Stafford38.Characteristic.ConcreteEquation33SourceMatricesSource ↗open Stafford38.Characteristic.ConcreteLocalizedTwoBlockSpecialFibreSource ↗open Stafford38.Characteristic.LocalizedTwoBlockQuotientSource ↗open Stafford38.Characteristic.LocalizedTwoBlockModuleExactnessSource ↗open Stafford38.Characteristic.SquareZeroLocalizedExactnessSource ↗open Stafford38.Characteristic.SquareZeroLocalizedRingSource ↗open Stafford38.Characteristic.SquareZeroTraceDataSource ↗open Stafford38.Characteristic.RightReesArtinianAdapterSource ↗open Stafford38.CharacteristicConcreteSquareZeroTraceDataSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicOrderReesTwoJetSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.EulerSurjectivitySource ↗universe uSource ↗
variable (k : Type u) [Field k] [CharZero k] {n : ℕ}Source ↗
variable (I : RightIdeal (PresentedWeyl k n))Source ↗
variable (P : Ideal (SymbolRing k n)) [P.IsPrime]Source ↗
local notation "D" => filteredQuotientTwoJetTraceData k ISource ↗
local notation "S" => P.primeComplSource ↗local notation "R" => Localization SSource ↗
local notation "G" => LocalizedModule S (OrderAssociatedGradedModule k I)Source ↗
local notation "Wₗ" => LocalizedDeformationModule D SSource ↗
local notation "C" => LocalizedTwoBlockRing D SSource ↗
variable [hOre : OreLocalization.OreSet (OppositeDenominators (filteredQuotientTwoJetTraceData k I) P.primeCompl)]Source ↗
omit hOre intheorem minimalPrime_isInvolutive (hP : P ∈ (Module.annihilator (SymbolRing k n) (OrderAssociatedGradedModule k I)).minimalPrimes) : IsInvolutive P := by obtain ⟨h, core⟩ := exists_concrete_localizedPreArtinianCore k I P hP letI := h letI : Nontrivial G := core.fibre_nontrivial letI : IsNoetherian R G := (isFiniteLength_iff_isNoetherian_isArtinian.mp core.fibre_finiteLength).1 obtain ⟨q, hq⟩ := core.annihilatingPower have hpow : IsLocalRing.maximalIdeal R ^ (q + 1) ≤ Module.annihilator R G := (Ideal.pow_le_pow_right (Nat.le_succ q)).trans hq intro f hf g hg obtain ⟨a, b, z, ha, hb, hab, hz⟩ := RightSquareZeroTraceData.exists_lifts_commutator_factor k _ _ D f g let φ : (OrderReesTwoJet (n := n) k)ᵐᵒᵖ →+* C (q + 1) := ((localizedTwoBlockIdeal D S (q + 1)).ringCon.mk').comp OreLocalization.numeratorRingHom let x := φ (MulOpposite.op b) let y := φ (MulOpposite.op a) let zq := φ (MulOpposite.op z) let H := localizedDoubledPower S (q + 1) have hspec (v : OrderReesTwoJet (n := n) k) : localizedTwoBlockSpecialization D S (q + 1) (φ (MulOpposite.op v)) = Ideal.Quotient.mk H (algebraMap (SymbolRing k n) R ((D).pi v)) := by rw [show φ (MulOpposite.op v) = (localizedTwoBlockIdeal D S (q + 1)).ringCon.mk' (OreLocalization.numeratorRingHom (MulOpposite.op v)) from rfl, localizedTwoBlockSpecialization_mk] change Ideal.Quotient.mk H (localizedOppositeSpecialization D S (OreLocalization.numeratorRingHom (MulOpposite.op v))) = _ exact congrArg (Ideal.Quotient.mk H) (localizedOppositeSpecialization_numerator D S (MulOpposite.op v)) have hx : localizedTwoBlockSpecialization D S (q + 1) x ∈ (IsLocalRing.maximalIdeal R).map (Ideal.Quotient.mk H) := by rw [show x = φ (MulOpposite.op b) from rfl, hspec, hb] exact Ideal.mem_map_of_mem _ ((IsLocalization.AtPrime.to_map_mem_maximal_iff R P g).mpr hg) have hy : localizedTwoBlockSpecialization D S (q + 1) y ∈ (IsLocalRing.maximalIdeal R).map (Ideal.Quotient.mk H) := by rw [show y = φ (MulOpposite.op a) from rfl, hspec, ha] exact Ideal.mem_map_of_mem _ ((IsLocalization.AtPrime.to_map_mem_maximal_iff R P f).mpr hf) have hop : MulOpposite.op b * MulOpposite.op a - MulOpposite.op a * MulOpposite.op b = MulOpposite.op (D).c * MulOpposite.op z := by apply MulOpposite.unop_injective change a * b - b * a = z * (D).c change a * b - b * a = (D).c * z at hab exact hab.trans ((D).c_center.comm z).eq have hxy : x * y - y * x = (concreteLeftPrincipalParameterReduction k I S (q + 1)).parameter * zq := by have hparam : (concreteLeftPrincipalParameterReduction k I S (q + 1)).parameter = φ (MulOpposite.op (D).c) := rfl rw [hparam] have hh := congrArg φ hop simpa only [map_sub, map_mul] using hh have hlocal := localized_cofactor_mem k I P (q + 1) (by omega) hpow x y zq hx hy hxy rw [show zq = φ (MulOpposite.op z) from rfl, hspec, hz, filteredQuotientTwoJetTraceData_bracket] at hlocal have hHle : H ≤ IsLocalRing.maximalIdeal R := Ideal.mul_le_left.trans (Ideal.pow_le_self (by omega)) have hneg : algebraMap (SymbolRing k n) R (-poissonBracket f g) ∈ IsLocalRing.maximalIdeal R := by have hc := (Ideal.mem_comap).mpr hlocal rw [Ideal.comap_map_of_surjective (Ideal.Quotient.mk H) Ideal.Quotient.mk_surjective, ← RingHom.ker_eq_comap_bot, Ideal.mk_ker, sup_eq_left.mpr hHle] at hc exact hc exact (P.neg_mem_iff).mp ((IsLocalization.AtPrime.to_map_mem_maximal_iff R P _).mp hneg) omit hOre in
Variables and notation
Source ↗open Stafford38.Characteristic.BGab001CoefficientFieldTraceSource ↗open Stafford38.Characteristic.ConcreteInducedZActionSource ↗open Stafford38.Characteristic.ConcreteEquation33SourceMatricesSource ↗open Stafford38.Characteristic.ConcreteLocalizedTwoBlockSpecialFibreSource ↗open Stafford38.Characteristic.LocalizedTwoBlockQuotientSource ↗open Stafford38.Characteristic.LocalizedTwoBlockModuleExactnessSource ↗open Stafford38.Characteristic.SquareZeroLocalizedExactnessSource ↗open Stafford38.Characteristic.SquareZeroLocalizedRingSource ↗open Stafford38.Characteristic.SquareZeroTraceDataSource ↗open Stafford38.Characteristic.RightReesArtinianAdapterSource ↗open Stafford38.CharacteristicConcreteSquareZeroTraceDataSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicOrderReesTwoJetSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.EulerSurjectivitySource ↗universe uSource ↗
variable (k : Type u) [Field k] [CharZero k] {n : ℕ}Source ↗
variable (I : RightIdeal (PresentedWeyl k n))Source ↗
variable (P : Ideal (SymbolRing k n)) [P.IsPrime]Source ↗
local notation "D" => filteredQuotientTwoJetTraceData k ISource ↗
local notation "S" => P.primeComplSource ↗local notation "R" => Localization SSource ↗
local notation "G" => LocalizedModule S (OrderAssociatedGradedModule k I)Source ↗
local notation "Wₗ" => LocalizedDeformationModule D SSource ↗
local notation "C" => LocalizedTwoBlockRing D SSource ↗
variable [hOre : OreLocalization.OreSet (OppositeDenominators (filteredQuotientTwoJetTraceData k I) P.primeCompl)]Source ↗
omit hOre in/-- **Definition 2.10 (involutive ideal).** `{J, J} ⊆ J`. -/ def IsInvolutive {n : ℕ} (J : Ideal (SymbolRing k n)) : Prop := ∀ f ∈ J, ∀ g ∈ J, poisson k f g ∈ J
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
Section2.theorem_2_11 states the theorem as printed: for a field K of characteristic zero and a finitely generated right A_n(K)-module M with a good order filtration (IsGoodFiltration, Definition 2.4: compatible, exhaustive, gr M finite over K[x, ξ]), √(Ann_{K[x,ξ]} gr M) is involutive. The paper cites it [Gab81, SK14, HTT Thm. D.3.4] without proof; the repository proves it by its own route: Support.Gabber.theorem_2_11_general reduces it to cyclic modules (independence of the good filtration, exactness of gr for induced filtrations, induction on the number of generators), and the cyclic case M = A_n(K)/I is GabberGlobalAssembly.associatedGraded_radical_isInvolutive, which follows Gabber and Singh–Kumar (order-Rees ring modulo T², Ore localization at a minimal prime, an Artinian special fibre with a coefficient field, and a trace argument). theorem_2_11_cyclic is the case used in Lemma 4.10 and Section 6.
Review notes
Theorem 2.11 — Hilbert's Nullstellensatz
- [(a)] the maximal ideals of are exactly , ; hence identifies with the closed points of , and with the closed points of ;
- [(b)] if and only if ;
- [(c)] .
/-- **Theorem 2.12 (Hilbert's Nullstellensatz).** Let `K` be algebraically closed and `J ⊆ K[z_1, …, z_N]` an ideal (the variables indexed by any finite type `σ`, e.g. `PhaseVar n` for `K[x, ξ]`). (a) The maximal ideals of `K[z]` are exactly the `𝔪_a`, `a ↦ 𝔪_a` is injective, and `V_K(J)` corresponds to the closed points of `V(J)`; (b) `V_K(J) = ∅` iff `1 ∈ J`; (c) `I(V_K(J)) = √J`. -/ theorem theorem_2_12 [IsAlgClosed K] [Finite σ] (J : Ideal (MvPolynomial σ K)) : ((∀ M : Ideal (MvPolynomial σ K), M.IsMaximal ↔ ∃ a, M = maxIdealAt K a) ∧ Function.Injective (maxIdealAt K (σ := σ)) ∧ ∀ a, a ∈ VK K J ↔ closedPoint K a ∈ V J) ∧ (VK K J = ∅ ↔ (1 : MvPolynomial σ K) ∈ J) ∧ vanishingIdealOf K (VK K J) = J.radical := by have hc : vanishingIdealOf K (VK K J) = J.radical := by rw [vanishingIdealOf, VK_eq_zeroLocus, MvPolynomial.vanishingIdeal_zeroLocus_eq_radical] refine ⟨⟨fun M => ?_, ?_, mem_VK_iff_closedPoint_mem_V K J⟩, ⟨fun h => ?_, fun h => ?_⟩, hc⟩ · simp only [maxIdealAt_eq_vanishingIdeal] exact MvPolynomial.isMaximal_iff_eq_vanishingIdeal_singleton · intro a b hab funext i have hmem : MvPolynomial.X i - MvPolynomial.C (a i) ∈ maxIdealAt K b := hab ▸ Ideal.subset_span ⟨i, rfl⟩ rw [mem_maxIdealAt_iff] at hmem simpa [sub_eq_zero, eq_comm] using hmem · have h1 : (1 : MvPolynomial σ K) ∈ J.radical := by rw [← hc, h, mem_vanishingIdealOf] simp exact (Ideal.radical_eq_top.mp ((Ideal.eq_top_iff_one _).mpr h1)).symm ▸ Submodule.mem_top · ext a simp only [Set.mem_empty_iff_false, iff_false] intro ha simpa using ha 1 h
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
Comparison notes
theorem_2_12 states (a)–(c) for an ideal J of K[z_1, …, z_N], K algebraically closed, the variables indexed by any finite type: (a) the maximal ideals are exactly the 𝔪_a, a ↦ 𝔪_a is injective, and a ∈ V_K(J) ⇔ 𝔪_a ∈ V(J); (b) V_K(J) = ∅ ⇔ 1 ∈ J; (c) I(V_K(J)) = √J. The paper cites the theorem; Lean derives it from Mathlib's Nullstellensatz.
Review notes
3. Mapping to a normal form
Paragraph — Scalar case N = 0
If , then is a nonzero scalar, hence a unit, and holds trivially with , , . We therefore assume .
/-- **Section 3, opening paragraph.** If `N = deg_B d = 0`, then `d ∈ B_0 = k` is a nonzero scalar `c`, hence a unit, and `1 = dR + FdS` with `F = 1`, `R = d⁻¹ = c⁻¹`, `S = 0`. Otherwise `N ≥ 1`. -/ theorem step_scalar_case {n : ℕ} {d : PresentedWeyl k n} (hd : d ≠ 0) (hN : bernsteinDegree k d = 0) : ∃ c : k, c ≠ 0 ∧ d = algebraMap k (PresentedWeyl k n) c ∧ (1 : PresentedWeyl k n) = d * algebraMap k _ c⁻¹ + 1 * d * 0 := by have hpiece : d ∈ bernsteinPiece k n 0 := (bernsteinDegree_le_iff k d).mp hN.le obtain ⟨c, rfl⟩ := eq_algebraMap_of_mem_bernsteinPiece_zero k hpiece have hc : c ≠ 0 := by rintro rfl exact hd (map_zero _) refine ⟨c, hc, rfl, ?_⟩ rw [mul_zero, add_zero, ← map_mul, mul_inv_cancel₀ hc, map_one]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
step_scalar_case: if deg_B d = 0 then d = c ∈ k with c ≠ 0, and 1 = d·c⁻¹ + 1·d·0, i.e. F = 1, R = d⁻¹, S = 0.
Review notes
Lemma 3.1 — Symplectic completion
Let . For : if , ; if , . Either way pairs nontrivially with something in , so is nondegenerate.
Since is nondegenerate, , and as [Coutinho 1995, Prop. 11.2.1] for a nondegenerate form, this means . If pairs to zero with all of , it also pairs to zero with (as ), hence with all of , forcing ; so is nondegenerate too.
By induction on (trivial for , where ), a symplectic basis of therefore exists [Lang 2012, XV.8.1 and XV.8.2]. Setting , gives a symplectic basis of , since cross-pairings between and vanish by definition of .
Consequently: given nonzero , we can complete both vectors to a symplectic basis as above. We then define the linear map to send to , and each subsequent -basis vector to the corresponding -basis vector, thus preserving all pairwise values of on basis vectors. Through the bilinearity of , then preserves on all of ; and . ∎
Section3.lemma_3_1
Let be a -linear automorphism of that preserves .
Since is identified with the span of the generators of ,
assigns to each generator a linear combination of generators.
Because is presented by generators and the Weyl relations
, and preserves , the elements satisfy
the same relations as the . Hence the substitution extends
to a unique -algebra endomorphism of [Coutinho 1995, Th. 1.3.1, Ex. 7.6.9]. Because is an automorphism, so is , and we have Lean: Section3.phiG,
Section3.step_phi_g
/-- **Lemma 3.1.** Let `(𝕍, ω)` be a symplectic `k`-space of dimension `2n` (`ω` antisymmetric and nondegenerate) and `0 ≠ v ∈ 𝕍`. Then `v` is the first vector `e_1` of a symplectic basis of `𝕍`; consequently, for every nonzero `e ∈ 𝕍` there is a `k`-linear automorphism `g` preserving `ω` with `g(e) = v`. -/ theorem lemma_3_1 [CharZero k] [Module.Finite k V] (n : ℕ) (ω : BilinForm k V) (hanti : ∀ x y, ω x y = -ω y x) (hnd : ω.Nondegenerate) (hdim : Module.finrank k V = 2 * n) {v : V} (hv : v ≠ 0) : (∃ (hn : 0 < n) (b : Basis (Fin n ⊕ Fin n) k V), IsSymplecticBasis ω b ∧ b (Sum.inl ⟨0, hn⟩) = v) ∧ ∀ e : V, e ≠ 0 → ∃ g : V ≃ₗ[k] V, (∀ x y, ω (g x) (g y) = ω x y) ∧ g e = v := by have hAlt := alternating_of_antisymmetric ω hanti -- every nonzero vector is the first vector of a symplectic basis have hcomplete : ∀ u : V, u ≠ 0 → ∃ m : ℕ, n = m + 1 ∧ ∃ b : Basis (Fin (m + 1) ⊕ Fin (m + 1)) k V, IsSymplecticBasis ω b ∧ b (Sum.inl 0) = u := by intro u hu obtain ⟨w, huw⟩ := step_lemma_3_1_partner ω hnd hu obtain ⟨-, hcompl, hWnd, hWdim⟩ := step_lemma_3_1_complement ω hAlt hnd huw obtain ⟨m, rfl⟩ : ∃ m, n = m + 1 := ⟨n - 1, by omega⟩ refine ⟨m, rfl, ?_⟩ obtain ⟨b, hb, hb0, -⟩ := step_lemma_3_1_basis m ω hAlt huw hcompl hWnd (by omega) exact ⟨b, hb, hb0⟩ obtain ⟨m, rfl, b, hb, hb0⟩ := hcomplete v hv refine ⟨⟨Nat.succ_pos m, b, hb, hb0⟩, fun e he => ?_⟩ obtain ⟨m', hm', b', hb', hb'0⟩ := hcomplete e he obtain rfl : m' = m := by omega obtain ⟨hpres, happ⟩ := step_lemma_3_1_transport ω hAlt b' b hb' hb exact ⟨b'.equiv b (Equiv.refl _), hpres, by rw [← hb'0, happ, hb0]⟩
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
More Lean proofs and definitions
/-- **Lemma 3.1, paragraph 1.** As `ω` is nondegenerate and `v ≠ 0`, some `w'` has `ω(v, w') = c ≠ 0`; `w := w'/c` has `ω(v, w) = 1`. -/ theorem step_lemma_3_1_partner (ω : BilinForm k V) (hnd : ω.Nondegenerate) {v : V} (hv : v ≠ 0) : ∃ w : V, ω v w = 1 := by have hpair : ∃ w', ω v w' ≠ 0 := by by_contra h exact hv (hnd.1 v fun y => by_contra fun hy => h ⟨y, hy⟩) obtain ⟨w', hw'⟩ := hpair refine ⟨(ω v w')⁻¹ • w', ?_⟩ rw [BilinForm.smul_right] exact inv_mul_cancel₀ hw'
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Lemma 3.1, paragraph 2.** On `U = span(v, w)`: for `av + bw ≠ 0`, if `a ≠ 0` then `ω(av + bw, w) = a ≠ 0`, and if `b ≠ 0` then `ω(av + bw, v) = -b ≠ 0`; so `ω|_U` is nondegenerate. -/ theorem step_lemma_3_1_plane (ω : BilinForm k V) (hAlt : ∀ x, ω x x = 0) {v w : V} (hvw : ω v w = 1) (a b : k) : ω (a • v + b • w) w = a ∧ ω (a • v + b • w) v = -b := by have hwv : ω w v = -1 := by have h := BilinForm.IsAlt.neg_eq (show ω.IsAlt from hAlt) v w simpa [hvw] using h.symm constructor · rw [BilinForm.add_left, BilinForm.smul_left, BilinForm.smul_left, hvw, hAlt w] ring · rw [BilinForm.add_left, BilinForm.smul_left, BilinForm.smul_left, hAlt v, hwv] ring
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Lemma 3.1, paragraph 3.** `U ∩ U^⊥ = 0` because `ω|_U` is nondegenerate (paragraph 2); with `dim U + dim U^⊥ = 2n` this gives `𝕍 = U ⊕ U^⊥`, and `ω|_{U^⊥}` is nondegenerate. The intersection is computed from paragraph 2; the dimension count and the nondegeneracy of `ω|_{U^⊥}` are `SymplecticComplement.pairComplementData`. -/ theorem step_lemma_3_1_complement [Module.Finite k V] (ω : BilinForm k V) (hAlt : ∀ x, ω x x = 0) (hnd : ω.Nondegenerate) {v w : V} (hvw : ω v w = 1) : let U : Submodule k V := Submodule.span k {v, w} Disjoint U (ω.orthogonal U) ∧ IsCompl U (ω.orthogonal U) ∧ (ω.restrict (ω.orthogonal U)).Nondegenerate ∧ Module.finrank k (ω.orthogonal U) + 2 = Module.finrank k V := by intro U have hdata := Stafford38.LinearAlgebra.SymplecticComplement.pairComplementData ω hAlt hnd v w hvw refine ⟨?_, hdata.2.1, hdata.2.2.1, hdata.2.2.2⟩ rw [Submodule.disjoint_def] intro x hxU hxO obtain ⟨a, b, rfl⟩ := Submodule.mem_span_pair.1 hxU have hvU : v ∈ U := Submodule.subset_span (by simp) have hwU : w ∈ U := Submodule.subset_span (by simp) have hskew (x y : V) : ω x y = -ω y x := ((show ω.IsAlt from hAlt).neg_eq y x).symm have hw0 : ω (a • v + b • w) w = 0 := by rw [hskew]; simp [hxO w hwU] have hv0 : ω (a • v + b • w) v = 0 := by rw [hskew]; simp [hxO v hvU] obtain ⟨ha, hb⟩ := step_lemma_3_1_plane ω hAlt hvw a b have ha0 : a = 0 := by rw [← ha, hw0] have hb0 : b = 0 := by rw [← neg_neg b, ← hb, hv0, neg_zero] simp [ha0, hb0]
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Lemma 3.1, paragraph 4.** By induction on `n`, `U^⊥` has a symplectic basis `e_2, …, f_n`; with `e_1 := v`, `f_1 := w` this is a symplectic basis of `𝕍 = U ⊕ U^⊥`, since cross pairings between `U` and `U^⊥` vanish. The induction is `CharacteristicPaperSymplecticBasis.exists_symplectic_basis` applied to `U^⊥` with the data of paragraph 3. -/ theorem step_lemma_3_1_basis [Module.Finite k V] (n : ℕ) (ω : BilinForm k V) (hAlt : ∀ x, ω x x = 0) {v w : V} (hvw : ω v w = 1) (hcompl : IsCompl (Submodule.span k {v, w}) (ω.orthogonal (Submodule.span k {v, w}))) (hWnd : (ω.restrict (ω.orthogonal (Submodule.span k {v, w}))).Nondegenerate) (hWdim : Module.finrank k (ω.orthogonal (Submodule.span k {v, w})) = 2 * n) : ∃ b : Basis (Fin (n + 1) ⊕ Fin (n + 1)) k V, IsSymplecticBasis ω b ∧ b (Sum.inl 0) = v ∧ b (Sum.inr 0) = w := by let U : Submodule k V := Submodule.span k {v, w} let W : Submodule k V := ω.orthogonal U have hAltW : ∀ x : W, (ω.restrict W) x x = 0 := fun x => hAlt x obtain ⟨bW, hbWcross, hbWleft, hbWright⟩ := Stafford38.CharacteristicPaperSymplecticBasis.exists_symplectic_basis n (ω.restrict W) hAltW hWnd hWdim have hUmemV : v ∈ U := Submodule.subset_span (by simp) have hUmemW : w ∈ U := Submodule.subset_span (by simp) have hli : LinearIndependent k (pairVectors v w) := by change LinearIndependent k ![v, w] rw [LinearIndependent.pair_iff] intro a b hab obtain ⟨ha, hb⟩ := step_lemma_3_1_plane ω hAlt hvw a b rw [hab, BilinForm.zero_left] at ha hb exact ⟨ha.symm, by simpa using hb.symm⟩ have hspan : Submodule.span k (Set.range (pairVectors v w)) = U := by congr 1 ext x simp only [Set.mem_range, Set.mem_insert_iff, Set.mem_singleton_iff] constructor · rintro ⟨i, rfl⟩ fin_cases i <;> simp [pairVectors] · rintro (rfl | rfl) · exact ⟨0, by simp [pairVectors]⟩ · exact ⟨1, by simp [pairVectors]⟩ let bU : Basis (Fin 2) k U := (Basis.span hli).map (LinearEquiv.ofEq _ _ hspan) have hbU0 : bU 0 = ⟨v, hUmemV⟩ := by apply Subtype.ext dsimp [bU] rw [Module.Basis.coe_span_apply] simp [pairVectors] have hbU1 : bU 1 = ⟨w, hUmemW⟩ := by apply Subtype.ext dsimp [bU] rw [Module.Basis.coe_span_apply] simp [pairVectors] let b : Basis (Fin (n + 1) ⊕ Fin (n + 1)) k V := ((bU.prod bW).map (Submodule.prodEquivOfIsCompl U W hcompl)).reindex (symplecticPairIndex n) have hbE0 : b (Sum.inl (0 : Fin (n + 1))) = v := by simp only [b, Module.Basis.reindex_apply, pairIndex_left0, Module.Basis.map_apply] simp [Module.Basis.prod_apply, Submodule.coe_prodEquivOfIsCompl', hbU0] have hbF0 : b (Sum.inr (0 : Fin (n + 1))) = w := by simp only [b, Module.Basis.reindex_apply, pairIndex_right0, Module.Basis.map_apply] simp [Module.Basis.prod_apply, Submodule.coe_prodEquivOfIsCompl', hbU1] have hbEs (i : Fin n) : b (Sum.inl i.succ) = (bW (Sum.inl i) : V) := by simp only [b, Module.Basis.reindex_apply, pairIndex_left_succ, Module.Basis.map_apply] simp [Module.Basis.prod_apply, Submodule.coe_prodEquivOfIsCompl'] have hbFs (i : Fin n) : b (Sum.inr i.succ) = (bW (Sum.inr i) : V) := by simp only [b, Module.Basis.reindex_apply, pairIndex_right_succ, Module.Basis.map_apply] simp [Module.Basis.prod_apply, Submodule.coe_prodEquivOfIsCompl'] have hskew (x y : V) : ω x y = -ω y x := ((show ω.IsAlt from hAlt).neg_eq y x).symm refine ⟨b, ⟨?_, ?_, ?_⟩, hbE0, hbF0⟩ · intro i j refine Fin.cases ?_ (fun i => ?_) i · refine Fin.cases ?_ (fun j => ?_) j · simpa [hbE0, hbF0] using hvw · simp only [eq_comm, Fin.succ_ne_zero, ite_false] rw [hbE0, hbFs] exact ((bW (Sum.inr j)).property v hUmemV).symm · refine Fin.cases ?_ (fun j => ?_) j · simp only [Fin.succ_ne_zero, ite_false] rw [hbEs, hbF0, hskew] simp [(bW (Sum.inl i)).property w hUmemW] · simpa [hbEs, hbFs] using hbWcross i j · intro i j refine Fin.cases ?_ (fun i => ?_) i · refine Fin.cases ?_ (fun j => ?_) j · rw [hbE0]; exact hAlt v · rw [hbE0, hbEs]; exact (bW (Sum.inl j)).property v hUmemV · refine Fin.cases ?_ (fun j => ?_) j · rw [hbEs, hbE0, hskew] simp [(bW (Sum.inl i)).property v hUmemV] · simpa [hbEs] using hbWleft i j · intro i j refine Fin.cases ?_ (fun i => ?_) i · refine Fin.cases ?_ (fun j => ?_) j · rw [hbF0]; exact hAlt w · rw [hbF0, hbFs]; simp [(bW (Sum.inr j)).property w hUmemW] · refine Fin.cases ?_ (fun j => ?_) j · rw [hbFs, hbF0, hskew] simp [(bW (Sum.inr i)).property w hUmemW] · simpa [hbFs] using hbWright i j
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Lemma 3.1, paragraph 5.** The linear map sending one symplectic basis to another, vector by vector, preserves all pairwise values of `ω` on basis vectors, hence by bilinearity preserves `ω`. -/ theorem step_lemma_3_1_transport {m : ℕ} (ω : BilinForm k V) (hAlt : ∀ x, ω x x = 0) (b₁ b₂ : Basis (Fin m ⊕ Fin m) k V) (h₁ : IsSymplecticBasis ω b₁) (h₂ : IsSymplecticBasis ω b₂) : (∀ x y, ω (b₁.equiv b₂ (Equiv.refl _) x) (b₁.equiv b₂ (Equiv.refl _) y) = ω x y) ∧ ∀ i, b₁.equiv b₂ (Equiv.refl _) (b₁ i) = b₂ i := by have hg := Stafford38.CharacteristicPaperSymplecticBasis.symplectic_basis_equiv m ω ω b₁ b₂ hAlt hAlt h₁.1 h₁.2.1 h₁.2.2 h₂.1 h₂.2.1 h₂.2.2 refine ⟨fun x y => ?_, fun i => Basis.equiv_apply b₁ i b₂ (Equiv.refl _)⟩ have h := congrArg (fun f : BilinForm k V => f x y) hg simpa [LinearMap.BilinForm.comp_apply] using h
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
Comparison notes
lemma_3_1 states the lemma for a finite-dimensional space V of dimension 2n with an antisymmetric nondegenerate form over a field of characteristic zero: v is the first vector of a symplectic basis, and for every nonzero e there is an ω-preserving linear automorphism g with g(e) = v. The proof follows the five paragraphs: step_lemma_3_1_partner, _plane, _complement, _basis, _transport. Lean calls _plane inside _complement and _basis, where the paper uses it.
Review notes
Construction — Symplectic automorphisms Φ_g and their symbols
Let be a -linear automorphism of that preserves . Since is identified with the span of the generators of , assigns to each generator a linear combination of generators. Because is presented by generators and the Weyl relations , and preserves , the elements satisfy the same relations as the . Hence the substitution extends to a unique -algebra endomorphism of [Coutinho 1995, Th. 1.3.1, Ex. 7.6.9]. Because is an automorphism, so is , and we have
Lean: Section3.phiG,
Section3.step_phi_g
/-- **The automorphism `Φ_g`.** For `g ∈ Sp(2n, k)`, given by the matrix of its rows `g(u_i) = ∑_j g_ij u_j`, `Φ_g` is the algebra automorphism of `A_n(k)` extending `u_i ↦ g(u_i)`; its inverse is built from `g⁻¹`. -/ def phiG {n : ℕ} (g : Matrix.symplecticGroup (Fin n) k) : PresentedWeyl k n ≃ₐ[k] PresentedWeyl k n := standardSymplecticAlgEquivOfInverse k g.1 (g⁻¹).1 (SymplecticGroup.mem_iff.mp g.2) (SymplecticGroup.mem_iff.mp (g⁻¹).2) (by calc g.1 * (g⁻¹).1 = ((g * g⁻¹ : Matrix.symplecticGroup (Fin n) k) : Matrix _ _ k) := rfl _ = 1 := by rw [mul_inv_cancel]; rfl) (by calc (g⁻¹).1 * g.1 = ((g⁻¹ * g : Matrix.symplecticGroup (Fin n) k) : Matrix _ _ k) := rfl _ = 1 := by rw [inv_mul_cancel]; rfl)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open StaffordSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Construction of `Φ_g` (paragraph after Lemma 3.1).** Because the `g(u_i)` satisfy the Weyl relations, `u ↦ g(u)` extends to a unique `k`-algebra endomorphism `Φ_g`; it is an automorphism with `Φ_g⁻¹ = Φ_{g⁻¹}`. -/ theorem step_phi_g {n : ℕ} (g : Matrix.symplecticGroup (Fin n) k) : (∀ i, phiG k g (freeWeylGenerator (standardForm k n) i) = freeWeylLinearCombination g.1 (freeWeylGenerator (standardForm k n)) i) ∧ (∀ f : PresentedWeyl k n →ₐ[k] PresentedWeyl k n, (∀ i, f (freeWeylGenerator (standardForm k n) i) = freeWeylLinearCombination g.1 (freeWeylGenerator (standardForm k n)) i) → f = (phiG k g).toAlgHom) ∧ (phiG k g).symm = phiG k g⁻¹ := by refine ⟨fun i => by rw [phiG_apply, standardSymplecticAlgHom_generator], ?_, ?_⟩ · intro f hf apply Stafford38.WeylUniversal.freeWeyl_algHom_ext intro i rw [hf i] change _ = phiG k g _ rw [phiG_apply, standardSymplecticAlgHom_generator] · apply AlgEquiv.ext intro z rw [phiG_symm_apply, phiG_apply]
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open StaffordSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
g is an element of Sp(2n, k), given by the matrix of its rows g(u_i) = Σ_j g_ij u_j, rather than an ω-preserving linear automorphism of V. step_phi_g: u ↦ g(u) extends to Φ_g = phiG g, the unique k-algebra endomorphism with these values on the generators, and Φ_g⁻¹ = Φ_{g⁻¹}; as printed, the extension exists because the g(u_i) satisfy the Weyl relations.
Review notes
Proposition 3.2 — Monic normal form in p
If preserves , then , where is the Gram matrix of . In particular is invertible, and inverting gives , i.e. . Since , i.e. , this is , so again preserves .
Let be the corresponding automorphism of , extending on generators and let be the matrix whose rows are the images of the generators, i.e.\ . Since sends each generator to a degree-one element, it maps into for all , and since Bernstein symbols are multiplicative, a standard monomial of degree is mapped to an element of whose class modulo is obtained by substituting in the corresponding commutative monomial. Hence .
As is invertible, is again a nonzero homogeneous polynomial of degree , so and the Bernstein symbol of is the function . Since the rows of are the coordinate vectors of the , the matrix of is ; as , so the matrix of is , i.e. . Hence the value of at is
Lean ():
Section3.proposition_3_2,
Section3.IsMonicNormalForm
/-- **Proposition 3.2.** Let `0 ≠ d ∈ A_n(k)` with `N := deg_B d ≥ 1`. There are an automorphism `Φ_g` of `A_n(k)` and `c ∈ k^×` with `d̃ := c⁻¹ Φ_g(d) = p^N + ∑_{j<N} p^j a_j`, `a_j ∈ C`, `deg_B a_j ≤ N - j` (eq. (3.1)), and `deg_B d̃ = N`. -/ theorem proposition_3_2 [CharZero k] {n : ℕ} {d : PresentedWeyl k (n + 1)} (hd : d ≠ 0) (hN : 1 ≤ bernsteinDegree k d) : ∃ (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k), c ≠ 0 ∧ IsMonicNormalForm k (bernsteinDegree k d) (c⁻¹ • phiG k g d) ∧ bernsteinDegree k (c⁻¹ • phiG k g d) = bernsteinDegree k d := by obtain ⟨hpiece, hne, hhom, -, -⟩ := bernsteinSymbol_ne_zero k hd obtain ⟨v, hv, hv0⟩ := step_3_2_nonvanishing k hhom hne hN obtain ⟨h, hpres, hhe⟩ := step_3_2_h k hv0 (.inr (0 : Fin (n + 1))) obtain ⟨g, -, hmul⟩ := step_3_2_transpose k h hpres obtain ⟨hpiece', heval, hdeg'⟩ := step_3_2_symbol k g hpiece (.inr (0 : Fin (n + 1))) rw [hmul, hhe] at heval hdeg' obtain ⟨hpiece'', hdeg'', hval⟩ := step_3_2_normalize k hpiece' (hdeg' hv) _ hv heval exact ⟨g, _, hv, step_3_2_monic k hpiece'' hval, hdeg''⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- **Equation (3.1).** `d̃ = p^N + ∑_{j<N} p^j a_j` with `a_j ∈ C` and `deg_B a_j ≤ N - j`. -/ def IsMonicNormalForm {n : ℕ} (N : ℕ) (dt : PresentedWeyl k (n + 1)) : Prop := ∃ a : ℕ → PresentedWeyl k (n + 1), (∀ j < N, a j ∈ C k n ∧ bernsteinDegree k (a j) ≤ N - j) ∧ dt = p k n ^ N + ∑ j ∈ Finset.range N, p k n ^ j * a j
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 1.** The Bernstein symbol `s` of `d` is a nonzero homogeneous polynomial of degree `N ≥ 1` on `𝕍`. A field of characteristic zero is infinite, so `s(v) ≠ 0` for some `v`; as `s(0) = 0`, `v ≠ 0`. -/ theorem step_3_2_nonvanishing [CharZero k] {n N : ℕ} {s : SymbolRing k n} (hs : s.IsHomogeneous N) (hne : s ≠ 0) (hN : 1 ≤ N) : ∃ v : PhaseVar n → k, MvPolynomial.eval v s ≠ 0 ∧ v ≠ 0 := by obtain ⟨v, hv⟩ := exists_eval_ne_zero_of_homogeneous k hs hne refine ⟨v, hv, ?_⟩ rintro rfl apply hv have hc : MvPolynomial.coeff 0 s = 0 := hs.coeff_eq_zero (by simp; omega) simpa [MvPolynomial.eval_zero', MvPolynomial.constantCoeff] using hc
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 2 (first part).** By Lemma 3.1 there is an `ω`-preserving automorphism `h` of `𝕍` with `h(e_p) = v`. -/ theorem step_3_2_h [CharZero k] {n : ℕ} {v : PhaseVar n → k} (hv : v ≠ 0) (t : PhaseVar n) : ∃ h : (PhaseVar n → k) ≃ₗ[k] (PhaseVar n → k), (∀ x y, omega k n (h x) (h y) = omega k n x y) ∧ h (phaseBasis k t) = v := by have he : phaseBasis k t ≠ 0 := by intro h have := congrFun h t simp [phaseBasis] at this exact (lemma_3_1 n (omega k n) (omega_isSymplectic k n).1 (omega_isSymplectic k n).2 (finrank_phaseSpace k n) hv).2 _ he
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 2 (second part).** If `h` preserves `ω`, then `hᵀ J h = J`; inverting and using `J⁻¹ = -J` gives `h J hᵀ = J`, so the matrix `M = h` (whose rows are the images `g(u_i)` of the generators for `g = hᵀ`) is symplectic, and `M e_p = h(e_p)`. -/ theorem step_3_2_transpose {n : ℕ} (h : (PhaseVar n → k) ≃ₗ[k] (PhaseVar n → k)) (hpres : ∀ x y, omega k n (h x) (h y) = omega k n x y) : ∃ g : Matrix.symplecticGroup (Fin n) k, g.1 = LinearMap.toMatrix' h.toLinearMap ∧ ∀ e, Matrix.mulVec g.1 e = h e := ⟨⟨_, toMatrix_mem_symplecticGroup k h.toLinearMap hpres⟩, rfl, fun e => LinearMap.toMatrix'_mulVec h.toLinearMap e⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 3.** `Φ_g` maps `B_m` into `B_m`, and modulo `B_{N-1}` it substitutes `u_i ↦ ∑_j M_ij u_j` in the commutative symbol, so the Bernstein symbol of `Φ_g(d)` is `s_g : a ↦ s(Ma)`; at `e_p` its value is `s(M e_p)`. If this value is nonzero, `deg_B Φ_g(d) = N`. -/ theorem step_3_2_symbol {n N : ℕ} (g : Matrix.symplecticGroup (Fin n) k) {d : PresentedWeyl k n} (hd : d ∈ bernsteinPiece k n N) (t : PhaseVar n) : phiG k g d ∈ bernsteinPiece k n N ∧ MvPolynomial.eval (axisPoint k t) (presentedPrincipalComponent k (@bernsteinWeight n) N (phiG k g d)) = MvPolynomial.eval (Matrix.mulVec g.1 (phaseBasis k t)) (presentedPrincipalComponent k (@bernsteinWeight n) N d) ∧ (MvPolynomial.eval (Matrix.mulVec g.1 (phaseBasis k t)) (presentedPrincipalComponent k (@bernsteinWeight n) N d) ≠ 0 → bernsteinDegree k (phiG k g d) = N) := by have hpiece := phiG_mem_bernsteinPiece k g hd have heval : MvPolynomial.eval (axisPoint k t) (presentedPrincipalComponent k (@bernsteinWeight n) N (phiG k g d)) = MvPolynomial.eval (Matrix.mulVec g.1 (phaseBasis k t)) (presentedPrincipalComponent k (@bernsteinWeight n) N d) := by rw [phiG_principal k g hd, eval_symbolLinearAlgHom_axis, mulVec_phaseBasis] refine ⟨hpiece, heval, fun hne => bernsteinDegree_eq_of_principal_ne_zero k hpiece ?_⟩ intro hzero rw [hzero, map_zero] at heval exact hne heval.symm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 4.** With `c := s(v) ≠ 0`, a central unit, `d̃ := c⁻¹ Φ_g(d)` satisfies `deg_B d̃ = N` and `σ_N^B(d̃)(e_p) = 1`. -/ theorem step_3_2_normalize {n N : ℕ} {e : PresentedWeyl k n} (he : e ∈ bernsteinPiece k n N) (hdeg : bernsteinDegree k e = N) (t : PhaseVar n) {c : k} (hc : c ≠ 0) (hval : MvPolynomial.eval (axisPoint k t) (presentedPrincipalComponent k (@bernsteinWeight n) N e) = c) : c⁻¹ • e ∈ bernsteinPiece k n N ∧ bernsteinDegree k (c⁻¹ • e) = N ∧ MvPolynomial.eval (axisPoint k t) (presentedPrincipalComponent k (@bernsteinWeight n) N (c⁻¹ • e)) = 1 := by have hpiece := (bernsteinPiece k n N).smul_mem c⁻¹ he have heval : MvPolynomial.eval (axisPoint k t) (presentedPrincipalComponent k (@bernsteinWeight n) N (c⁻¹ • e)) = 1 := by rw [map_smul, MvPolynomial.smul_eval, hval, inv_mul_cancel₀ hc] refine ⟨hpiece, ?_, heval⟩ · subst hdeg refine bernsteinDegree_eq_of_principal_ne_zero k hpiece ?_ intro hzero rw [hzero, map_zero] at heval exact zero_ne_one heval
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 5 (decomposition).** Using the normal form `A_n(k) = ⊕_j p^j C` (Convention 2.2), `d̃ = ∑_{j=0}^N p^j a_j` with `a_j ∈ C`; the Bernstein filtration is compatible with this decomposition, so `deg_B a_j ≤ N - j`, in particular `a_N ∈ C ∩ B_0 = k`. -/ theorem step_3_2_decomposition {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hdt : dt ∈ bernsteinPiece k (n + 1) N) : ∃ a : ℕ → PresentedWeyl k (n + 1), (∀ j, a j ∈ C k n ∧ bernsteinDegree k (a j) ≤ N - j) ∧ dt = ∑ j ∈ Finset.range (N + 1), p k n ^ j * a j ∧ ∃ c : k, a N = algebraMap k _ c := by obtain ⟨a, ha, hdec⟩ := exists_bounded_decomposition k n N hdt refine ⟨a, ha, hdec, eq_algebraMap_of_mem_bernsteinPiece_zero k ?_⟩ rw [← bernsteinDegree_le_iff] simpa using (ha N).2
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 5 (leading coefficient).** The symbol is `σ_N^B(d̃) = ∑_{j + deg_B a_j = N} σ(p)^j σ(a_j)`. For `j < N` the factor `σ_{N-j}^B(a_j)` has positive degree and contains no `ξ₁` (`a_j ∈ C`), so it vanishes at `e_p`; evaluating at `e_p` gives `a_N = σ_N^B(d̃)(e_p) = 1`. -/ theorem step_3_2_leading {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (a : ℕ → PresentedWeyl k (n + 1)) (ha : ∀ j, a j ∈ C k n ∧ bernsteinDegree k (a j) ≤ N - j) (hdec : dt = ∑ j ∈ Finset.range (N + 1), p k n ^ j * a j) (c : k) (hc : a N = algebraMap k _ c) (hval : MvPolynomial.eval (axisPoint k (.inr (0 : Fin (n + 1)))) (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) = 1) : a N = 1 := by have hsum : MvPolynomial.eval (axisPoint k (.inr (0 : Fin (n + 1)))) (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) = ∑ j ∈ Finset.range (N + 1), MvPolynomial.coeff (Finsupp.single (.inr (0 : Fin (n + 1))) (N - j)) (presentedNormalFormLinearEquiv k (n + 1) (a j)) := by rw [hdec, map_sum, map_sum] apply Finset.sum_congr rfl intro j hj have hjN : j ≤ N := Nat.lt_succ_iff.mp (Finset.mem_range.mp hj) rw [(principal_bernstein_p_pow_mul k hjN ((bernsteinDegree_le_iff k (a j)).mp (ha j).2)).2, map_mul, map_pow, MvPolynomial.eval_X, eval_ep_principal] simp [axisPoint] rw [hsum, Finset.sum_range_succ, Finset.sum_eq_zero, zero_add, Nat.sub_self, Finsupp.single_zero, hc, Algebra.algebraMap_eq_smul_one, map_smul, presentedNormalFormLinearEquiv_one] at hval · have hc1 : c = 1 := by simpa using hval rw [hc, hc1, map_one] · intro j hj have hjN := Finset.mem_range.mp hj by_contra hne have := normalForm_C_no_p k (ha j).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2, paragraph 5.** Hence `d̃ = p^N + ∑_{j<N} p^j a_j` with `a_j ∈ C` and `deg_B a_j ≤ N - j`. -/ theorem step_3_2_monic {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hdt : dt ∈ bernsteinPiece k (n + 1) N) (hval : MvPolynomial.eval (axisPoint k (.inr (0 : Fin (n + 1)))) (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) = 1) : IsMonicNormalForm k N dt := by obtain ⟨a, ha, hdec, c, hc⟩ := step_3_2_decomposition k hdt have hN := step_3_2_leading k a ha hdec c hc hval refine ⟨a, fun j _ => ha j, ?_⟩ rw [hdec, Finset.sum_range_succ, hN, mul_one] exact add_comm _ _
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- In (3.1), `N = deg_B d̃`: `d̃ ∈ B_N` and its Bernstein symbol has the coefficient `1` at `ξ₁^N`. -/ theorem bernsteinDegree_eq_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : bernsteinDegree k dt = N := by have hd := isPBWMonicAt_of_isMonicNormalForm k h refine bernsteinDegree_eq_of_principal_ne_zero k hd.1 fun h0 => ?_ have hc : AddMonoidAlgebra.coeff (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) (Finsupp.single (.inr (0 : Fin (n + 1))) N) = 1 := by rw [coeff_presentedPrincipalComponent, ite_eq_left (by simp [monomialWeight, bernsteinWeight])] exact hd.2 rw [h0] at hc exact zero_ne_one hc
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.1.** Let `(𝕍, ω)` be a symplectic `k`-space of dimension `2n` (`ω` antisymmetric and nondegenerate) and `0 ≠ v ∈ 𝕍`. Then `v` is the first vector `e_1` of a symplectic basis of `𝕍`; consequently, for every nonzero `e ∈ 𝕍` there is a `k`-linear automorphism `g` preserving `ω` with `g(e) = v`. -/ theorem lemma_3_1 [CharZero k] [Module.Finite k V] (n : ℕ) (ω : BilinForm k V) (hanti : ∀ x y, ω x y = -ω y x) (hnd : ω.Nondegenerate) (hdim : Module.finrank k V = 2 * n) {v : V} (hv : v ≠ 0) : (∃ (hn : 0 < n) (b : Basis (Fin n ⊕ Fin n) k V), IsSymplecticBasis ω b ∧ b (Sum.inl ⟨0, hn⟩) = v) ∧ ∀ e : V, e ≠ 0 → ∃ g : V ≃ₗ[k] V, (∀ x y, ω (g x) (g y) = ω x y) ∧ g e = v := by have hAlt := alternating_of_antisymmetric ω hanti -- every nonzero vector is the first vector of a symplectic basis have hcomplete : ∀ u : V, u ≠ 0 → ∃ m : ℕ, n = m + 1 ∧ ∃ b : Basis (Fin (m + 1) ⊕ Fin (m + 1)) k V, IsSymplecticBasis ω b ∧ b (Sum.inl 0) = u := by intro u hu obtain ⟨w, huw⟩ := step_lemma_3_1_partner ω hnd hu obtain ⟨-, hcompl, hWnd, hWdim⟩ := step_lemma_3_1_complement ω hAlt hnd huw obtain ⟨m, rfl⟩ : ∃ m, n = m + 1 := ⟨n - 1, by omega⟩ refine ⟨m, rfl, ?_⟩ obtain ⟨b, hb, hb0, -⟩ := step_lemma_3_1_basis m ω hAlt huw hcompl hWnd (by omega) exact ⟨b, hb, hb0⟩ obtain ⟨m, rfl, b, hb, hb0⟩ := hcomplete v hv refine ⟨⟨Nat.succ_pos m, b, hb, hb0⟩, fun e he => ?_⟩ obtain ⟨m', hm', b', hb', hb'0⟩ := hcomplete e he obtain rfl : m' = m := by omega obtain ⟨hpres, happ⟩ := step_lemma_3_1_transport ω hAlt b' b hb' hb exact ⟨b'.equiv b (Equiv.refl _), hpres, by rw [← hb'0, happ, hb0]⟩
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.Paper.Section2Source ↗universe u vSource ↗
variable {k : Type u} {V : Type v} [Field k] [AddCommGroup V] [Module k V]
/-- **Proposition 3.2, paragraph 5 (the decomposition).** Using `A_n(k) = ⊕_j p^j C` (Convention 2.2: existence and uniqueness of the normal form), `d ∈ B_N` is `d = ∑_{j=0}^N p^j a_j` with `a_j ∈ C`; the Bernstein filtration is compatible with the decomposition, so `deg_B a_j ≤ N - j`. -/ theorem exists_bounded_decomposition (n N : ℕ) {d : PresentedWeyl k (n + 1)} (hd : d ∈ bernsteinPiece k (n + 1) N) : ∃ a : ℕ → PresentedWeyl k (n + 1), (∀ j, a j ∈ C k n ∧ bernsteinDegree k (a j) ≤ N - j) ∧ d = ∑ j ∈ Finset.range (N + 1), p k n ^ j * a j := by -- The normal form of Convention 2.2. obtain ⟨L, a, haC, hda⟩ := convention_2_2_exists k n d -- Its coefficients are those bounded by the Bernstein filtration. obtain ⟨b, hb, hdb⟩ := step_bounded_coefficients k n N hd let M := max L (N + 1) let a' : ℕ → PresentedWeyl k (n + 1) := fun j => if j < L then a j else 0 let b' : ℕ → PresentedWeyl k (n + 1) := fun j => if j < N + 1 then b j else 0 have ha'C : ∀ j, a' j ∈ C k n := fun j => by by_cases hj : j < L · simp only [a', hj, ↓reduceIte]; exact haC j · simp only [a', hj, ↓reduceIte]; exact Subalgebra.zero_mem _ have hb'C : ∀ j, b' j ∈ C k n := fun j => by by_cases hj : j < N + 1 · simp only [b', hj, ↓reduceIte]; exact (hb j).1 · simp only [b', hj, ↓reduceIte]; exact Subalgebra.zero_mem _ have hsum : ∑ j ∈ Finset.range M, p k n ^ j * (a' j - b' j) = 0 := by simp only [mul_sub, Finset.sum_sub_distrib] rw [sum_range_pad _ a (le_max_left L (N + 1)), sum_range_pad _ b (le_max_right L (N + 1)), ← hda, ← hdb, sub_self] have huniq := convention_2_2_unique k n M (fun j => a' j - b' j) (fun j => Subalgebra.sub_mem _ (ha'C j) (hb'C j)) hsum have hab : ∀ j, a' j = b' j := fun j => by by_cases hj : j < M · exact sub_eq_zero.mp (huniq j hj) · have h1 : ¬ j < L := fun h => hj (lt_of_lt_of_le h (le_max_left _ _)) have h2 : ¬ j < N + 1 := fun h => hj (lt_of_lt_of_le h (le_max_right _ _)) simp only [a', b', h1, h2, ↓reduceIte] refine ⟨a', fun j => ⟨ha'C j, ?_⟩, ?_⟩ · rw [hab j] by_cases hj : j < N + 1 · simp only [b', hj, ↓reduceIte]; exact (hb j).2 · simp only [b', hj, ↓reduceIte] exact (bernsteinDegree_le_iff k 0).mpr (Submodule.zero_mem _) · rw [hdb] apply Finset.sum_congr rfl intro j hj rw [hab j] simp only [b', Finset.mem_range.mp hj, ↓reduceIte]
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open AlgebraicAnalysis.OreRightQuotientSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
/-- **Convention 2.2, normal form (existence).** Every element of `A_n(k)` can be written as `∑_j p^j c_j` with `c_j ∈ C`. -/ theorem convention_2_2_exists (n : ℕ) (d : PresentedWeyl k (n + 1)) : ∃ (L : ℕ) (a : ℕ → PresentedWeyl k (n + 1)), (∀ j, a j ∈ C k n) ∧ d = ∑ j ∈ Finset.range L, p k n ^ j * a j := by let b := pairStage_rightOrePBWBasis (B := IteratedPairStage k n) let c := b.repr (presentedToIterated k (n + 1) d) let L := c.support.sup id + 1 have hsub : c.support ⊆ Finset.range L := by intro j hj exact Finset.mem_range.mpr (Nat.lt_succ_of_le (Finset.le_sup (f := id) hj)) refine ⟨L, fun j => coordinateStageToPresented k n (c j).unop, fun j => (C_eq_range k n).symm ▸ ⟨_, rfl⟩, ?_⟩ have hd : presentedToIterated k (n + 1) d = ∑ j ∈ Finset.range L, c j • rightPBWMonomial (coordinateDerivation : OreDivisionDerivation (CoordinateStage (B := IteratedPairStage k n))) j := by conv_lhs => rw [← b.linearCombination_repr (presentedToIterated k (n + 1) d)] rw [Finsupp.linearCombination_apply, Finsupp.sum_of_support_subset _ hsub _ (by simp)] apply Finset.sum_congr rfl intro j _ simp [b, pairStage_rightOrePBWBasis] rfl have hback : d = stageToPresented k n (presentedToIterated k (n + 1) d) := (DFunLike.congr_fun (iteratedToPresented_comp_presentedToIterated k (n + 1)) d).symm rw [hback, hd, map_sum] apply Finset.sum_congr rfl intro j _ exact stageToPresented_rightTerm k n j (c j).unop
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
Comparison notes
proposition_3_2 states the proposition for A_{n+1}(k), k of characteristic zero: for 0 ≠ d with N = deg_B d ≥ 1 there are g ∈ Sp(2(n+1), k) and c ≠ 0 with c⁻¹Φ_g(d) in the normal form (3.1) (IsMonicNormalForm: p^N + Σ_{j<N} p^j a_j, a_j ∈ C, deg_B a_j ≤ N − j) and deg_B(c⁻¹Φ_g(d)) = deg_B d. The proof follows the five paragraphs, one step each, and the fifth paragraph in three steps; step_3_2_h applies Lemma 3.1, and step_3_2_decomposition the normal form of Convention 2.2 with the degree bound of Section3.exists_bounded_decomposition. Section3.bernsteinDegree_eq_of_isMonicNormalForm shows the last sentence for every d̃ in the form (3.1): deg_B d̃ = N.
Review notes
Remark 3.3 — Transport back along Φ_g
Lean (): Section3.remark_3_3,
Section3.remark_3_3_of_Q_eq_zero
We will now define the modules
We now use the order filtration on (where coordinates have degree zero and derivatives have degree one) to characterize , the order principal symbol of and show that annihilates the graded module .
/-- **Remark 3.3.** Since `Φ_g` is a ring automorphism and `c` a central unit, `1 = d̃R + x₁^N d̃S` for `d̃ = c⁻¹Φ_g(d)` transports back to `1 = d (c⁻¹Φ_g^{-1}(R)) + F d (c⁻¹Φ_g^{-1}(S))` with `F = Φ_g^{-1}(x₁)^N`. -/ theorem remark_3_3 {n N : ℕ} (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k) (d R S : PresentedWeyl k (n + 1)) (h : (1 : PresentedWeyl k (n + 1)) = (c⁻¹ • phiG k g d) * R + x₁ k n ^ N * (c⁻¹ • phiG k g d) * S) : (1 : PresentedWeyl k (n + 1)) = d * (c⁻¹ • (phiG k g).symm R) + (phiG k g).symm (x₁ k n) ^ N * d * (c⁻¹ • (phiG k g).symm S) := by have ht := congrArg (phiG k g).symm h rw [map_one] at ht rw [ht, map_add, map_mul, map_mul, map_mul, map_pow, map_smul, AlgEquiv.symm_apply_apply, smul_mul_assoc, mul_smul_comm, mul_smul_comm, smul_mul_assoc, mul_smul_comm]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Remark 3.3 with (3.2).** If `Q = A/𝓘 = 0` for `d̃ = c⁻¹Φ_g(d)`, then `1 = dR' + F d S'` with `F = ℓ^N` for the linear symplectic coordinate `ℓ = Φ_g^{-1}(x₁)`. -/ theorem remark_3_3_of_Q_eq_zero {n N : ℕ} (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k) (d : PresentedWeyl k (n + 1)) (hQ : Subsingleton (Q k N (c⁻¹ • phiG k g d))) : ∃ ell R S : PresentedWeyl k (n + 1), Stafford38FixedSourceChallenge.IsLinearWeylCoordinate k n ell ∧ (1 : PresentedWeyl k (n + 1)) = d * R + ell ^ N * d * S := by obtain ⟨R, S, hRS⟩ := exists_eq_of_Q_eq_zero k hQ exact ⟨_, _, _, phiG_symm_x₁_isLinearWeylCoordinate k g, remark_3_3 k g c d R S hRS⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- **Section 3 for the final assembly.** For `0 ≠ d ∈ A_n(k)`, either `deg_B d = 0` and `d` is a unit with `1 = dR + FdS` (opening paragraph), or `N = deg_B d ≥ 1` and Proposition 3.2 gives the normal form `d̃ = c⁻¹Φ_g(d)` of (3.1) with `deg_B d̃ = N`, for which `Q = A/𝓘 = 0` yields `1 = dR + ℓ^N dS` with a linear symplectic coordinate `ℓ` (Remark 3.3). -/ theorem section_3_reduction [CharZero k] {n : ℕ} {d : PresentedWeyl k (n + 1)} (hd : d ≠ 0) : (bernsteinDegree k d = 0 ∧ Stafford.Reduction.Stafford38 d) ∨ (1 ≤ bernsteinDegree k d ∧ ∃ (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k), c ≠ 0 ∧ IsMonicNormalForm k (bernsteinDegree k d) (c⁻¹ • phiG k g d) ∧ bernsteinDegree k (c⁻¹ • phiG k g d) = bernsteinDegree k d ∧ (Subsingleton (Q k (bernsteinDegree k d) (c⁻¹ • phiG k g d)) → ∃ ell R S : PresentedWeyl k (n + 1), Stafford38FixedSourceChallenge.IsLinearWeylCoordinate k n ell ∧ (1 : PresentedWeyl k (n + 1)) = d * R + ell ^ bernsteinDegree k d * d * S)) := by by_cases hN : bernsteinDegree k d = 0 · obtain ⟨c, -, -, hcert⟩ := step_scalar_case k hd hN exact Or.inl ⟨hN, 1, _, 0, hcert⟩ · obtain ⟨g, c, hc, hform, hdeg⟩ := proposition_3_2 k hd (Nat.one_le_iff_ne_zero.mpr hN) exact Or.inr ⟨Nat.one_le_iff_ne_zero.mpr hN, g, c, hc, hform, hdeg, remark_3_3_of_Q_eq_zero k g c d⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
remark_3_3: if 1 = d̃R + x₁^N d̃S for d̃ = c⁻¹Φ_g(d), then 1 = d(c⁻¹Φ_g⁻¹(R)) + Φ_g⁻¹(x₁)^N d (c⁻¹Φ_g⁻¹(S)). remark_3_3_of_Q_eq_zero combines it with (3.2): Q = 0 gives the identity with F = ℓ^N, ℓ = Φ_g⁻¹(x₁) a linear symplectic coordinate. section_3_reduction collects the scalar case, Proposition 3.2 and this remark for the proof of Theorem 1.1.
Review notes
Equation 3.2 — Canonical right ideal and quotient
We will now define the modules
/-- **Equation (3.2), the ideal.** `𝓘 = d̃A_n + x₁^N d̃A_n`, a right ideal. -/ abbrev idealI {n : ℕ} (N : ℕ) (dt : PresentedWeyl k (n + 1)) : RightIdeal (PresentedWeyl k (n + 1)) := canonicalRightIdeal (x₁ k n) dt N
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Equation (3.2), the module.** `Q = A_n/𝓘`, a right `A_n`-module. -/ abbrev Q {n : ℕ} (N : ℕ) (dt : PresentedWeyl k (n + 1)) := RightQuotient (idealI k N dt)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
idealI N d̃ is the right ideal d̃A_n + x₁^N d̃A_n and Q its quotient A_n/𝓘, a right A_n-module, as in (3.2).
Review notes
Definition 3.4 — Monic of degree N in ξ₁
Section3.IsMonicInXi1
/-- **Definition 3.4.** `P = ∑_{i=0}^N c_i ξ₁^i ∈ k[ξ_1, …, ξ_n]`, with `c_i ∈ k[ξ_2, …, ξ_n]`, is monic of degree `N` in `ξ₁` if `c_N = 1`. Here `finSuccEquiv` writes `P` as a polynomial in `ξ₁ = X 0` with coefficients in the remaining variables. -/ def IsMonicInXi1 {n : ℕ} (N : ℕ) (P : MvPolynomial (Fin (n + 1)) k) : Prop := (MvPolynomial.finSuccEquiv k n P).natDegree ≤ N ∧ (MvPolynomial.finSuccEquiv k n P).coeff N = 1
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
IsMonicInXi1 N P: writing P ∈ k[ξ_1, …, ξ_n] as a polynomial in ξ₁ with coefficients in k[ξ_2, …, ξ_n] (finSuccEquiv), its coefficient c_N is 1.
Review notes
Lemma 3.5 — Order symbol P(ξ), monic in ξ₁
The monomial occurs in with coefficient . Every other order- monomial comes from some with and (hence involving no further factor of ), so it carries a positive power of some transverse , . Such a monomial contributes only to coefficients with , never to ; hence receives no contribution besides the coefficient of itself, giving identically, i.e. is monic of degree in . Evaluating at then gives . ∎
Section3.lemma_3_5
/-- **Lemma 3.5.** Let `d̃ = p^N + ∑_{j<N} p^j a_j` with `a_j ∈ C`, `deg_B a_j ≤ N - j` as in (3.1), and `N = deg_B d̃`. Then `d̃` has order `N`, and its order principal symbol is a polynomial in `ξ` alone, `σ_N^ord(d̃) = P(ξ)` with `P ∈ k[ξ_1, …, ξ_n]`, `deg P = N` (eq. (3.3)); `P` is homogeneous, monic of degree `N` in `ξ₁`, and `P(1, 0, …, 0) = 1` (eq. (3.4)). -/ theorem lemma_3_5 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : (dt ∈ orderPiece k (n + 1) N ∧ ∀ j < N, dt ∉ orderPiece k (n + 1) j) ∧ ∃ P : MvPolynomial (Fin (n + 1)) k, orderSymbol k N dt = MvPolynomial.rename Sum.inr P ∧ P.IsHomogeneous N ∧ P.totalDegree = N ∧ IsMonicInXi1 k N P ∧ MvPolynomial.eval (fun i => if i = 0 then 1 else 0) P = 1 := by obtain ⟨hF, hfibre, hhom⟩ := step_3_5_fibre k hdeg obtain ⟨P, hsP, hP⟩ := step_3_5_fibre_polynomial k hfibre hhom have hpure := step_3_5_pure_coefficient k hform have hne : orderSymbol k N dt ≠ 0 := by intro h rw [h] at hpure simp at hpure have hcoeff : MvPolynomial.coeff (Finsupp.single 0 N) P = 1 := by have h := MvPolynomial.coeff_rename_mapDomain (Sum.inr : Fin (n + 1) → Fin (n + 1) ⊕ Fin (n + 1)) Sum.inr_injective P (Finsupp.single 0 N) rw [Finsupp.mapDomain_single, ← hsP] at h simpa [MvPolynomial.coeff] using h.symm.trans hpure obtain ⟨hmonic, htot, heval⟩ := step_3_5_monic k hP hcoeff exact ⟨step_3_5_order k hF hne, P, hsP, hP, htot, hmonic, heval⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- **Lemma 3.5, paragraph 1.** Since `deg_B d̃ = N`, every PBW monomial `x^α ∂^β` of `d̃` has `|α| + |β| ≤ N`; a monomial contributing to the order-`N` part has `|β| = N`, hence `|α| = 0`. So `d̃ ∈ F_N` and its order-`N` part is a `k`-linear combination of `∂^β` with `|β| = N`: `σ_N^ord(d̃)` involves no `x_i` and is homogeneous of degree `N`. -/ theorem step_3_5_fibre {n N : ℕ} {dt : PresentedWeyl k n} (hdeg : bernsteinDegree k dt = N) : dt ∈ orderPiece k n N ∧ IsFibreOnly k (orderSymbol k N dt) ∧ (orderSymbol k N dt).IsHomogeneous N := by have hB : dt ∈ bernsteinPiece k n N := (bernsteinDegree_le_iff k dt).mp hdeg.le have hfibre : IsFibreOnly k (orderSymbol k N dt) := by intro m hm i change AddMonoidAlgebra.coeff (orderSymbol k N dt) m ≠ 0 at hm rw [coeff_presentedPrincipalComponent] at hm by_cases horder : monomialWeight (@orderWeight n) m = N · rw [ite_eq_left horder] at hm exact coordinateExponent_eq_zero_of_order_eq_of_bernstein_le m horder ((mem_presentedWeightPiece k (@bernsteinWeight n) N dt).mp hB m hm) i · rw [ite_eq_right horder] at hm exact (hm rfl).elim refine ⟨bernsteinPiece_le_orderPiece k n N hB, hfibre, ?_⟩ apply isHomogeneous_of_orderHomogeneous_of_isFibreOnly _ hfibre exact MvPolynomial.weightedHomogeneousComponent_isWeightedHomogeneous (w := @orderWeight n) (n := N) (presentedNormalFormLinearEquiv k n dt)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5, paragraph 1 (conclusion).** A symbol without `x_i` is a polynomial `P(ξ)` in the fibre variables alone, homogeneous of the same degree. -/ theorem step_3_5_fibre_polynomial {n N : ℕ} {s : SymbolRing k n} (hfibre : IsFibreOnly k s) (hs : s.IsHomogeneous N) : ∃ P : MvPolynomial (Fin n) k, s = MvPolynomial.rename Sum.inr P ∧ P.IsHomogeneous N := by obtain ⟨P, hP⟩ := MvPolynomial.exists_rename_eq_of_vars_subset_range s Sum.inr Sum.inr_injective (by intro i hi rw [Finset.mem_coe, MvPolynomial.mem_vars_iff_mem_support] at hi obtain ⟨m, hm, hmi⟩ := hi rcases i with i | i · exact absurd (hfibre m (MvPolynomial.mem_support_iff.mp hm) i) (Finsupp.mem_support_iff.mp hmi) · exact ⟨i, rfl⟩) refine ⟨P, hP.symm, ?_⟩ rw [← MvPolynomial.IsHomogeneous.rename_isHomogeneous_iff Sum.inr_injective, hP] exact hs
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- `d̃` has order exactly `N`: `d̃ ∈ F_N` and `d̃ ∉ F_j` for `j < N`, since its order-`N` symbol is nonzero. -/ theorem step_3_5_order {n N : ℕ} {dt : PresentedWeyl k n} (hF : dt ∈ orderPiece k n N) (hne : orderSymbol k N dt ≠ 0) : dt ∈ orderPiece k n N ∧ ∀ j < N, dt ∉ orderPiece k n j := ⟨hF, fun _ hj hmem => hne (presentedPrincipalComponent_eq_zero_of_mem_of_lt k _ dt hmem hj)⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5, paragraph 2 (the terms `p^j a_j`, `j < N`).** Since `a_j ∈ C` involves no further factor of `p = ∂₁`, every order-`N` monomial of `p^j a_j` has `∂₁`-exponent exactly `j < N` (and so carries a positive power of some transverse `∂_i`); it contributes only to the coefficient `c_j` of `ξ₁^j`, never to `c_N` (tex lines 494–500). The monomials are read in the order symbol, `σ_N^ord(p^j a_j) = ξ₁^j σ_{N-j}^ord(a_j)`, whose monomials are the order-`N` PBW monomials of `p^j a_j`. -/ theorem step_3_5_lower_term {n j N : ℕ} (hj : j < N) {a : PresentedWeyl k (n + 1)} (haC : a ∈ C k n) (ha : bernsteinDegree k a ≤ N - j) (m : PhaseVar (n + 1) →₀ ℕ) (hm : MvPolynomial.coeff m (orderSymbol k N (p k n ^ j * a)) ≠ 0) : m (.inr 0) = j := by have haF : a ∈ orderPiece k (n + 1) (N - j) := bernsteinPiece_le_orderPiece k (n + 1) _ ((bernsteinDegree_le_iff k a).mp ha) rw [orderSymbol, principal_order_p_pow_mul k hj.le haF] at hm simp only [MvPolynomial.coeff] at hm rw [MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial_mul'] at hm split_ifs at hm with hle · rw [one_mul, coeff_presentedPrincipalComponent] at hm split_ifs at hm swap · exact absurd rfl hm have h0 := normalForm_C_no_p k haC _ hm have h1 := Finsupp.le_def.mp hle (.inr 0) simp only [Finsupp.coe_tsub, Pi.sub_apply, Finsupp.single_eq_same] at h0 h1 omega · exact absurd rfl hm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5, paragraph 2 (the coefficient `c_N`).** The monomial `p^N = ∂₁^N` occurs with coefficient `1`, and by `step_3_5_lower_term` no other term contributes to `c_N`: the coefficient of `ξ₁^N` in `σ_N^ord(d̃)` is `1` (tex lines 494–503). -/ theorem step_3_5_pure_coefficient {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) : MvPolynomial.coeff (Finsupp.single (.inr (0 : Fin (n + 1))) N) (orderSymbol k N dt) = 1 := by obtain ⟨a, ha, rfl⟩ := hform obtain ⟨-, hσN⟩ := p_pow_piece_principal k (@orderWeight (n + 1)) rfl N rw [orderSymbol, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj by_contra hne have := step_3_5_lower_term k hjN (ha j hjN).1 (ha j hjN).2 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5, paragraph 2 (conclusion).** If `P` is homogeneous of degree `N` and its coefficient of `ξ₁^N` is `1`, then `c_N = 1` identically, so `P` is monic of degree `N` in `ξ₁` (Definition 3.4), `deg P = N`, and evaluating at `(1, 0, …, 0)` gives `P(1, 0, …, 0) = 1`. -/ theorem step_3_5_monic {n N : ℕ} {P : MvPolynomial (Fin (n + 1)) k} (hP : P.IsHomogeneous N) (hcoeff : MvPolynomial.coeff (Finsupp.single 0 N) P = 1) : IsMonicInXi1 k N P ∧ P.totalDegree = N ∧ MvPolynomial.eval (fun i => if i = 0 then 1 else 0) P = 1 := by have hne : P ≠ 0 := by rintro rfl simp at hcoeff refine ⟨⟨?_, ?_⟩, hP.totalDegree hne, ?_⟩ · rw [MvPolynomial.natDegree_finSuccEquiv] exact (MvPolynomial.degreeOf_le_totalDegree P 0).trans hP.totalDegree_le · ext m rw [MvPolynomial.finSuccEquiv_coeff_coeff] by_cases hm : m = 0 · subst m have h0 : (Finsupp.cons N (0 : Fin n →₀ ℕ)) = Finsupp.single 0 N := by ext i refine Fin.cases ?_ (fun i => ?_) i <;> simp rw [h0] simpa using hcoeff · rw [MvPolynomial.coeff_one, ite_eq_right (Ne.symm hm)] apply hP.coeff_eq_zero rw [degree_cons] have : m.degree ≠ 0 := fun h => hm ((Finsupp.degree_eq_zero_iff m).mp h) omega · have hsym := eval_axis_eq_one_of_pureCoefficient_one (k := k) (n := n + 1) (N := N) (.inr (0 : Fin (n + 1))) (P := MvPolynomial.rename Sum.inr P) ((MvPolynomial.IsHomogeneous.rename_isHomogeneous_iff Sum.inr_injective).mpr hP) (by have h := MvPolynomial.coeff_rename_mapDomain (Sum.inr : Fin (n + 1) → Fin (n + 1) ⊕ Fin (n + 1)) Sum.inr_injective P (Finsupp.single 0 N) rw [Finsupp.mapDomain_single] at h simpa [MvPolynomial.coeff] using h.trans hcoeff) have hpt : (axisPoint k (.inr (0 : Fin (n + 1))) ∘ Sum.inr : Fin (n + 1) → k) = fun i => if i = 0 then 1 else 0 := by funext i simp [axisPoint] rwa [MvPolynomial.eval_rename, hpt] at hsym
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- Equation (3.4) read in `k[x, ξ]`: `σ_N^ord(d̃)` takes the value `1` at the covector `ξ = (1, 0, …, 0)` over every base point `x = 0`, i.e. at `axisPoint (.inr 0)`. -/ theorem lemma_3_5_eval_axis {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : MvPolynomial.eval (axisPoint k (.inr (0 : Fin (n + 1)))) (orderSymbol k N dt) = 1 := by obtain ⟨-, P, hsP, -, -, -, heval⟩ := lemma_3_5 k hform hdeg have hpt : (axisPoint k (.inr (0 : Fin (n + 1))) ∘ Sum.inr : Fin (n + 1) → k) = fun i => if i = 0 then 1 else 0 := by funext i simp [axisPoint] rw [hsP, MvPolynomial.eval_rename, hpt] exact heval
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
lemma_3_5: for d̃ in the form (3.1) with deg_B d̃ = N, d̃ ∈ F_N and d̃ ∉ F_j for j < N (order N), and there is P ∈ k[ξ] with σ_N^ord(d̃) = P(ξ), P homogeneous of degree N, monic of degree N in ξ₁ and P(1, 0, …, 0) = 1. The proof follows the two paragraphs; Lean proves the clause “d̃ has order N” (step_3_5_order) after the symbol computation, from the nonzero order-N symbol. lemma_3_5_eval_axis reads (3.4) in k[x, ξ].
Review notes
Proposition 3.6 — P annihilates gr Q
/-- **Proposition 3.6.** Equip `Q` with the order filtration `Q_m = qF_m` induced by the cyclic generator `q = [1]`, and write `P = σ_N^ord(d̃)`. Then `P ∈ Ann_{k[x,ξ]} gr Q` (eq. (3.5)). -/ theorem proposition_3_6 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : orderSymbol k N dt ∈ Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt) := by obtain ⟨⟨hF, -⟩, -⟩ := lemma_3_5 k hform hdeg rw [Module.mem_annihilator] intro m obtain ⟨f, rfl⟩ := (step_3_6_generated k dt).2.2 m rw [smul_comm, step_3_6_kills k hF, smul_zero]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
More Lean proofs and definitions
/-- **Proposition 3.6, proof, first sentence.** `Q_•` is good because `Q` is cyclic: `Q_m = qF_m` is a compatible filtration of `Q` (Definition 2.4), and `gr Q` is finite over `k[x, ξ]`, generated by `q̄`. -/ theorem step_3_6_generated {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : IsCompatibleFiltration (orderFiltration k (n + 1)) (cyclicFiltration (orderFiltration k (n + 1)) (qClass k (idealI k N dt))) ∧ Module.Finite (SymbolRing k (n + 1)) (grQ k N dt) ∧ ∀ m : grQ k N dt, ∃ f : SymbolRing k (n + 1), f • qBar k (idealI k N dt) = m := ⟨cyclicFiltration_order_isCompatible k _, grCyclic_good k _, grCyclic_generated k _⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.6, proof, second sentence.** `q d̃ = [d̃] = 0` in `Q` because `d̃ ∈ 𝓘`, and `d̃ ∈ F_N`, so `q d̃ ∈ Q_N`; its image `q̄ · P` in `Q_N/Q_{N-1}` is zero. -/ theorem step_3_6_kills {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hF : dt ∈ orderPiece k (n + 1) N) : orderSymbol k N dt • qBar k (idealI k N dt) = 0 := by have h := smul_orderAssociatedGradedOf_mk_eq_of_mul k (idealI k N dt) (orderPieceOne (n := n + 1) k) (⟨dt, hF⟩ : orderPiece k (n + 1) N) change orderSymbol k N dt • qBar k (idealI k N dt) = _ at h rw [h] have hzero : orderPieceToQuotientGraded k (idealI k N dt) (0 + N) ⟨(orderPieceOne (n := n + 1) k : PresentedWeyl k (n + 1)) * dt, mul_mem_orderPiece k (orderPieceOne (n := n + 1) k).property hF⟩ = 0 := by have hpiece : orderPieceToQuotientPiece k (idealI k N dt) (0 + N) ⟨(orderPieceOne (n := n + 1) k : PresentedWeyl k (n + 1)) * dt, mul_mem_orderPiece k (orderPieceOne (n := n + 1) k).property hF⟩ = 0 := by apply Subtype.ext change Submodule.Quotient.mk ((1 : PresentedWeyl k (n + 1)) * dt) = 0 rw [one_mul, Submodule.Quotient.mk_eq_zero] exact firstGenerator_mem (x₁ k n) dt N rw [orderPieceToQuotientGraded, LinearMap.comp_apply, hpiece, map_zero] rw [hzero] exact map_zero _
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5.** Let `d̃ = p^N + ∑_{j<N} p^j a_j` with `a_j ∈ C`, `deg_B a_j ≤ N - j` as in (3.1), and `N = deg_B d̃`. Then `d̃` has order `N`, and its order principal symbol is a polynomial in `ξ` alone, `σ_N^ord(d̃) = P(ξ)` with `P ∈ k[ξ_1, …, ξ_n]`, `deg P = N` (eq. (3.3)); `P` is homogeneous, monic of degree `N` in `ξ₁`, and `P(1, 0, …, 0) = 1` (eq. (3.4)). -/ theorem lemma_3_5 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : (dt ∈ orderPiece k (n + 1) N ∧ ∀ j < N, dt ∉ orderPiece k (n + 1) j) ∧ ∃ P : MvPolynomial (Fin (n + 1)) k, orderSymbol k N dt = MvPolynomial.rename Sum.inr P ∧ P.IsHomogeneous N ∧ P.totalDegree = N ∧ IsMonicInXi1 k N P ∧ MvPolynomial.eval (fun i => if i = 0 then 1 else 0) P = 1 := by obtain ⟨hF, hfibre, hhom⟩ := step_3_5_fibre k hdeg obtain ⟨P, hsP, hP⟩ := step_3_5_fibre_polynomial k hfibre hhom have hpure := step_3_5_pure_coefficient k hform have hne : orderSymbol k N dt ≠ 0 := by intro h rw [h] at hpure simp at hpure have hcoeff : MvPolynomial.coeff (Finsupp.single 0 N) P = 1 := by have h := MvPolynomial.coeff_rename_mapDomain (Sum.inr : Fin (n + 1) → Fin (n + 1) ⊕ Fin (n + 1)) Sum.inr_injective P (Finsupp.single 0 N) rw [Finsupp.mapDomain_single, ← hsP] at h simpa [MvPolynomial.coeff] using h.symm.trans hpure obtain ⟨hmonic, htot, heval⟩ := step_3_5_monic k hP hcoeff exact ⟨step_3_5_order k hF hne, P, hsP, hP, htot, hmonic, heval⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
proposition_3_6: for d̃ in the form (3.1) with deg_B d̃ = N, σ_N^ord(d̃) ∈ Ann_{k[x,ξ]} gr Q, where gr Q is the graded module of Q_m = qF_m (Definition 2.4). The two sentences of the proof are step_3_6_generated (Q_• good, gr Q generated by q̄) and step_3_6_kills (q d̃ = 0 in Q_N, so q̄·P = 0); Lemma 3.5 gives the order.
Review notes
4. x₁ as a surjective map
Paragraph — Euler grading, A_{≥0} and normality of x₁
The first goal of this section will be to show that for defined as in Eq. (3.2), we have (Proposition 4.3). We will then use this result, as well as Gabber's theorem (Theorem 2.10) to establish two contradictory length estimations for the graded module that we have assumed to be nonzero, concluding that therefore, also (Corollary 4.11).
Lean: Section3.idealI
as a right ideal,
Section3.Q
We define . The commutator of defines a grading of : has Euler degree , has Euler degree , and the transverse Weyl algebra has Euler degree zero. Let denote the nonnegative part of in this grading. In PBW form (with on the left of ), is generated by the transverse algebra, , and .
By the commutation relations of the Weyl algebra, it holds that
Section4.eulerDegree,
Section4.nonnegativePart,
Section4.positivePart,
Section4.step_nonnegative_pbw,
Section4.step_coordinate_normal
/-- **Equation (nonnegative-pbw).** `E x₁ = x₁(E+1)`, `x₁E = (E−1)x₁`, `Ep = p(E−1)`, `pE = (E+1)p`, from `p x₁ = x₁ p + 1`. -/ theorem eq_nonnegative_pbw (x p : A) (hweyl : p * x = x * p + 1) : eulerElement x p * x = x * (eulerElement x p + 1) ∧ x * eulerElement x p = (eulerElement x p - 1) * x ∧ eulerElement x p * p = p * (eulerElement x p - 1) ∧ p * eulerElement x p = (eulerElement x p + 1) * p := by simp only [eulerElement] refine ⟨?_, ?_, ?_, ?_⟩ · rw [mul_assoc, hweyl] · rw [sub_mul, one_mul, mul_assoc x p x, hweyl] noncomm_ring · rw [mul_sub, mul_one, ← mul_assoc p x p, hweyl] noncomm_ring · rw [← mul_assoc, hweyl]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- **`A_{≥0}` in PBW form.** `A_{≥0}` is generated by the transverse algebra, `E` and `x₁`: these have Euler degree `0, 0, 1`; conversely an element of Euler degree `m ≥ 0` is, in PBW form, a sum of monomials `b x₁^a p^c` with `a − c = m ≥ 0`, and each is `b x₁^{a−c} (x₁^c p^c)` with `x₁^c p^c` a polynomial in `E`. -/ theorem step_nonnegative_pbw [Algebra ℚ k] : presentedNonnegativePart k n = eulerSubring (transverse k n) (presentedCoordinate k n) (presentedMomentum k n) := by let := Stafford38.WeylEulerRemainder.iteratedRatAlgebra k n apply le_antisymm · intro z hz have h := pair_nonnegativePart_le (IteratedPairStage k n) (map_mem_nonnegativePart (toIterated k n) (toIterated_coordinate k n) (toIterated_momentum k n) hz) have h' := iterated_eulerSubring_le k n _ h rwa [fromIterated_toIterated_ringHom] at h' · exact eulerSubring_le_nonnegativePart _ _ _ (Stafford38.Paper.Section2.convention_2_2_relations k n).1 (fun c hc => (Stafford38.Paper.Section2.convention_2_2_relations k n).2.2 c hc)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
More Lean proofs and definitions
/-- The Euler element `E = x₁ p`. -/ abbrev eulerElement (x p : A) : A := x * p
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The part of Euler degree `m` of the grading defined by the commutator of `E`: the elements `z` with `[E, z] = m z`. -/ def eulerDegree (x p : A) (m : ℤ) : Submodule ℤ A := Module.End.eigenspace (eulerCommutator x p) m
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The commutator `[E, ·]` with `E = x₁ p`, an additive endomorphism of `A`. -/ def eulerCommutator (x p : A) : Module.End ℤ A := LinearMap.mulLeft ℤ (eulerElement x p) - LinearMap.mulRight ℤ (eulerElement x p)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `A_{≥0}`, the nonnegative part of `A` in the Euler grading: the sum of the parts of Euler degree `m ≥ 0`. It is a subring because the grading is multiplicative. -/ def nonnegativePart (x p : A) : Subring A := { (⨆ m : ℕ, eulerDegree x p m).toAddSubgroup with one_mem' := Submodule.mem_iSup_of_mem 0 (by rw [mem_eulerDegree]; simp) mul_mem' := by intro a b ha hb change a * b ∈ ⨆ m : ℕ, eulerDegree x p m refine Submodule.iSup_induction _ (motive := fun a => a * b ∈ ⨆ m : ℕ, eulerDegree x p m) ha (fun m a ha => ?_) (by simp) (fun a a' h h' => by rw [add_mul]; exact add_mem h h') refine Submodule.iSup_induction _ (motive := fun b => a * b ∈ ⨆ m : ℕ, eulerDegree x p m) hb (fun m' b hb => ?_) (by simp) (fun b b' h h' => by rw [mul_add]; exact add_mem h h') exact Submodule.mem_iSup_of_mem (m + m') (by simpa only [Nat.cast_add] using eulerDegree_mul ha hb) }
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The elements of Euler degree `≥ 1`: the sum of the parts of degree `m ≥ 1`. -/ def positivePart (x p : A) : Submodule ℤ A := ⨆ m : ℕ, eulerDegree x p (m + 1 : ℕ)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `A_{≥0}` in `A_n`: the nonnegative part of the Euler grading. -/ abbrev presentedNonnegativePart : Subring (PresentedWeyl k (n + 1)) := nonnegativePart (presentedCoordinate k n) (presentedMomentum k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- The transverse Weyl algebra `A_{n-1}` inside `A_n`. -/ abbrev transverse : Subring (PresentedWeyl k (n + 1)) := (previousWeylEmbedding k n).range.toSubring
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- **Normality of `x₁` in `A_{≥0}`.** `x₁` commutes with the transverse algebra `C`; once `A_{≥0}` is known to be generated by `C`, `E` and `x₁` (`step_nonnegative_pbw` for `A_n`), for every `r ∈ A_{≥0}` there is `r' ∈ A_{≥0}` with `x₁ r = r' x₁`. Only this one-sided normality is used later. -/ theorem step_coordinate_normal (C : Subring A) (x p : A) (hweyl : p * x = x * p + 1) (hcomm : ∀ c : C, x * (c : A) = (c : A) * x) (hpbw : nonnegativePart x p = eulerSubring C x p) : ∀ r ∈ nonnegativePart x p, ∃ r' ∈ nonnegativePart x p, x * r = r' * x := by rw [hpbw] intro r hr obtain ⟨s, hs⟩ := coordinate_normal_eulerSubring C x p hweyl hcomm ⟨r, hr⟩ exact ⟨s, s.property, hs⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- **Normality of `x₁` in `A_{≥0}` for `A_n`.** -/ theorem presented_coordinate_normal [Algebra ℚ k] : ∀ r ∈ presentedNonnegativePart k n, ∃ r' ∈ presentedNonnegativePart k n, presentedCoordinate k n * r = r' * presentedCoordinate k n := step_coordinate_normal _ _ _ (Stafford38.Paper.Section2.convention_2_2_relations k n).1 (transverse_commutes_coordinate k n) (step_nonnegative_pbw k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- **Elements of Euler degree `≥ 1` lie in `A_{≥0} x₁`.** In PBW form such an element is a sum of monomials `b x₁^a p^c` with `a > c`, and `b x₁^a p^c = b x₁^{a−c−1} · x₁ (x₁^c p^c)`, where `x₁` crosses the polynomial `x₁^c p^c` in `E` to the right by normality. -/ theorem step_positive_mem_nonnegative_mul_coordinate [Algebra ℚ k] : ∀ z ∈ presentedPositivePart k n, ∃ u₀ ∈ presentedNonnegativePart k n, z = u₀ * presentedCoordinate k n := by let := Stafford38.WeylEulerRemainder.iteratedRatAlgebra k n intro z hz obtain ⟨U, hU, hzU⟩ := pair_positivePart_eq_mul_coordinate (IteratedPairStage k n) (map_mem_positivePart (toIterated k n) (toIterated_coordinate k n) (toIterated_momentum k n) hz) refine ⟨fromIterated k n U, ?_, ?_⟩ · rw [step_nonnegative_pbw k n] exact iterated_eulerSubring_le k n U hU · have h := congrArg (fromIterated k n) hzU rwa [map_mul, fromIterated_coordinate, fromIterated_toIterated_ringHom] at h
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- The subring generated by transverse coefficients, the coordinate, and the Euler element. In the concrete Weyl tower this is the Euler-nonnegative subring used by the quotient argument. -/ def eulerSubring (C : Subring A) (x p : A) : Subring A := Subring.closure ((C : Set A) ∪ {x, x * p})
theorem mem_eulerDegree {x p z : A} {m : ℤ} : z ∈ eulerDegree x p m ↔ eulerElement x p * z - z * eulerElement x p = m • z := by rw [eulerDegree, Module.End.mem_eigenspace_iff] rfl
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The grading is multiplicative: `[E, ·]` is a derivation. -/ theorem eulerDegree_mul {x p y z : A} {m m' : ℤ} (hy : y ∈ eulerDegree x p m) (hz : z ∈ eulerDegree x p m') : y * z ∈ eulerDegree x p (m + m') := by rw [mem_eulerDegree] at hy hz ⊢ calc eulerElement x p * (y * z) - y * z * eulerElement x p = (eulerElement x p * y - y * eulerElement x p) * z + y * (eulerElement x p * z - z * eulerElement x p) := by noncomm_ring _ = (m + m') • (y * z) := by rw [hy, hz, smul_mul_assoc, mul_smul_comm, add_smul]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem eulerDegree_pow {x p z : A} {m : ℤ} (hz : z ∈ eulerDegree x p m) : ∀ l : ℕ, z ^ l ∈ eulerDegree x p (l * m) | 0 => by rw [pow_zero, mem_eulerDegree]; simp | l + 1 => by rw [pow_succ] convert eulerDegree_mul (eulerDegree_pow hz l) hz using 2 push_cast ring
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `x₁` has Euler degree `1`. -/ theorem coordinate_mem_eulerDegree {x p : A} (hweyl : p * x = x * p + 1) : x ∈ eulerDegree x p 1 := by rw [mem_eulerDegree, eulerElement, mul_assoc, hweyl] noncomm_ring
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `E` has Euler degree `0`. -/ theorem euler_mem_eulerDegree (x p : A) : eulerElement x p ∈ eulerDegree x p 0 := by rw [mem_eulerDegree] simp
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- Elements commuting with `x₁` and `p` have Euler degree `0`. -/ theorem mem_eulerDegree_zero_of_commute {x p c : A} (hx : c * x = x * c) (hp : c * p = p * c) : c ∈ eulerDegree x p 0 := by rw [mem_eulerDegree, zero_smul, sub_eq_zero, eulerElement, mul_assoc, ← hp, ← mul_assoc, ← hx, mul_assoc]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem mem_nonnegativePart {x p z : A} : z ∈ nonnegativePart x p ↔ z ∈ ⨆ m : ℕ, eulerDegree x p m := Iff.rfl
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `x₁` and `E` lie in `A_{≥0}`. -/ theorem coordinate_mem_nonnegativePart {x p : A} (hweyl : p * x = x * p + 1) : x ∈ nonnegativePart x p := Submodule.mem_iSup_of_mem 1 (by simpa using coordinate_mem_eulerDegree hweyl)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem euler_mem_nonnegativePart (x p : A) : eulerElement x p ∈ nonnegativePart x p := Submodule.mem_iSup_of_mem 0 (by simpa using euler_mem_eulerDegree x p)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The transverse algebra `C`, `E` and `x₁` lie in `A_{≥0}`, hence so does the subring they generate. -/ theorem eulerSubring_le_nonnegativePart (C : Subring A) (x p : A) (hweyl : p * x = x * p + 1) (hcomm : ∀ c ∈ C, c * x = x * c ∧ c * p = p * c) : eulerSubring C x p ≤ nonnegativePart x p := by apply Subring.closure_le.2 rintro a (ha | ha) · exact Submodule.mem_iSup_of_mem 0 (by simpa using mem_eulerDegree_zero_of_commute (hcomm a ha).1 (hcomm a ha).2) · simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at ha rcases ha with rfl | rfl · exact coordinate_mem_nonnegativePart hweyl · exact euler_mem_nonnegativePart _ _
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem mem_positivePart_of_mem {x p z : A} {m : ℤ} (hm : 1 ≤ m) (hz : z ∈ eulerDegree x p m) : z ∈ positivePart x p := by obtain ⟨l, rfl⟩ : ∃ l : ℕ, m = (l + 1 : ℕ) := ⟨m.toNat - 1, by omega⟩ exact Submodule.mem_iSup_of_mem l hz
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- Multiplying an element of degree `≥ 1` by one of degree `0` keeps degree `≥ 1`. -/ theorem positivePart_mul_eulerDegree_zero {x p z y : A} (hz : z ∈ positivePart x p) (hy : y ∈ eulerDegree x p 0) : z * y ∈ positivePart x p := by refine Submodule.iSup_induction _ (motive := fun z => z * y ∈ positivePart x p) hz (fun m z hz => ?_) (by simp) (fun z z' h h' => by rw [add_mul]; exact add_mem h h') exact Submodule.mem_iSup_of_mem m (by simpa only [add_zero] using eulerDegree_mul hz hy)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The elements of `A_n` of Euler degree `≥ 1`. -/ abbrev presentedPositivePart : Submodule ℤ (PresentedWeyl k (n + 1)) := positivePart (presentedCoordinate k n) (presentedMomentum k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- A ring map sending `x₁, p` to `x₁', p'` preserves Euler degrees. -/ theorem map_mem_eulerDegree {A' : Type*} [Ring A'] (φ : A →+* A') {x p : A} {x' p' : A'} (hx : φ x = x') (hp : φ p = p') {m : ℤ} {z : A} (hz : z ∈ eulerDegree x p m) : φ z ∈ eulerDegree x' p' m := by subst hx hp rw [mem_eulerDegree] at hz ⊢ simpa only [eulerElement, map_sub, map_mul, map_zsmul] using congrArg φ hz
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem map_mem_nonnegativePart {A' : Type*} [Ring A'] (φ : A →+* A') {x p : A} {x' p' : A'} (hx : φ x = x') (hp : φ p = p') {z : A} (hz : z ∈ nonnegativePart x p) : φ z ∈ nonnegativePart x' p' := by refine Submodule.iSup_induction _ (motive := fun z => φ z ∈ nonnegativePart x' p') (mem_nonnegativePart.mp hz) (fun m z hz => ?_) (by simp) (fun z z' h h' => by rw [map_add]; exact add_mem h h') exact Submodule.mem_iSup_of_mem m (map_mem_eulerDegree φ hx hp hz)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
theorem map_mem_positivePart {A' : Type*} [Ring A'] (φ : A →+* A') {x p : A} {x' p' : A'} (hx : φ x = x') (hp : φ p = p') {z : A} (hz : z ∈ positivePart x p) : φ z ∈ positivePart x' p' := by refine Submodule.iSup_induction _ (motive := fun z => φ z ∈ positivePart x' p') hz (fun m z hz => ?_) (by simp) (fun z z' h h' => by rw [map_add]; exact add_mem h h') exact Submodule.mem_iSup_of_mem m (map_mem_eulerDegree φ hx hp hz)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- Induction over the PBW form `∑ b x^a p^j` of a pair-stage element whose outer exponents `j` satisfy `S`. -/ theorem pairStage_pbw_induction (S : ℕ → Prop) (P : PairStage (B := B) → Prop) (h0 : P 0) (hadd : ∀ y z, P y → P z → P (y + z)) (hmono : ∀ (b : B) (a j : ℕ), S j → P (pairCoefficient b * pairCoordinate ^ a * pairMomentum ^ j)) (z : PairStage (B := B)) (hS : ∀ j ∈ ((normalFormAddEquiv coordinateDerivation).symm z).support, S j) : P z := by have hcoeff : ∀ (c : CoordinateStage (B := B)) (j : ℕ), S j → P (normalCoefficient coordinateDerivation c * pairMomentum ^ j) := by intro c j hj let r := (normalFormAddEquiv (zeroDerivation (B := B))).symm c have hc : c = normalForm (zeroDerivation (B := B)) r := ((normalFormAddEquiv (zeroDerivation (B := B))).apply_symm_apply c).symm rw [hc] induction r using Polynomial.induction_on' with | add r s hr hs => rw [normalForm_add, (normalCoefficient (coordinateDerivation (B := B))).map_add, add_mul] exact hadd _ _ hr hs | monomial a b => rw [normalForm_monomial, (normalCoefficient (coordinateDerivation (B := B))).map_mul, (normalCoefficient (coordinateDerivation (B := B))).map_pow] exact hmono b a j hj let q := (normalFormAddEquiv coordinateDerivation).symm z have hz : z = normalForm coordinateDerivation q := ((normalFormAddEquiv coordinateDerivation).apply_symm_apply z).symm rw [hz, ← Polynomial.sum_monomial_eq q] change P (normalFormAddHom (coordinateDerivation (B := B)) (∑ j ∈ q.support, Polynomial.monomial j (q.coeff j))) rw [map_sum] refine Finset.sum_induction _ P hadd h0 fun j hj => ?_ change P (normalForm (coordinateDerivation (B := B)) (Polynomial.monomial j (q.coeff j))) rw [normalForm_monomial] exact hcoeff (q.coeff j) j (hS j hj)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- The PBW monomial `b x^a p^c` has Euler degree `a − c`. -/ theorem pairMonomial_mem_eulerDegree (b : B) (a c : ℕ) : pairCoefficient b * pairCoordinate ^ a * pairMomentum ^ c ∈ eulerDegree (pairCoordinate (B := B)) pairMomentum ((a : ℤ) - c) := mem_eulerDegree.mpr (Stafford38.PaperEulerGrading.orderedPBWMonomial_eulerWeight B b a c)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- The span of the PBW monomials of Euler degree `m`. -/ def pairWeightSpan (m : ℤ) : Submodule ℤ (PairStage (B := B)) := Submodule.span ℤ {z | ∃ (b : B) (a c : ℕ), (a : ℤ) - c = m ∧ z = pairCoefficient b * pairCoordinate ^ a * pairMomentum ^ c}
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
theorem pairWeightSpan_le (m : ℤ) : pairWeightSpan B m ≤ eulerDegree (pairCoordinate (B := B)) pairMomentum m := by rw [pairWeightSpan, Submodule.span_le] rintro _ ⟨b, a, c, rfl, rfl⟩ exact pairMonomial_mem_eulerDegree B b a c
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- PBW: every element is a sum of homogeneous PBW monomials. -/ theorem mem_iSup_pairWeightSpan (z : PairStage (B := B)) : z ∈ ⨆ m, pairWeightSpan B m := by refine pairStage_pbw_induction B (fun _ => True) (fun z => z ∈ ⨆ m, pairWeightSpan B m) (zero_mem _) (fun _ _ hy hz => add_mem hy hz) (fun b a c _ => ?_) z fun _ _ => trivial exact Submodule.mem_iSup_of_mem ((a : ℤ) - c) (Submodule.subset_span ⟨b, a, c, rfl, rfl⟩)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- An element of Euler degree `m` is a sum of PBW monomials of degree `m`: the parts of different degree are independent (eigenspaces of `[E, ·]`, the pair stage being torsion-free as a `ℚ`-algebra). -/ theorem mem_pairWeightSpan_of_mem_eulerDegree [Algebra ℚ B] {m : ℤ} {z : PairStage (B := B)} (hz : z ∈ eulerDegree pairCoordinate pairMomentum m) : z ∈ pairWeightSpan B m := by let : Algebra ℚ (CoordinateStage (B := B)) := coordinateStageAlgebra let : Algebra ℚ (PairStage (B := B)) := pairStageAlgebra have : IsAddTorsionFree (PairStage (B := B)) := .of_module_rat _ have hsplit := mem_iSup_pairWeightSpan B z rw [iSup_split_single _ m, Submodule.mem_sup] at hsplit obtain ⟨y, hy, r, hr, hyr⟩ := hsplit have hr' : r ∈ ⨆ (ν : ℤ) (_ : ν ≠ m), Module.End.eigenspace (eulerCommutator (pairCoordinate (B := B)) pairMomentum) ν := iSup₂_mono (fun j _ => pairWeightSpan_le B j) hr have hrm : r ∈ Module.End.eigenspace (eulerCommutator (pairCoordinate (B := B)) pairMomentum) m := by have hr_eq : r = z - y := by rw [← hyr]; abel rw [hr_eq] exact sub_mem hz (pairWeightSpan_le B m hy) have h0 := Submodule.disjoint_def.mp (Module.End.eigenspaces_iSupIndep (eulerCommutator (pairCoordinate (B := B)) pairMomentum) m) r hrm hr' rw [← hyr, h0, add_zero] exact hy
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- Monomials of Euler degree `≥ 0` lie in the subring generated by the coefficients, `x` and `E`. -/ theorem pairWeightSpan_le_eulerSubring {m : ℤ} (hm : 0 ≤ m) {z : PairStage (B := B)} (hz : z ∈ pairWeightSpan B m) : z ∈ pairEulerSubring B := by induction hz using Submodule.span_induction with | mem z hz => obtain ⟨b, a, c, hac, rfl⟩ := hz exact pairOrderedMonomial_mem_eulerSubring B b a c (by omega) | zero => exact zero_mem _ | add y z _ _ hy hz => exact add_mem hy hz | smul l z _ hz => exact zsmul_mem hz l
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]Source ↗
variable [Algebra ℚ B]
/-- Monomials of Euler degree `≥ 1` have a right factor `x` with cofactor in that subring, by normality of `x` (`pairOrderedMonomial_eq_eulerSubring_mul_coordinate`). -/ theorem pairWeightSpan_eq_mul_coordinate {m : ℤ} (hm : 1 ≤ m) {z : PairStage (B := B)} (hz : z ∈ pairWeightSpan B m) : ∃ U ∈ pairEulerSubring B, z = U * pairCoordinate := by induction hz using Submodule.span_induction with | mem z hz => obtain ⟨b, a, c, hac, rfl⟩ := hz obtain ⟨U, hU⟩ := pairOrderedMonomial_eq_eulerSubring_mul_coordinate B b a c (by omega) exact ⟨U, U.2, hU⟩ | zero => exact ⟨0, zero_mem _, by simp⟩ | add y z _ _ hy hz => obtain ⟨U, hU, rfl⟩ := hy obtain ⟨V, hV, rfl⟩ := hz exact ⟨U + V, add_mem hU hV, by rw [add_mul]⟩ | smul l z _ hz => obtain ⟨U, hU, rfl⟩ := hz exact ⟨l • U, zsmul_mem hU l, by rw [smul_mul_assoc]⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]Source ↗
variable [Algebra ℚ B]
/-- At the pair stage, `A_{≥0}` lies in the subring generated by the coefficients, `x` and `E`. -/ theorem pair_nonnegativePart_le : nonnegativePart (pairCoordinate (B := B)) pairMomentum ≤ pairEulerSubring B := by intro z hz exact Submodule.iSup_induction _ (motive := fun z => z ∈ pairEulerSubring B) (mem_nonnegativePart.mp hz) (fun m z hz => pairWeightSpan_le_eulerSubring B (Int.natCast_nonneg m) (mem_pairWeightSpan_of_mem_eulerDegree B hz)) (zero_mem _) (fun _ _ => add_mem)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]Source ↗
variable [Algebra ℚ B]
/-- At the pair stage, elements of Euler degree `≥ 1` lie in `R x` for the subring `R` generated by the coefficients, `x` and `E`. -/ theorem pair_positivePart_eq_mul_coordinate {z : PairStage (B := B)} (hz : z ∈ positivePart (pairCoordinate (B := B)) pairMomentum) : ∃ U ∈ pairEulerSubring B, z = U * pairCoordinate := by refine Submodule.iSup_induction _ (motive := fun z => ∃ U ∈ pairEulerSubring B, z = U * pairCoordinate) hz (fun m z hz => pairWeightSpan_eq_mul_coordinate B (by omega) (mem_pairWeightSpan_of_mem_eulerDegree B hz)) ⟨0, zero_mem _, by simp⟩ ?_ rintro _ _ ⟨U, hU, rfl⟩ ⟨V, hV, rfl⟩ exact ⟨U + V, add_mem hU hV, by rw [add_mul]⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]Source ↗
variable [Algebra ℚ B]
/-- The equivalence `A_n ≃ PairStage(A_{n-1})`, as a ring map. -/ abbrev toIterated : PresentedWeyl k (n + 1) →+* PairStage (B := IteratedPairStage k n) := (presentedIteratedEquiv k (n + 1)).toRingEquiv.toRingHom
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
theorem toIterated_coordinate : toIterated k n (presentedCoordinate k n) = pairCoordinate := by rw [← fromIterated_coordinate k n] exact (presentedIteratedEquiv k (n + 1)).apply_symm_apply _
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
theorem toIterated_momentum : toIterated k n (presentedMomentum k n) = pairMomentum := by rw [← fromIterated_momentum k n] exact (presentedIteratedEquiv k (n + 1)).apply_symm_apply _
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
theorem fromIterated_toIterated_ringHom (d : PresentedWeyl k (n + 1)) : fromIterated k n (toIterated k n d) = d := (presentedIteratedEquiv k (n + 1)).symm_apply_apply d
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- The equivalence `A_n ≃ PairStage(A_{n-1})` sends the Ore-tower Euler subring into the subring of `A_n` generated by `A_{n-1}`, `x₁` and `E`. -/ theorem iterated_eulerSubring_le (z : PairStage (B := IteratedPairStage k n)) (hz : z ∈ pairEulerSubring (IteratedPairStage k n)) : fromIterated k n z ∈ eulerSubring (transverse k n) (presentedCoordinate k n) (presentedMomentum k n) := by have hle : pairEulerSubring (IteratedPairStage k n) ≤ (eulerSubring (transverse k n) (presentedCoordinate k n) (presentedMomentum k n)).comap (fromIterated k n) := by apply Subring.closure_le.2 intro a ha rcases ha with ⟨b, rfl⟩ | ha · exact Subring.subset_closure (Or.inl (fromIterated_coefficient_mem_transverse k n b)) · simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at ha rcases ha with rfl | rfl · rw [SetLike.mem_coe, Subring.mem_comap, fromIterated_coordinate] exact coordinate_mem_eulerSubring _ _ _ · rw [SetLike.mem_coe, Subring.mem_comap, map_mul, fromIterated_coordinate, fromIterated_momentum] exact euler_mem_eulerSubring _ _ _ exact hle hz
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
Comparison notes
Lean defines the grading as printed: eulerDegree is the part of Euler degree m, the eigenspace [E, z] = m z of the commutator [E, ·] (eulerCommutator) with E = x₁p (eulerElement), and the grading is multiplicative (eulerDegree_mul). x₁ has Euler degree 1 (coordinate_mem_eulerDegree), elements commuting with x₁ and p, such as the transverse Weyl algebra, have degree 0 (mem_eulerDegree_zero_of_commute), and a PBW monomial b x₁^a p^c has degree a − c (pairMonomial_mem_eulerDegree). Lean does not state separately for A_n that p has degree −1 or that A_n is the sum of its degree parts; both are used at the pair stage, as the degree a − c of b x₁^a p^c and as the decomposition into homogeneous PBW monomials (mem_iSup_pairWeightSpan). A_{≥0} is the nonnegative part, the sum of the parts of degree m ≥ 0 (nonnegativePart; presentedNonnegativePart in A_n). step_nonnegative_pbw proves that A_{≥0} is generated by the transverse algebra, E and x₁ (WeylEulerSubring.eulerSubring): these lie in A_{≥0} (eulerSubring_le_nonnegativePart), and in PBW form an element of degree m ≥ 0 is a sum of monomials b x₁^a p^c with a − c = m (mem_pairWeightSpan_of_mem_eulerDegree), each in the pair-stage subring generated by the coefficients, x₁ and E (pairWeightSpan_le_eulerSubring), which A_n ≃ PairStage(A_{n−1}) maps into the subring of A_n generated by the transverse algebra, x₁ and E (iterated_eulerSubring_le). eq_nonnegative_pbw gives the four relations (eq:nonnegative-pbw) from p x₁ = x₁ p + 1. step_coordinate_normal derives the normality x₁r = r′x₁ (r, r′ ∈ A_{≥0}) from x₁ commuting with the transverse algebra and the PBW description, and presented_coordinate_normal is the case of A_n. step_positive_mem_nonnegative_mul_coordinate proves that every element of Euler degree ≥ 1 (presentedPositivePart) lies in A_{≥0}x₁: in PBW form b x₁^a p^c = b x₁^{a−c−1} · x₁(x₁^c p^c) with a > c, where x₁^c p^c is a polynomial in E and x₁ crosses it by normality (pairWeightSpan_eq_mul_coordinate).
Review notes
Lemma 4.1 — Euler products
∎
Section4.lemma_4_1
/-- **Lemma 4.1.** For `x₁, p = ∂₁ ∈ A_n(k)` and `E = x₁p`, for every `m ≥ 1`, `x₁^m p^m = ∏_{i=0}^{m-1}(E−i)` and `p^m x₁^m = ∏_{i=1}^{m}(E+i)`, the factors in each product commuting with one another. The products are taken in the listed order of `i` (`A_n(k)` is not commutative); by the last two clauses the order does not matter. Paper `n` is Lean `n + 1`. -/ theorem lemma_4_1 (k : Type*) [Field k] (n m : ℕ) (hm : 1 ≤ m) : x₁ k n ^ m * p k n ^ m = ((List.range m).map fun i : ℕ => eulerElement (x₁ k n) (p k n) - i).prod ∧ p k n ^ m * x₁ k n ^ m = ((List.range' 1 m).map fun i : ℕ => eulerElement (x₁ k n) (p k n) + i).prod ∧ (∀ i j : ℕ, Commute (eulerElement (x₁ k n) (p k n) - i) (eulerElement (x₁ k n) (p k n) - j)) ∧ ∀ i j : ℕ, Commute (eulerElement (x₁ k n) (p k n) + i) (eulerElement (x₁ k n) (p k n) + j) := lemma_4_1_of_weylPair _ _ (convention_2_2_relations k n).1 m hm /-! ## The PBW form at one Ore pair stage Every element of `PairStage B` is a sum of PBW monomials `b x^a p^c` (coefficient on the left; `b x^a p^c` and `p^c x^a b` have the same Euler degree `a − c`). This section proves that the part of Euler degree `m` is spanned by the monomials of degree `m`. -/
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageMore Lean proofs and definitions
/-- Lemma 4.1 for any ring with elements `x, p` satisfying `p x = x p + 1`: for every `m ≥ 1`, `x^m p^m = ∏_{i=0}^{m-1}(E−i)` and `p^m x^m = ∏_{i=1}^{m}(E+i)` with `E = x p`, the factors in each product commuting with one another. The products are taken in the listed order of `i`; by the last two clauses the order does not matter. -/ theorem lemma_4_1_of_weylPair (x p : A) (hweyl : p * x = x * p + 1) (m : ℕ) (hm : 1 ≤ m) : x ^ m * p ^ m = ((List.range m).map fun i : ℕ => eulerElement x p - i).prod ∧ p ^ m * x ^ m = ((List.range' 1 m).map fun i : ℕ => eulerElement x p + i).prod ∧ (∀ i j : ℕ, Commute (eulerElement x p - i) (eulerElement x p - j)) ∧ ∀ i j : ℕ, Commute (eulerElement x p + i) (eulerElement x p + j) := by refine ⟨?_, ?_, step_4_1_factors_commute x p⟩ · -- induction from the trivial case `m = 1` induction m, hm using Nat.le_induction with | base => simp [eulerElement] | succ m _ ih => exact step_4_1_falling_succ x p hweyl m ih · induction m, hm using Nat.le_induction with | base => simp [eulerElement, hweyl] | succ m _ ih => exact step_4_1_rising_succ x p hweyl m ih
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- The factors `E − i` and `E + i` are polynomials in the single element `E`, hence commute with one another. -/ theorem step_4_1_factors_commute (x p : A) : (∀ i j : ℕ, Commute (eulerElement x p - i) (eulerElement x p - j)) ∧ ∀ i j : ℕ, Commute (eulerElement x p + i) (eulerElement x p + j) := by refine ⟨fun i j => ?_, fun i j => ?_⟩ · exact ((Commute.refl _).sub_right (Nat.cast_commute j _).symm).sub_left ((Nat.cast_commute i _).sub_right (Nat.cast_commute i (j : A))) · exact ((Commute.refl _).add_right (Nat.cast_commute j _).symm).add_left ((Nat.cast_commute i _).add_right (Nat.cast_commute i (j : A)))
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `x₁ ∏(E − c) = ∏(E − 1 − c) x₁` for any list of shifts, from `x₁E = (E−1)x₁`. -/ theorem step_4_1_coordinate_shift (x p : A) (hweyl : p * x = x * p + 1) (l : List ℕ) : x * (l.map fun i : ℕ => eulerElement x p - i).prod = (l.map fun i : ℕ => eulerElement x p - 1 - i).prod * x := by have hxE := (eq_nonnegative_pbw x p hweyl).2.1 induction l with | nil => simp | cons a l ih => simp only [List.map_cons, List.prod_cons] rw [← mul_assoc, mul_sub, hxE, ← (Nat.cast_commute a x).eq, ← sub_mul, mul_assoc, ih, mul_assoc]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- Induction step of the first identity: `x₁^{m+1}p^{m+1} = x₁(∏_{i<m}(E−i))p = (∏_{i<m}(E−1−i))x₁p = ∏_{i≤m}(E−i)`. -/ theorem step_4_1_falling_succ (x p : A) (hweyl : p * x = x * p + 1) (m : ℕ) (ih : x ^ m * p ^ m = ((List.range m).map fun i : ℕ => eulerElement x p - i).prod) : x ^ (m + 1) * p ^ (m + 1) = ((List.range (m + 1)).map fun i : ℕ => eulerElement x p - i).prod := by calc x ^ (m + 1) * p ^ (m + 1) = x * (x ^ m * p ^ m) * p := by rw [pow_succ', pow_succ]; simp only [mul_assoc] _ = ((List.range m).map fun i : ℕ => eulerElement x p - 1 - i).prod * eulerElement x p := by rw [ih, step_4_1_coordinate_shift x p hweyl, mul_assoc] _ = eulerElement x p * ((List.range m).map fun i : ℕ => eulerElement x p - 1 - i).prod := by refine (Commute.list_prod_left _ _ fun y hy => ?_).eq obtain ⟨i, -, rfl⟩ := List.mem_map.mp hy exact ((Commute.refl _).sub_left (Commute.one_left _)).sub_left (Nat.cast_commute i _) _ = ((List.range (m + 1)).map fun i : ℕ => eulerElement x p - i).prod := by rw [List.range_succ_eq_map, List.map_cons, List.prod_cons, List.map_map] congr 1 · simp · congr 1 refine List.map_congr_left fun i _ => ?_ simp only [Function.comp_apply, Nat.succ_eq_add_one] push_cast abel
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- `p ∏(E + c) = ∏(E + 1 + c) p` for any list of shifts, from `pE = (E+1)p`. -/ theorem step_4_1_momentum_shift (x p : A) (hweyl : p * x = x * p + 1) (l : List ℕ) : p * (l.map fun i : ℕ => eulerElement x p + i).prod = (l.map fun i : ℕ => eulerElement x p + 1 + i).prod * p := by have hpE := (eq_nonnegative_pbw x p hweyl).2.2.2 induction l with | nil => simp | cons a l ih => simp only [List.map_cons, List.prod_cons] rw [← mul_assoc, mul_add, hpE, ← (Nat.cast_commute a p).eq, ← add_mul, mul_assoc, ih, mul_assoc]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- Induction step of the second identity, from `Ep = p(E−1)`, i.e. `pE = (E+1)p`: `p^{m+1}x₁^{m+1} = p(∏_{i=1}^m(E+i))x₁ = (∏_{i=1}^m(E+1+i))px₁ = ∏_{i=1}^{m+1}(E+i)`. -/ theorem step_4_1_rising_succ (x p : A) (hweyl : p * x = x * p + 1) (m : ℕ) (ih : p ^ m * x ^ m = ((List.range' 1 m).map fun i : ℕ => eulerElement x p + i).prod) : p ^ (m + 1) * x ^ (m + 1) = ((List.range' 1 (m + 1)).map fun i : ℕ => eulerElement x p + i).prod := by have hpx : p * x = eulerElement x p + 1 := hweyl calc p ^ (m + 1) * x ^ (m + 1) = p * (p ^ m * x ^ m) * x := by rw [pow_succ', pow_succ]; simp only [mul_assoc] _ = ((List.range' 1 m).map fun i : ℕ => eulerElement x p + 1 + i).prod * (eulerElement x p + 1) := by rw [ih, step_4_1_momentum_shift x p hweyl, mul_assoc, hpx] _ = (eulerElement x p + 1) * ((List.range' 1 m).map fun i : ℕ => eulerElement x p + 1 + i).prod := by refine (Commute.list_prod_left _ _ fun y hy => ?_).eq obtain ⟨i, -, rfl⟩ := List.mem_map.mp hy exact (Commute.refl (eulerElement x p + 1)).add_left (Nat.cast_commute i _) _ = ((List.range' 1 (m + 1)).map fun i : ℕ => eulerElement x p + i).prod := by rw [List.range'_succ, List.map_cons, List.prod_cons] congr 1 · simp · rw [List.range'_eq_map_range, List.range'_eq_map_range, List.map_map, List.map_map] congr 1 refine List.map_congr_left fun i _ => ?_ simp only [Function.comp_apply] push_cast rw [add_assoc _ 1, ← add_assoc 1 1, one_add_one_eq_two]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- **Equation (nonnegative-pbw).** `E x₁ = x₁(E+1)`, `x₁E = (E−1)x₁`, `Ep = p(E−1)`, `pE = (E+1)p`, from `p x₁ = x₁ p + 1`. -/ theorem eq_nonnegative_pbw (x p : A) (hweyl : p * x = x * p + 1) : eulerElement x p * x = x * (eulerElement x p + 1) ∧ x * eulerElement x p = (eulerElement x p - 1) * x ∧ eulerElement x p * p = p * (eulerElement x p - 1) ∧ p * eulerElement x p = (eulerElement x p + 1) * p := by simp only [eulerElement] refine ⟨?_, ?_, ?_, ?_⟩ · rw [mul_assoc, hweyl] · rw [sub_mul, one_mul, mul_assoc x p x, hweyl] noncomm_ring · rw [mul_sub, mul_one, ← mul_assoc p x p, hweyl] noncomm_ring · rw [← mul_assoc, hweyl]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
/-- **Convention 2.2, relations.** `x₁` is central in `C`, `[p, x₁] = 1`, and the transverse algebra commutes with `x₁` and `p`. -/ theorem convention_2_2_relations (n : ℕ) : p k n * x₁ k n = x₁ k n * p k n + 1 ∧ (∀ c ∈ C k n, x₁ k n * c = c * x₁ k n) ∧ ∀ t ∈ transverse k n, t * x₁ k n = x₁ k n * t ∧ t * p k n = p k n * t := by refine ⟨presentedMomentum_mul_coordinate k n, ?_, ?_⟩ · intro c hc induction hc using Algebra.adjoin_induction with | mem x hx => rcases hx with ⟨w, rfl⟩ | hx · exact presentedCoordinate_commutes_previous k n w · rw [Set.mem_singleton_iff.mp hx] | algebraMap r => exact (Algebra.commutes r _).symm | add x y _ _ hx hy => rw [mul_add, add_mul, hx, hy] | mul x y _ _ hx hy => rw [← mul_assoc, hx, mul_assoc, hy, mul_assoc] · rintro _ ⟨w, rfl⟩ exact ⟨(presentedCoordinate_commutes_previous k n w).symm, (presentedMomentum_commutes_previous k n w).symm⟩
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (k : Type u) [Field k]Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra k (PairStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Nontrivial (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Nontrivial (CoordinateStage (B := IteratedPairStage k n)) :=
Comparison notes
lemma_4_1 states the lemma for x₁ and p = ∂₁ in A_n(k), with E = x₁p: for m ≥ 1, x₁^m p^m = Π_{i=0}^{m−1}(E − i) and p^m x₁^m = Π_{i=1}^{m}(E + i), the products taken in the listed order, and the factors in each product commute. the paper's n is Lean's n + 1. It is lemma_4_1_of_weylPair, the same identities in any ring with elements x, p such that px = xp + 1, applied to p x₁ = x₁ p + 1 (convention_2_2_relations). The proof is the printed induction: step_4_1_factors_commute, then the induction steps step_4_1_falling_succ (x₁E = (E − 1)x₁, step_4_1_coordinate_shift) and step_4_1_rising_succ (pE = (E + 1)p, step_4_1_momentum_shift).
Review notes
Lemma 4.2 — Euler residue 1 + U x₁ ∈ I
Section4.lemma_4_2/-- **Lemma 4.2.** For `d̃` in the form (3.1) and `𝓘 = d̃A_n + x₁^N d̃A_n` (Equation (3.2)), there is `u₀ ∈ A_{≥0}` with `1 + u₀ x₁ ∈ 𝓘`. The proof uses (3.1) only through its consequence `IsPBWMonicAt` (`Section3.isPBWMonicAt_of_isMonicNormalForm`), and cites Lemma 4.1 for `m = N`. -/ theorem lemma_4_2 {N : ℕ} {d : PresentedWeyl k (n + 1)} (hform : Stafford38.Paper.Section3.IsMonicNormalForm k N d) : ∃ u₀ ∈ presentedNonnegativePart k n, 1 + u₀ * presentedCoordinate k n ∈ Stafford38.Paper.Section3.idealI k N d := by let := presentedRatAlgebra k n have hd := Stafford38.Paper.Section3.isPBWMonicAt_of_isMonicNormalForm k hform obtain ⟨hright, hleft⟩ := step_4_2_presented_remainder hd exact lemma_4_2_of_positive (presentedCoordinate k n) (presentedMomentum k n) d (d - presentedMomentum k n ^ N) N (fun hN => ⟨(lemma_4_1 k n N hN).1, (lemma_4_1 k n N hN).2.1⟩) (by abel) hright hleft (step_positive_mem_nonnegative_mul_coordinate k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] [Algebra ℚ k] (n : ℕ)Source ↗
variable {k n}More Lean proofs and definitions
/-- `R_+(E) = ∏_{r=1}^N (E+r)` and `R_-(E) = ∏_{r=0}^{N-1}(E−r)` have the disjoint root sets `{−1,…,−N}` and `{0,…,N−1}`, so `gcd(R_+, R_-) = 1` in `ℚ[E]`. -/ theorem step_4_2_coprime (N : ℕ) : IsCoprime ((List.range' 1 N).map fun r : ℕ => (Polynomial.X + Polynomial.C (r : ℚ))).prod ((List.range N).map fun r : ℕ => (Polynomial.X - Polynomial.C (r : ℚ))).prod := by apply isCoprime_list_prod intro a ha b hb obtain ⟨r, hr, rfl⟩ := List.mem_map.mp ha obtain ⟨s, -, rfl⟩ := List.mem_map.mp hb have hr1 : 1 ≤ r := (List.mem_range'_1.mp hr).1 have hunit : IsUnit (-(r : ℚ) - s) := by apply isUnit_iff_ne_zero.mpr have : (1 : ℚ) ≤ r := by exact_mod_cast hr1 have : (0 : ℚ) ≤ s := by positivity linarith have := Polynomial.isCoprime_X_sub_C_of_isUnit_sub hunit simpa [sub_eq_add_neg] using this
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStage/-- Evaluating `R_±` at `E` gives the products of Lemma 4.1. -/ theorem step_4_2_eval_products (x p : A) (N : ℕ) : eulerPolynomialEval x p ((List.range' 1 N).map fun r : ℕ => (Polynomial.X + Polynomial.C (r : ℚ))).prod = ((List.range' 1 N).map fun i : ℕ => eulerElement x p + i).prod ∧ eulerPolynomialEval x p ((List.range N).map fun r : ℕ => (Polynomial.X - Polynomial.C (r : ℚ))).prod = ((List.range N).map fun i : ℕ => eulerElement x p - i).prod := by constructor <;> · rw [map_list_prod, List.map_map] congr 1 refine List.map_congr_left fun r _ => ?_ simp [eulerElement]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- **First paragraph of the proof of Lemma 4.2.** By Lemma 4.1 (the hypothesis `h41`, the product identities for `m = N`) and the extended Euclidean algorithm in `ℚ[E]` there are `u, v ∈ ℚ[E]` with `p^N x₁^N u(E) + x₁^N p^N v(E) = 1`. -/ theorem step_4_2_bezout (x p : A) (N : ℕ) (h41 : 1 ≤ N → x ^ N * p ^ N = ((List.range N).map fun i : ℕ => eulerElement x p - i).prod ∧ p ^ N * x ^ N = ((List.range' 1 N).map fun i : ℕ => eulerElement x p + i).prod) : ∃ u v : Polynomial ℚ, (p ^ N * x ^ N) * eulerPolynomialEval x p u + (x ^ N * p ^ N) * eulerPolynomialEval x p v = 1 := by obtain ⟨u, v, huv⟩ := step_4_2_coprime N refine ⟨u, v, ?_⟩ rw [mul_comm u, mul_comm v] at huv have h := congrArg (eulerPolynomialEval x p) huv rw [map_add, map_mul, map_mul, map_one] at h rcases Nat.eq_zero_or_pos N with rfl | hN · simpa [add_comm] using h obtain ⟨hminus, hplus⟩ := h41 hN rw [(step_4_2_eval_products x p N).1, (step_4_2_eval_products x p N).2, ← hplus, ← hminus] at h exact h
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- **Second paragraph.** With `d̃ = p^N + Σ`, `(d̃ x₁^N) u(E) + (x₁^N d̃) v(E) = 1 + w ∈ 𝓘`, where `w = Σ x₁^N u(E) + x₁^N Σ v(E)` and `𝓘 = d̃A + x₁^N d̃A`. -/ theorem step_4_2_residue (x p d sigma : A) (N : ℕ) (u v : Polynomial ℚ) (hd : d = p ^ N + sigma) (hbezout : (p ^ N * x ^ N) * eulerPolynomialEval x p u + (x ^ N * p ^ N) * eulerPolynomialEval x p v = 1) : (d * x ^ N) * eulerPolynomialEval x p u + (x ^ N * d) * eulerPolynomialEval x p v = 1 + ((sigma * x ^ N) * eulerPolynomialEval x p u + (x ^ N * sigma) * eulerPolynomialEval x p v) ∧ 1 + ((sigma * x ^ N) * eulerPolynomialEval x p u + (x ^ N * sigma) * eulerPolynomialEval x p v) ∈ canonicalRightIdeal x d N := by have hid := positiveEulerResidue_identity x p d sigma (eulerPolynomialEval x p u) (eulerPolynomialEval x p v) N hd hbezout rw [← mul_assoc] at hid refine ⟨hid, ?_⟩ rw [← hid] refine (canonicalRightIdeal x d N).add_mem ?_ ?_ · rw [mul_assoc] change MulOpposite.op (x ^ N * eulerPolynomialEval x p u) • d ∈ canonicalRightIdeal x d N exact (canonicalRightIdeal x d N).smul_mem _ (firstGenerator_mem x d N) · change MulOpposite.op (eulerPolynomialEval x p v) • (x ^ N * d) ∈ canonicalRightIdeal x d N exact (canonicalRightIdeal x d N).smul_mem _ (secondGenerator_mem x d N)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- **Positivity of the remainder for `d̃`.** Write `Σ = d̃ − p^N` in PBW form: a sum of terms `b x₁^a p^j` with `j < N` (`presentedRemainder_outer_support_lt`). Such a term has Euler degree `a − j`, so multiplied by `x₁^N` on either side it has Euler degree `a + N − j ≥ N − j > 0`. Hence `Σ x₁^N` and `x₁^N Σ` have positive Euler degree. -/ theorem step_4_2_presented_remainder {N : ℕ} {d : PresentedWeyl k (n + 1)} (hd : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N d) : (d - presentedMomentum k n ^ N) * presentedCoordinate k n ^ N ∈ presentedPositivePart k n ∧ presentedCoordinate k n ^ N * (d - presentedMomentum k n ^ N) ∈ presentedPositivePart k n := by have hxN := eulerDegree_pow (coordinate_mem_eulerDegree (pairMomentum_mul_coordinate (B := IteratedPairStage k n))) N have hpair := pairStage_pbw_induction (IteratedPairStage k n) (· < N) (fun z => z * pairCoordinate ^ N ∈ positivePart pairCoordinate pairMomentum ∧ pairCoordinate ^ N * z ∈ positivePart pairCoordinate pairMomentum) ⟨by simp, by simp⟩ (fun y z hy hz => ⟨by rw [add_mul]; exact add_mem hy.1 hz.1, by rw [mul_add]; exact add_mem hy.2 hz.2⟩) (fun b a j hj => by have hmono := pairMonomial_mem_eulerDegree (IteratedPairStage k n) b a j exact ⟨mem_positivePart_of_mem (by omega) (eulerDegree_mul hmono hxN), mem_positivePart_of_mem (by omega) (eulerDegree_mul hxN hmono)⟩) (presentedToIterated k (n + 1) d - stageMomentum k n ^ N) (presentedRemainder_outer_support_lt k n N hd) have hσ : fromIterated k n (presentedToIterated k (n + 1) d - stageMomentum k n ^ N) = d - presentedMomentum k n ^ N := by change (presentedIteratedEquiv k (n + 1)).symm (presentedIteratedEquiv k (n + 1) d - stageMomentum k n ^ N) = _ rw [map_sub, AlgEquiv.symm_apply_apply, map_pow] congr 2 exact fromIterated_momentum k n have hright := map_mem_positivePart (fromIterated k n) (fromIterated_coordinate k n) (fromIterated_momentum k n) hpair.1 have hleft := map_mem_positivePart (fromIterated k n) (fromIterated_coordinate k n) (fromIterated_momentum k n) hpair.2 rw [RingHom.map_mul (fromIterated k n) (presentedToIterated k (n + 1) d - stageMomentum k n ^ N), RingHom.map_pow (fromIterated k n) pairCoordinate N, hσ, fromIterated_coordinate] at hright rw [RingHom.map_mul (fromIterated k n) (pairCoordinate ^ N) (presentedToIterated k (n + 1) d - stageMomentum k n ^ N), RingHom.map_pow (fromIterated k n) pairCoordinate N, hσ, fromIterated_coordinate] at hleft exact ⟨hright, hleft⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] [Algebra ℚ k] (n : ℕ)Source ↗
variable {k n}/-- `u(E)` and `v(E)` depend only on `E`, so they have Euler degree `0`. -/ theorem step_4_2_eulerPolynomial_degree_zero (x p : A) (f : Polynomial ℚ) : eulerPolynomialEval x p f ∈ eulerDegree x p 0 := by induction f using Polynomial.induction_on' with | add f g hf hg => rw [map_add] exact add_mem hf hg | monomial l q => rw [← Polynomial.C_mul_X_pow_eq_monomial, map_mul, map_pow, eulerPolynomialEval_C, eulerPolynomialEval_X] have hq : algebraMap ℚ A q ∈ eulerDegree x p 0 := mem_eulerDegree_zero_of_commute (Algebra.commutes q x) (Algebra.commutes q p) simpa using eulerDegree_mul hq (eulerDegree_pow (euler_mem_eulerDegree x p) l)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- **Third paragraph.** `Σ x₁^N` and `x₁^N Σ` have positive Euler degree (`step_4_2_presented_remainder` for the presented `d̃`) and `u(E), v(E)` have degree `0`, so every term of `w = Σ x₁^N u(E) + x₁^N Σ v(E)` has positive Euler degree; hence `w ∈ A_{≥0} x₁` (the hypothesis `hpos`, which is `step_positive_mem_nonnegative_mul_coordinate` for `A_n`). -/ theorem step_4_2_positive (x p sigma : A) (N : ℕ) (hright : sigma * x ^ N ∈ positivePart x p) (hleft : x ^ N * sigma ∈ positivePart x p) (hpos : ∀ z ∈ positivePart x p, ∃ u₀ ∈ nonnegativePart x p, z = u₀ * x) (u v : Polynomial ℚ) : ∃ u₀ ∈ nonnegativePart x p, (sigma * x ^ N) * eulerPolynomialEval x p u + (x ^ N * sigma) * eulerPolynomialEval x p v = u₀ * x := hpos _ (add_mem (positivePart_mul_eulerDegree_zero hright (step_4_2_eulerPolynomial_degree_zero x p u)) (positivePart_mul_eulerDegree_zero hleft (step_4_2_eulerPolynomial_degree_zero x p v)))
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- Lemma 4.2 for a Weyl pair, assembled from the three paragraphs, given the identities of Lemma 4.1 for `m = N`. -/ theorem lemma_4_2_of_positive (x p d sigma : A) (N : ℕ) (h41 : 1 ≤ N → x ^ N * p ^ N = ((List.range N).map fun i : ℕ => eulerElement x p - i).prod ∧ p ^ N * x ^ N = ((List.range' 1 N).map fun i : ℕ => eulerElement x p + i).prod) (hd : d = p ^ N + sigma) (hright : sigma * x ^ N ∈ positivePart x p) (hleft : x ^ N * sigma ∈ positivePart x p) (hpos : ∀ z ∈ positivePart x p, ∃ u₀ ∈ nonnegativePart x p, z = u₀ * x) : ∃ u₀ ∈ nonnegativePart x p, 1 + u₀ * x ∈ canonicalRightIdeal x d N := by obtain ⟨u, v, hbez⟩ := step_4_2_bezout x p N h41 obtain ⟨-, hmem⟩ := step_4_2_residue x p d sigma N u v hd hbez obtain ⟨u₀, hu₀, hw⟩ := step_4_2_positive x p sigma N hright hleft hpos u v exact ⟨u₀, hu₀, hw ▸ hmem⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] [Algebra ℚ A]
/-- **Lemma 4.1.** For `x₁, p = ∂₁ ∈ A_n(k)` and `E = x₁p`, for every `m ≥ 1`, `x₁^m p^m = ∏_{i=0}^{m-1}(E−i)` and `p^m x₁^m = ∏_{i=1}^{m}(E+i)`, the factors in each product commuting with one another. The products are taken in the listed order of `i` (`A_n(k)` is not commutative); by the last two clauses the order does not matter. Paper `n` is Lean `n + 1`. -/ theorem lemma_4_1 (k : Type*) [Field k] (n m : ℕ) (hm : 1 ≤ m) : x₁ k n ^ m * p k n ^ m = ((List.range m).map fun i : ℕ => eulerElement (x₁ k n) (p k n) - i).prod ∧ p k n ^ m * x₁ k n ^ m = ((List.range' 1 m).map fun i : ℕ => eulerElement (x₁ k n) (p k n) + i).prod ∧ (∀ i j : ℕ, Commute (eulerElement (x₁ k n) (p k n) - i) (eulerElement (x₁ k n) (p k n) - j)) ∧ ∀ i j : ℕ, Commute (eulerElement (x₁ k n) (p k n) + i) (eulerElement (x₁ k n) (p k n) + j) := lemma_4_1_of_weylPair _ _ (convention_2_2_relations k n).1 m hm /-! ## The PBW form at one Ore pair stage Every element of `PairStage B` is a sum of PBW monomials `b x^a p^c` (coefficient on the left; `b x^a p^c` and `p^c x^a b` have the same Euler degree `a − c`). This section proves that the part of Euler degree `m` is spanned by the monomials of degree `m`. -/
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStage/-- An element of the form (3.1) is PBW-monic: it lies in `B_N` and the PBW coefficient of `∂₁^N` is `1`. This connects (3.1) to the repository's `WeylPBWMonicBridge.IsPBWMonicAt`, the hypothesis of `UniversalAssembly.CanonicalSupportVanishing`. -/ theorem isPBWMonicAt_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N dt := by obtain ⟨a, ha, rfl⟩ := h have hterm (j : ℕ) (hj : j ∈ Finset.range N) := principal_bernstein_p_pow_mul k (Finset.mem_range.mp hj).le ((bernsteinDegree_le_iff k (a j)).mp (ha j (Finset.mem_range.mp hj)).2) obtain ⟨hpN, hσN⟩ := p_pow_piece_principal k (@bernsteinWeight (n + 1)) rfl N refine ⟨add_mem hpN (Submodule.sum_mem _ fun j hj => (hterm j hj).1), ?_⟩ rw [← coeff_principal_p_pow k (@bernsteinWeight (n + 1)) rfl, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj rw [(hterm j hj).2] have h := coeff_X_pow_mul_single k hjN.le (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) (N - j) (a j)) simp only [MvPolynomial.coeff] at h rw [h] have h2 := coeff_principal_p_pow k (L := N - j) (@bernsteinWeight (n + 1)) rfl (a j) simp only [MvPolynomial.coeff] at h2 rw [h2] by_contra hne have := normalForm_C_no_p k (ha j hjN).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Elements of Euler degree `≥ 1` lie in `A_{≥0} x₁`.** In PBW form such an element is a sum of monomials `b x₁^a p^c` with `a > c`, and `b x₁^a p^c = b x₁^{a−c−1} · x₁ (x₁^c p^c)`, where `x₁` crosses the polynomial `x₁^c p^c` in `E` to the right by normality. -/ theorem step_positive_mem_nonnegative_mul_coordinate [Algebra ℚ k] : ∀ z ∈ presentedPositivePart k n, ∃ u₀ ∈ presentedNonnegativePart k n, z = u₀ * presentedCoordinate k n := by let := Stafford38.WeylEulerRemainder.iteratedRatAlgebra k n intro z hz obtain ⟨U, hU, hzU⟩ := pair_positivePart_eq_mul_coordinate (IteratedPairStage k n) (map_mem_positivePart (toIterated k n) (toIterated_coordinate k n) (toIterated_momentum k n) hz) refine ⟨fromIterated k n U, ?_, ?_⟩ · rw [step_nonnegative_pbw k n] exact iterated_eulerSubring_le k n U hU · have h := congrArg (fromIterated k n) hzU rwa [map_mul, fromIterated_coordinate, fromIterated_toIterated_ringHom] at h
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- The outer momentum exponents of the remainder `d - p^N` of a normalized presented Weyl operator are below `N`. -/ theorem presentedRemainder_outer_support_lt (n N : ℕ) {d : PresentedWeyl k (n + 1)} (hd : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N d) : ∀ j ∈ ((normalFormAddEquiv (coordinateDerivation (B := IteratedPairStage k n))).symm (presentedToIterated k (n + 1) d - stageMomentum k n ^ N)).support, j < N := by let H := presentedOuterPolynomial k n d have hnormal : normalForm (coordinateDerivation (B := IteratedPairStage k n)) (H - Polynomial.X ^ N) = presentedToIterated k (n + 1) d - stageMomentum k n ^ N := by change normalFormAddHom coordinateDerivation (H - Polynomial.X ^ N) = _ have hH : normalForm (coordinateDerivation (B := IteratedPairStage k n)) H = presentedToIterated k (n + 1) d := by exact (normalFormAddEquiv coordinateDerivation).apply_symm_apply _ have hH' : normalFormAddHom coordinateDerivation H = presentedToIterated k (n + 1) d := hH rw [map_sub, hH'] congr 1 rw [Polynomial.X_pow_eq_monomial] change normalForm coordinateDerivation (Polynomial.monomial N 1) = _ rw [normalForm_monomial] rw [map_one, one_mul] change pairMomentum (B := IteratedPairStage k n) ^ N = stageMomentum k n ^ N rfl rw [← hnormal] intro j hj change j ∈ ((normalFormAddEquiv (coordinateDerivation (B := IteratedPairStage k n))).symm ((normalFormAddEquiv (coordinateDerivation (B := IteratedPairStage k n))) (H - Polynomial.X ^ N))).support at hj rw [(normalFormAddEquiv (coordinateDerivation (B := IteratedPairStage k n))).symm_apply_apply] at hj exact support_sub_X_pow_lt (IteratedPairStage k n) H N (outer_coeff_eq_one_at_bound k n N hd) (fun j hj => outer_coeff_eq_zero_of_exponent_gt k n N hd hj) j hj
Variables and notation
Source ↗open PolynomialSource ↗open StaffordSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.EulerSurjectivitySource ↗variable (B : Type*) [Ring B] [Algebra ℚ B]Source ↗
local instance : Algebra ℚ (CoordinateStage (B := B)) :=Source ↗
local instance : Algebra ℚ (PairStage (B := B)) :=Source ↗
open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗universe uSource ↗
variable (k : Type u) [Field k] [Algebra ℚ k]Source ↗
local instance iteratedRatAlgebra : (n : ℕ) → Algebra ℚ (IteratedPairStage k n) | 0 => by change Algebra ℚ k infer_instance | n + 1 => by letI : Algebra ℚ (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra k (IteratedPairStage k n) :=Source ↗
local instance (n : ℕ) : Algebra ℚ (CoordinateStage (B := IteratedPairStage k n)) :=Source ↗
local instance (n : ℕ) : Algebra ℚ (PairStage (B := IteratedPairStage k n)) :=
/-- The PBW monomial `b x^a p^c` has Euler degree `a − c`. -/ theorem pairMonomial_mem_eulerDegree (b : B) (a c : ℕ) : pairCoefficient b * pairCoordinate ^ a * pairMomentum ^ c ∈ eulerDegree (pairCoordinate (B := B)) pairMomentum ((a : ℤ) - c) := mem_eulerDegree.mpr (Stafford38.PaperEulerGrading.orderedPBWMonomial_eulerWeight B b a c)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗variable (B : Type*) [Ring B]
/-- Multiplying an element of degree `≥ 1` by one of degree `0` keeps degree `≥ 1`. -/ theorem positivePart_mul_eulerDegree_zero {x p z y : A} (hz : z ∈ positivePart x p) (hy : y ∈ eulerDegree x p 0) : z * y ∈ positivePart x p := by refine Submodule.iSup_induction _ (motive := fun z => z * y ∈ positivePart x p) hz (fun m z hz => ?_) (by simp) (fun z z' h h' => by rw [add_mul]; exact add_mem h h') exact Submodule.mem_iSup_of_mem m (by simpa only [add_zero] using eulerDegree_mul hz hy)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A]
Comparison notes
lemma_4_2 states the lemma for d̃ in the form (3.1) (Section3.IsMonicNormalForm) and 𝓘 = d̃A_n + x₁^N d̃A_n (Section3.idealI, Equation (3.2)), over a field k of characteristic zero: there is u₀ ∈ A_{≥0} with 1 + u₀x₁ ∈ 𝓘. The proof uses (3.1) through its consequence that d̃ ∈ B_N has coefficient 1 at ∂₁^N (Section3.isPBWMonicAt_of_isMonicNormalForm). Lemma 4.1 is cited for m = N. The proof follows the three paragraphs: step_4_2_bezout (R_± coprime in ℚ[E], step_4_2_coprime; R_± are the products of Lemma 4.1, step_4_2_eval_products), step_4_2_residue (1 + w ∈ 𝓘) and step_4_2_positive (every term of w has positive Euler degree, so w = u₀x₁), assembled in lemma_4_2_of_positive. The third paragraph counts Euler degrees as printed: each PBW term b x₁^a p^j of Σ = d̃ − p^N has j < N, because d̃ ∈ B_N has coefficient 1 at ∂₁^N (WeylEulerRemainder.presentedRemainder_outer_support_lt), so after multiplication by x₁^N on either side it has Euler degree a + N − j ≥ N − j > 0; hence Σx₁^N and x₁^NΣ are sums of elements of Euler degree ≥ 1 (step_4_2_presented_remainder); u(E) and v(E) have Euler degree 0 (step_4_2_eulerPolynomial_degree_zero); and an element of positive Euler degree lies in A_{≥0}x₁ (step_positive_mem_nonnegative_mul_coordinate, from the paragraph before Lemma 4.1).
Review notes
Proposition 4.3 — Right multiplication by x₁ on Q is surjective
is stable under right multiplication by : If , then
Section4.proposition_4_3/-- **Proposition 4.3.** For `d̃` in the form (3.1), right multiplication by `x₁` on `Q = A_n/𝓘` (Equation (3.2)) is surjective, i.e. `Q = Qx₁`. -/ theorem proposition_4_3 [Algebra ℚ k] {N : ℕ} {d : PresentedWeyl k (n + 1)} (hform : Stafford38.Paper.Section3.IsMonicNormalForm k N d) : Function.Surjective (rightMul (Stafford38.Paper.Section3.idealI k N d) (presentedCoordinate k n)) := by obtain ⟨u₀, hu₀, hres⟩ := lemma_4_2 hform exact proposition_4_3_of (transverseCoordinateAlgebra k n) (presentedCoordinate k n) (presentedMomentum k n) u₀ (coordinate_mem_nonnegativePart (Stafford38.Paper.Section2.convention_2_2_relations k n).1) (euler_mem_nonnegativePart _ _) (transverseCoordinateAlgebra_le k n) (presented_coordinate_normal k n) hu₀ hres (closure_transverseCoordinateAlgebra_momentum k n)
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)Source ↗
variable {k n}More Lean proofs and definitions
/-- **`q ∈ (qA_{≥0})x₁`.** From `1 + u₀x₁ ∈ 𝓘`, `q = −q u₀ x₁`. -/ theorem step_4_3_q_mem (x u₀ : A) (hres : 1 + u₀ * x ∈ I) : qmk I 1 = rightMul I x (qmk I (-u₀)) := by rw [← qmk_right_mul] refine ((Submodule.Quotient.eq I).2 ?_).symm convert I.neg_mem hres using 1 noncomm_ring
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- **Equation (normal-surj): `qA_{≥0} = (qA_{≥0})x₁`.** `⊇` because `x₁ ∈ A_{≥0}`; `⊆` because `qr = −(qu₀)x₁r = −(qu₀r')x₁` for `x₁r = r'x₁`. -/ theorem step_4_3_normal_surj (x u₀ : A) (hx : x ∈ R) (hu₀ : u₀ ∈ R) (hres : 1 + u₀ * x ∈ I) (hnormal : ∀ r ∈ R, ∃ r' ∈ R, x * r = r' * x) : classSet I R = rightMul I x '' classSet I R := by ext z constructor · rintro ⟨r, hr, rfl⟩ obtain ⟨r', hr', hxr⟩ := hnormal r hr refine ⟨qmk I (-(u₀ * r')), ⟨_, R.neg_mem (R.mul_mem hu₀ hr'), rfl⟩, ?_⟩ calc rightMul I x (qmk I (-(u₀ * r'))) = qmk I (-u₀ * (r' * x)) := by rw [← qmk_right_mul]; congr 1; noncomm_ring _ = qmk I (-u₀ * x * r) := by rw [← hxr, mul_assoc] _ = rightMul I r (rightMul I x (qmk I (-u₀))) := by rw [← qmk_right_mul, ← qmk_right_mul] _ = qmk I r := by rw [← step_4_3_q_mem x u₀ hres, ← qmk_right_mul, one_mul] · rintro ⟨w, ⟨r, hr, rfl⟩, rfl⟩ exact ⟨r * x, R.mul_mem hr hx, (qmk_right_mul I r x).symm⟩
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- **Stability under `p`.** If `m = m'x₁` with `m' ∈ qA_{≥0}`, then `mp = m'(x₁p) = m'E ∈ qA_{≥0}`. -/ theorem step_4_3_stable_momentum (x p : A) (hE : x * p ∈ R) (hsurj : classSet I R = rightMul I x '' classSet I R) : ∀ z ∈ classSet I R, rightMul I p z ∈ classSet I R := by intro z hz rw [hsurj] at hz obtain ⟨_, ⟨r, hr, rfl⟩, rfl⟩ := hz refine ⟨r * (x * p), R.mul_mem hr hE, ?_⟩ rw [← qmk_right_mul, ← qmk_right_mul, mul_assoc]
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- **Stability under `C`.** `C = A_{n-1}[x₁] ⊆ A_{≥0}`. -/ theorem step_4_3_stable_coefficients (C : Subring A) (hC : C ≤ R) : ∀ c ∈ C, ∀ z ∈ classSet I R, rightMul I c z ∈ classSet I R := by rintro c hc _ ⟨r, hr, rfl⟩ exact ⟨r * c, R.mul_mem hr (hC hc), (qmk_right_mul I r c).symm⟩
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- **`qA_{≥0} = Q`.** `qA_{≥0}` is stable under `C` and `p`, which generate `A`. -/ theorem step_4_3_classSet_eq_univ (C : Subring A) (p : A) (hC : ∀ c ∈ C, ∀ z ∈ classSet I R, rightMul I c z ∈ classSet I R) (hp : ∀ z ∈ classSet I R, rightMul I p z ∈ classSet I R) (hgenerate : Subring.closure ((C : Set A) ∪ {p}) = ⊤) : classSet I R = Set.univ := by have htop : classSetStabilizer I R = ⊤ := by apply top_unique rw [← hgenerate] apply Subring.closure_le.2 rintro a (ha | ha) · exact hC a ha · rw [Set.mem_singleton_iff.mp ha] exact hp apply Set.eq_univ_of_forall intro z obtain ⟨a, rfl⟩ := Submodule.Quotient.mk_surjective I z have ha : a ∈ classSetStabilizer I R := htop ▸ Subring.mem_top a have h := ha (qmk I 1) ⟨1, R.one_mem, rfl⟩ rwa [← qmk_right_mul, one_mul] at h
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- `C = A_{n-1}(k)[x₁]`. -/ abbrev transverseCoordinateAlgebra : Subring (PresentedWeyl k (n + 1)) := Subring.closure ((transverse k n : Set (PresentedWeyl k (n + 1))) ∪ {presentedCoordinate k n})
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- `C ⊆ A_{≥0}`: no element of `C` contains `p`. -/ theorem transverseCoordinateAlgebra_le : transverseCoordinateAlgebra k n ≤ presentedNonnegativePart k n := by have hweyl := (Stafford38.Paper.Section2.convention_2_2_relations k n).1 apply Subring.closure_le.2 rintro a (ha | ha) · exact eulerSubring_le_nonnegativePart _ _ _ hweyl (fun c hc => (Stafford38.Paper.Section2.convention_2_2_relations k n).2.2 c hc) (coefficient_mem_eulerSubring _ _ _ ⟨a, ha⟩) · rw [Set.mem_singleton_iff.mp ha] exact coordinate_mem_nonnegativePart hweyl
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- `C` and `p` together generate `A_n`. -/ theorem closure_transverseCoordinateAlgebra_momentum [Algebra ℚ k] : Subring.closure ((transverseCoordinateAlgebra k n : Set (PresentedWeyl k (n + 1))) ∪ {presentedMomentum k n}) = ⊤ := by apply top_unique intro z _ have hz : presentedToIterated k (n + 1) z ∈ (⊤ : Subring (PairStage (B := IteratedPairStage k n))) := trivial let := Stafford38.WeylEulerRemainder.iteratedRatAlgebra k n rw [← closure_pairOldSubring_union_coordinate_momentum_eq_top] at hz have hle : Subring.closure (((pairOldSubring (IteratedPairStage k n) : Subring (PairStage (B := IteratedPairStage k n))) : Set _) ∪ {pairCoordinate, pairMomentum}) ≤ (Subring.closure ((transverseCoordinateAlgebra k n : Set (PresentedWeyl k (n + 1))) ∪ {presentedMomentum k n})).comap (fromIterated k n) := by apply Subring.closure_le.2 rintro a (⟨b, rfl⟩ | ha) · exact Subring.subset_closure (Or.inl (Subring.subset_closure (Or.inl (fromIterated_coefficient_mem_transverse k n b)))) · simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at ha rcases ha with rfl | rfl · rw [SetLike.mem_coe, Subring.mem_comap, fromIterated_coordinate] exact Subring.subset_closure (Or.inl (Subring.subset_closure (Or.inr rfl))) · rw [SetLike.mem_coe, Subring.mem_comap, fromIterated_momentum] exact Subring.subset_closure (Or.inr rfl) have h : fromIterated k n (presentedToIterated k (n + 1) z) ∈ Subring.closure ((transverseCoordinateAlgebra k n : Set (PresentedWeyl k (n + 1))) ∪ {presentedMomentum k n}) := Subring.mem_comap.mp (hle hz) rwa [fromIterated_toIterated] at h
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
/-- Proposition 4.3 for a ring with a right ideal, assembled from the paragraphs: equation (normal-surj) with `qA_{≥0} = Q` reads `Q = Qx₁`. -/ theorem proposition_4_3_of (C : Subring A) (x p u₀ : A) (hx : x ∈ R) (hE : x * p ∈ R) (hC : C ≤ R) (hnormal : ∀ r ∈ R, ∃ r' ∈ R, x * r = r' * x) (hu₀ : u₀ ∈ R) (hres : 1 + u₀ * x ∈ I) (hgenerate : Subring.closure ((C : Set A) ∪ {p}) = ⊤) : Function.Surjective (rightMul I x) := by have hsurj := step_4_3_normal_surj x u₀ hx hu₀ hres hnormal have huniv := step_4_3_classSet_eq_univ C p (step_4_3_stable_coefficients C hC) (step_4_3_stable_momentum x p hE hsurj) hgenerate intro z have hz : z ∈ rightMul I x '' classSet I R := by rw [← hsurj, huniv] trivial obtain ⟨w, -, hw⟩ := hz exact ⟨w, hw⟩
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗variable {A : Type*} [Ring A] (I : RightIdeal A) (R : Subring A)Source ↗
variable {I R}/-- **Lemma 4.2.** For `d̃` in the form (3.1) and `𝓘 = d̃A_n + x₁^N d̃A_n` (Equation (3.2)), there is `u₀ ∈ A_{≥0}` with `1 + u₀ x₁ ∈ 𝓘`. The proof uses (3.1) only through its consequence `IsPBWMonicAt` (`Section3.isPBWMonicAt_of_isMonicNormalForm`), and cites Lemma 4.1 for `m = N`. -/ theorem lemma_4_2 {N : ℕ} {d : PresentedWeyl k (n + 1)} (hform : Stafford38.Paper.Section3.IsMonicNormalForm k N d) : ∃ u₀ ∈ presentedNonnegativePart k n, 1 + u₀ * presentedCoordinate k n ∈ Stafford38.Paper.Section3.idealI k N d := by let := presentedRatAlgebra k n have hd := Stafford38.Paper.Section3.isPBWMonicAt_of_isMonicNormalForm k hform obtain ⟨hright, hleft⟩ := step_4_2_presented_remainder hd exact lemma_4_2_of_positive (presentedCoordinate k n) (presentedMomentum k n) d (d - presentedMomentum k n ^ N) N (fun hN => ⟨(lemma_4_1 k n N hN).1, (lemma_4_1 k n N hN).2.1⟩) (by abel) hright hleft (step_positive_mem_nonnegative_mul_coordinate k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] [Algebra ℚ k] (n : ℕ)Source ↗
variable {k n}/-- **Normality of `x₁` in `A_{≥0}` for `A_n`.** -/ theorem presented_coordinate_normal [Algebra ℚ k] : ∀ r ∈ presentedNonnegativePart k n, ∃ r' ∈ presentedNonnegativePart k n, presentedCoordinate k n * r = r' * presentedCoordinate k n := step_coordinate_normal _ _ _ (Stafford38.Paper.Section2.convention_2_2_relations k n).1 (transverse_commutes_coordinate k n) (step_nonnegative_pbw k n)
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)
Comparison notes
proposition_4_3 states the proposition for d̃ in the form (3.1) (Section3.IsMonicNormalForm) over a field k of characteristic zero: right multiplication by x₁ on Q = A_n/𝓘 (Section3.idealI, Equation (3.2)) is surjective. The proof follows the printed order: step_4_3_q_mem (q = −qu₀x₁, from Lemma 4.2), step_4_3_normal_surj (equation (normal-surj), using the normality of x₁), step_4_3_stable_momentum (stable under p), step_4_3_stable_coefficients (stable under C) and step_4_3_classSet_eq_univ (qA_{≥0} = Q), assembled in proposition_4_3_of. C = A_{n−1}[x₁] (transverseCoordinateAlgebra) lies in A_{≥0} (transverseCoordinateAlgebra_le), and C and p generate A_n (closure_transverseCoordinateAlgebra_momentum).
Review notes
Definition 4.4 — Localization at a multiplicative set
/-- **Definition 4.4 (localization).** Mathlib's `Localization S` and `LocalizedModule S N` are the rings and modules of fractions with `a/s = a'/s' ↔ t(s'a − sa') = 0` and `y/s = y'/s' ↔ t(s'y − sy') = 0` for some `t ∈ S`, and localization is exact. -/ theorem definition_4_4 (S : Submonoid R) {N : Type v} [AddCommGroup N] [Module R N] : (∀ (a a' : R) (s s' : S), Localization.mk a s = Localization.mk a' s' ↔ ∃ t : S, (t : R) * ((s' : R) * a - (s : R) * a') = 0) ∧ (∀ (y y' : N) (s s' : S), LocalizedModule.mk y s = LocalizedModule.mk y' s' ↔ ∃ t : S, t • (s' • y - s • y') = 0) ∧ ∀ {N₀ N₂ : Type v} [AddCommGroup N₀] [Module R N₀] [AddCommGroup N₂] [Module R N₂] (g : N₀ →ₗ[R] N) (h : N →ₗ[R] N₂), Function.Exact g h → Function.Exact (LocalizedModule.map S g) (LocalizedModule.map S h) := by refine ⟨fun a a' s s' => ?_, fun y y' s s' => ?_, fun g h ex => ?_⟩ · rw [Localization.mk_eq_mk_iff, Localization.r_iff_exists] simp only [mul_sub, sub_eq_zero] · rw [LocalizedModule.mk_eq] simp only [smul_sub, sub_eq_zero] · exact LocalizedModule.map_exact S g h ex
Comparison notes
definition_4_4: Mathlib's Localization S and LocalizedModule S N satisfy the printed equality of fractions (a/s = a′/s′ ⇔ t(s′a − sa′) = 0 for some t ∈ S, and likewise y/s = y′/s′), and localization of modules is exact. R_𝔭, N_𝔭 are these localizations at S = R ∖ 𝔭.
Review notes
Definition 4.5 — Support of a module
/-- **Definition 4.5 (support).** `Supp_R(M) = {𝔭 | M_𝔭 ≠ 0}`. -/ theorem definition_4_5 {M : Type*} [AddCommGroup M] [Module R M] (p : PrimeSpectrum R) : p ∈ Module.support R M ↔ ∃ z : LocalizedModule p.asIdeal.primeCompl M, z ≠ 0 := by rw [Module.mem_support_iff] exact nontrivial_iff_exists_ne 0
Comparison notes
definition_4_5: 𝔭 ∈ Module.support R M ⇔ M_𝔭 has a nonzero element, i.e. Supp_R(M) = {𝔭 : M_𝔭 ≠ 0}.
Review notes
Definition 4.6 — Length
Section4.definition_4_6;
Module.length (Mathlib); finite length of the localized modules:
AlgebraicAnalysis.localized_kernel_and_cokernel_isFiniteLength/-- **Definition 4.6 (length).** `Module.length` is the number of steps of a composition series `0 = N₀ ⊊ ⋯ ⊊ N_ℓ = N` (successive steps are coverings, i.e. have simple quotients, `covBy_iff_quot_is_simple`), independent of the series, and length is additive on short exact sequences. -/ theorem definition_4_6 {T : Type*} {M : Type v} [Ring T] [AddCommGroup M] [Module T M] : (∀ s : CompositionSeries (Submodule T M), s.head = ⊥ → s.last = ⊤ → (s.length : ℕ∞) = Module.length T M) ∧ ∀ {N P : Type v} [AddCommGroup N] [Module T N] [AddCommGroup P] [Module T P] (f : N →ₗ[T] M) (g : M →ₗ[T] P), Function.Injective f → Function.Surjective g → Function.Exact f g → Module.length T M = Module.length T N + Module.length T P := ⟨fun s h₁ h₂ => Module.length_compositionSeries s h₁ h₂, fun f g hf hg H => Module.length_eq_add_of_exact f g hf hg H⟩ /-! ## Lemma 4.7 -/
Comparison notes
definition_4_6: Mathlib's Module.length equals the length of every composition series from 0 to M (so it does not depend on the series), and length is additive on short exact sequences. Module.length is defined for every module, with value ∞ when there is no finite composition series; the paper defines it for modules of finite length.
Review notes
Lemma 4.7 — Support via the annihilator of an element
Section4.lemma_4_7
Suppose . Pick a nonzero element of . It has the form with and . The element is invertible in , so
For the next part, we write and denote the right multiplication by by . We furthermore define
/-- **Lemma 4.7.** `𝔭 ∈ Supp_R(N)` iff there is `y ∈ N` with `Ann_R(y) ⊆ 𝔭`. -/ theorem lemma_4_7 (p : PrimeSpectrum R) : p ∈ Module.support R N ↔ ∃ y : N, (R ∙ y).annihilator ≤ p.asIdeal := by rw [definition_4_5] constructor · -- (⇒) a nonzero `y/s` gives `y/1 ≠ 0` rintro ⟨z, hz⟩ induction z using LocalizedModule.induction_on with | h y s => refine ⟨y, (step_4_7_mk_one_ne_zero_iff p y).1 ?_⟩ rwa [ne_eq, ← step_4_7_mk_eq_zero_iff p y s] · -- (⇐) `y/1 ≠ 0` rintro ⟨y, hy⟩ exact ⟨_, (step_4_7_mk_one_ne_zero_iff p y).2 hy⟩
More Lean proofs and definitions
/-- In `N_𝔭`, `y/1 = 0` iff some `u ∉ 𝔭` kills `y`, i.e. iff `Ann_R(y) ⊄ 𝔭`. -/ theorem step_4_7_mk_one_eq_zero_iff (p : PrimeSpectrum R) (y : N) : LocalizedModule.mk y (1 : p.asIdeal.primeCompl) = 0 ↔ ¬ (R ∙ y).annihilator ≤ p.asIdeal := by rw [← LocalizedModule.zero_mk (1 : p.asIdeal.primeCompl), (definition_4_4 p.asIdeal.primeCompl (N := N)).2.1] simp only [one_smul, smul_zero, sub_zero, IsConcreteLE.le_iff, Submodule.mem_annihilator_span_singleton, not_forall] constructor · rintro ⟨⟨u, hu⟩, h⟩ exact ⟨u, h, hu⟩ · rintro ⟨u, h, hu⟩ exact ⟨⟨u, hu⟩, h⟩
/-- The contrapositive: `y/1 ≠ 0 ↔ Ann_R(y) ⊆ 𝔭`. -/ theorem step_4_7_mk_one_ne_zero_iff (p : PrimeSpectrum R) (y : N) : LocalizedModule.mk y (1 : p.asIdeal.primeCompl) ≠ 0 ↔ (R ∙ y).annihilator ≤ p.asIdeal := by rw [ne_eq, step_4_7_mk_one_eq_zero_iff, not_not]
/-- `s ∉ 𝔭` is invertible in `N_𝔭`, so `y/s = 0 ↔ y/1 = 0`. -/ theorem step_4_7_mk_eq_zero_iff (p : PrimeSpectrum R) (y : N) (s : p.asIdeal.primeCompl) : LocalizedModule.mk y s = 0 ↔ LocalizedModule.mk y (1 : p.asIdeal.primeCompl) = 0 := by have h : LocalizedModule.mk y s = (IsLocalization.mk' (Localization p.asIdeal.primeCompl) (1 : R) s) • LocalizedModule.mk y (1 : p.asIdeal.primeCompl) := by rw [← Localization.mk_eq_mk', LocalizedModule.mk_smul_mk, one_smul, mul_one] rw [h] constructor · intro h0 have := congrArg (fun z => (algebraMap R (Localization p.asIdeal.primeCompl) s) • z) h0 simp only [smul_zero, smul_smul] at this rwa [IsLocalization.mul_mk'_eq_mk'_of_mul, mul_one, IsLocalization.mk'_self, one_smul] at this · intro h0 rw [h0, smul_zero]
/-- **Definition 4.5 (support).** `Supp_R(M) = {𝔭 | M_𝔭 ≠ 0}`. -/ theorem definition_4_5 {M : Type*} [AddCommGroup M] [Module R M] (p : PrimeSpectrum R) : p ∈ Module.support R M ↔ ∃ z : LocalizedModule p.asIdeal.primeCompl M, z ≠ 0 := by rw [Module.mem_support_iff] exact nontrivial_iff_exists_ne 0
Comparison notes
lemma_4_7: for a commutative ring R, an R-module N and a prime 𝔭, 𝔭 ∈ Supp_R(N) ⇔ there is y ∈ N with Ann_R(y) ⊆ 𝔭 (Ann of the span of y). The proof is the printed one: y/1 = 0 ⇔ Ann_R(y) ⊄ 𝔭 (step_4_7_mk_one_eq_zero_iff), its contrapositive (step_4_7_mk_one_ne_zero_iff), and y/s = 0 ⇔ y/1 = 0 (step_4_7_mk_eq_zero_iff) for (⇒). Lemma 4.8 uses it.
Review notes
Setup — M = gr Q, T, V, U
For the next part, we write and denote the right multiplication by by . We furthermore define
/-- `M = gr Q`, the order-associated graded module of `Q = A_n/𝓘`. -/ abbrev GrQ := Graded k n N d
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
/-- `T = k[x₂,…,xₙ,ξ₂,…,ξₙ]`. -/ abbrev Tangential := T k n
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
More Lean proofs and definitions
/-- `f : M → M`, right multiplication by `x₁`, as a `T`-linear map. -/ abbrev coordinateMap : GrQ k n N d →ₗ[Tangential k n] GrQ k n N d := oldCoordinateMap (k := k) n (GrQ k n N d)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
/-- `V = ker f`. -/ abbrev kernelV := LinearMap.ker (coordinateMap k n N d)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
/-- `U = coker f = M/x₁M`. -/ abbrev cokernelU := GrQ k n N d ⧸ LinearMap.range (coordinateMap k n N d)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
/-- On `M = gr Q`, `f` is the action of the symbol `x₁`, i.e. right multiplication by `x₁` on `gr Q`. -/ theorem setup_coordinateMap_apply (z : GrQ k n N d) : coordinateMap k n N d z = symbolCoordinate k n • z := rfl
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
/-- `f` is the restriction to `T` of the `k[x,ξ]`-linear multiplication by `x₁`. -/ theorem setup_coordinateMap_eq : (LinearMap.lsmul (Symbol k n) (GrQ k n N d) (symbolCoordinate k n)).restrictScalars (Tangential k n) = coordinateMap k n N d := rfl
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable (k : Type*) [Field k] (n N : ℕ) (d : PresentedWeyl k (n + 1))
Comparison notes
Lean indexes coordinates from 0, so the paper's n is Lean's n + 1 and T = k[x₂, …, x_n, ξ₂, …, ξ_n] is MvPolynomial (Fin n ⊕ Fin n) k (T = k for paper n = 1). GrQ is M = gr Q, coordinateMap is f, the T-linear action of the symbol x₁ (setup_coordinateMap_apply, setup_coordinateMap_eq), kernelV = ker f and cokernelU = M/x₁M.
Review notes
Lemma 4.8 — Tangential finiteness
Let . By Lemma 4.7, we can choose with
. Let us now consider the annihilators of in , defining as the image of in : , because were and in , then would be annihilated by , which in turn would imply that is also in . Since , but we assumed , this is a contradiction. There must therefore exist a prime with and . Since , by Lemma 4.7, it follows that
From Vasconcelos' theorem [Matsumura 1986, Theorem 2.4] [Vasconcelos 1970] follows that :
We prove the contrapositive. Let be a prime of with
, i.e.\
.
Since localization is exact, , so
: the endomorphism
is therefore surjective.
The localization is a finite module over the Noetherian ring
, and is -linear, because
is multiplication by and is commutative. By Vasconcelos'
theorem, a surjective endomorphism of a finite module is injective. Hence
is injective, and by exactness of localization,
, i.e.\
. The contrapositive is therefore ,
then By
Lemma 4.7 there therefore exists an with
.
To get back to the support over , we define . Then , so
. If ,
the image of would lie in , which is
impossible. Hence , i.e.\
, and
Lemma 4.7 gives
. Since was arbitrary,
.
Let now be a minimal prime in . The primes of correspond to the primes of [Atiyah 1969, Prop 3.11], and . Hence corresponds to the primes lying in ; by minimality of the only such prime is itself, so , the set consisting only of the maximal ideal of . The same argument applies to : a prime in lies in by the inclusion just proved, hence equals , so .
Now is a Noetherian local ring with maximal ideal , and , are finite modules (localizations of the finite -modules , ) whose support is contained in . For a finite module over is equal to the vanishing locus of its annihilators [Authors 2026, Tag 00L2], so the only prime containing is the maximal ideal (if ). Hence is a Noetherian ring in which every prime is maximal, i.e. an Artinian ring, and a finite module over an Artinian ring has finite length [Atiyah 1969, Thm. 8.5][Authors 2026, Tag 00KH]. Thus and have finite length over ( possibly zero). ∎
Section4.lemma_4_8/-- **Lemma 4.8.** `U` and `V` are finite over `T`, `Supp_T V ⊆ Supp_T U`, and at a minimal prime `𝔮` of `Supp_T U`, `U_𝔮` and `V_𝔮` have finite length over `T_𝔮`. Here `d̃` is in the form (3.1), used through `IsPBWMonicAt`. -/ theorem lemma_4_8 (hform : IsMonicNormalForm k N d) : Module.Finite (Tangential k n) (cokernelU k n N d) ∧ Module.Finite (Tangential k n) (kernelV k n N d) ∧ Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) ∧ ∀ q ∈ Module.support (Tangential k n) (cokernelU k n N d), (∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) → IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ∧ IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by obtain ⟨hU, hV⟩ := step_4_8_finite hform have hsupp : Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) := by have h := step_4_8_support_inclusion (R := Tangential k n) (LinearMap.lsmul (Symbol k n) (GrQ k n N d) (symbolCoordinate k n)) rwa [setup_coordinateMap_eq] at h exact ⟨hU, hV, hsupp, fun q hqmem hq => step_4_8_finite_length hsupp q hqmem hq⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
More Lean proofs and definitions
/-- **First paragraph: `U` and `V` are finite over `T`.** `M = gr Q` is finite over `k[x,ξ] = T[x₁,ξ₁]` (cyclic generator `q̄`). Since `P` annihilates `M` (Proposition 3.6) and is monic of degree `N` in `ξ₁`, `M` is finite over `T[x₁]` (`canonical_orderAssociatedGradedModule_finite_normalCoeffRing_of_mem_annihilator`); `V` and `U` are then finite over the Noetherian `T[x₁]` and, being killed by `x₁`, over `T` (`canonical_finite_old_coordinate_kernel_cokernel_of_mem_annihilator`). -/ theorem step_4_8_finite (hform : IsMonicNormalForm k N d) : Module.Finite (Tangential k n) (cokernelU k n N d) ∧ Module.Finite (Tangential k n) (kernelV k n N d) := (canonical_finite_old_coordinate_kernel_cokernel_of_mem_annihilator (isPBWMonicAt_of_isMonicNormalForm k hform) (Stafford38.Paper.Section3.proposition_3_6 k hform (Stafford38.Paper.Section3.bernsteinDegree_eq_of_isMonicNormalForm k hform))).symm omit [Algebra ℚ k] in
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Second paragraph: `Supp_T V ⊆ Supp_T U`.** For `𝔭 ∈ Supp_T V` pick `v` with `Ann_T(v) ⊆ 𝔭` (Lemma 4.7); `Ann_{k[x,ξ]}(v)` misses `S = T∖𝔭`, so lies in a prime `𝔔` with `𝔔 ∩ S = ∅`; `𝔔 ∈ Supp(ker f)` (Lemma 4.7), hence `𝔔 ∈ Supp(coker f)` (Vasconcelos), so some `u ∈ U` has `Ann(u) ⊆ 𝔔` (Lemma 4.7), whence `Ann_T(u) ⊆ 𝔭` and `𝔭 ∈ Supp_T U` (Lemma 4.7). -/ theorem step_4_8_support_inclusion {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E] [IsNoetherianRing C] [Module.Finite C E] (f : Module.End C E) : Module.support R (LinearMap.ker (f.restrictScalars R)) ⊆ Module.support R (E ⧸ LinearMap.range (f.restrictScalars R)) := by intro p hp obtain ⟨v, hv⟩ := (lemma_4_7 p).mp hp let vC : LinearMap.ker f := ⟨v.1, v.2⟩ let I : Ideal C := (C ∙ vC).annihilator have hdisj : Disjoint (I : Set C) (p.asIdeal.primeCompl.map (algebraMap R C)) := by rw [Set.disjoint_left] rintro c hc ⟨r, hr, rfl⟩ have hrv : r • v = 0 := by apply Subtype.ext change r • v.1 = 0 have hcv : algebraMap R C r • vC = 0 := (Submodule.mem_annihilator_span_singleton _ _).mp hc simpa [vC, IsScalarTower.algebraMap_smul C r v.1] using congrArg Subtype.val hcv exact hr (hv ((Submodule.mem_annihilator_span_singleton _ _).mpr hrv)) obtain ⟨Q, hQprime, hIQ, hQdisj⟩ := I.exists_le_prime_disjoint _ hdisj let Q' : PrimeSpectrum C := ⟨Q, hQprime⟩ have hQker : Q' ∈ Module.support C (LinearMap.ker f) := (lemma_4_7 Q').mpr ⟨vC, hIQ⟩ have hQcoker := step_4_8_vasconcelos f hQker obtain ⟨u, hu⟩ := (lemma_4_7 Q').mp hQcoker refine (lemma_4_7 p).mpr ⟨u, fun r hr => ?_⟩ have hmap : algebraMap R C r ∈ (C ∙ u).annihilator := by rw [Submodule.mem_annihilator_span_singleton] have hru : r • u = 0 := (Submodule.mem_annihilator_span_singleton _ _).mp hr simpa [IsScalarTower.algebraMap_smul C r u] using hru by_contra hrp exact Set.disjoint_left.mp hQdisj (hu hmap) ⟨r, hrp, rfl⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)
/-- **Vasconcelos paragraph.** For a finite module `M` over a commutative Noetherian ring `C` and `f ∈ End_C M`, `Supp_C(ker f) ⊆ Supp_C(coker f)`: if `(coker f)_𝔔 = coker(f_𝔔) = 0`, then `f_𝔔` is a surjective endomorphism of a finite module, hence injective, and `(ker f)_𝔔 = ker(f_𝔔) = 0`. -/ theorem step_4_8_vasconcelos {C E : Type*} [CommRing C] [IsNoetherianRing C] [AddCommGroup E] [Module C E] [Module.Finite C E] (f : Module.End C E) : Module.support C (LinearMap.ker f) ⊆ Module.support C (E ⧸ LinearMap.range f) := by intro Q hQ by_contra hc let S := Q.asIdeal.primeCompl have hzero : Subsingleton (LocalizedModule S (E ⧸ LinearMap.range f)) := Module.notMem_support_iff.mp hc -- exactness of localization: `(coker f)_𝔔 = coker(f_𝔔)` have hexc := (definition_4_4 S).2.2 f (LinearMap.range f).mkQ (LinearMap.exact_map_mkQ_range f) have hsurj : Function.Surjective (LocalizedModule.map S f) := by intro z exact (hexc z).mp (Subsingleton.elim _ _) -- Vasconcelos: a surjective endomorphism of a finite module is injective have hinj : Function.Injective (LocalizedModule.map S f) := IsNoetherian.injective_of_surjective_endomorphism _ hsurj -- exactness again: `(ker f)_𝔔 = ker(f_𝔔) = 0` have hexk := (definition_4_4 S).2.2 (LinearMap.ker f).subtype f (LinearMap.exact_subtype_ker_map f) have hsub : Subsingleton (LocalizedModule S (LinearMap.ker f)) := by refine ⟨fun a b => LocalizedModule.map_injective S _ (Submodule.subtype_injective _) ?_⟩ have ha : LocalizedModule.map S f (LocalizedModule.map S (LinearMap.ker f).subtype a) = LocalizedModule.map S f (LocalizedModule.map S (LinearMap.ker f).subtype b) := by rw [hexk.apply_apply_eq_zero, hexk.apply_apply_eq_zero] exact hinj ha exact Module.notMem_support_iff.mpr hsub hQ
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)
/-- A minimal prime of `Supp_T L` (`L` finite over the Noetherian `T`) is minimal over `Ann_T L`, since `Supp L = V(Ann L)` [Stacks 00L2], so `L_𝔮` has finite length [Atiyah–Macdonald 8.5, Stacks 00KH]. -/ theorem step_4_8_localized_isFiniteLength {R L : Type*} [CommRing R] [IsNoetherianRing R] [AddCommGroup L] [Module R L] [Module.Finite R L] (q : PrimeSpectrum R) (hqmem : q ∈ Module.support R L) (hq : ∀ p ∈ Module.support R L, p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl L) := by have hmin : q.asIdeal ∈ (Module.annihilator R L).minimalPrimes := by obtain ⟨p, hp, hpq⟩ := Ideal.exists_minimalPrimes_le (Module.mem_support_iff_of_finite.mp hqmem) have hp' : (⟨p, hp.1.1⟩ : PrimeSpectrum R) ∈ Module.support R L := Module.mem_support_iff_of_finite.mpr hp.1.2 have heq : p = q.asIdeal := le_antisymm hpq (hq _ hp' hpq) simpa [heq] using hp exact AlgebraicAnalysis.MinimalPrimeFiniteLengthLocalization.localizedModule_isFiniteLength q.asIdeal hmin
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)
/-- **Third paragraph.** At a minimal prime `𝔮` of `Supp_T U`, the supports of `U_𝔮` and `V_𝔮` are contained in `{𝔮T_𝔮}` (for `V` by `Supp V ⊆ Supp U`), so both have finite length over `T_𝔮` (`V_𝔮` possibly zero). -/ theorem step_4_8_finite_length {R U V : Type*} [CommRing R] [IsNoetherianRing R] [AddCommGroup U] [Module R U] [Module.Finite R U] [AddCommGroup V] [Module R V] [Module.Finite R V] (hsupp : Module.support R V ⊆ Module.support R U) (q : PrimeSpectrum R) (hqmem : q ∈ Module.support R U) (hq : ∀ p ∈ Module.support R U, p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl U) ∧ IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl V) := by refine ⟨step_4_8_localized_isFiniteLength q hqmem hq, ?_⟩ by_cases hk : q ∈ Module.support R V · exact step_4_8_localized_isFiniteLength q hk fun p hp hpq => hq p (hsupp hp) hpq · have : Subsingleton (LocalizedModule q.asIdeal.primeCompl V) := Module.notMem_support_iff.mp hk exact IsFiniteLength.of_subsingleton
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)
/-- **Lemma 4.7.** `𝔭 ∈ Supp_R(N)` iff there is `y ∈ N` with `Ann_R(y) ⊆ 𝔭`. -/ theorem lemma_4_7 (p : PrimeSpectrum R) : p ∈ Module.support R N ↔ ∃ y : N, (R ∙ y).annihilator ≤ p.asIdeal := by rw [definition_4_5] constructor · -- (⇒) a nonzero `y/s` gives `y/1 ≠ 0` rintro ⟨z, hz⟩ induction z using LocalizedModule.induction_on with | h y s => refine ⟨y, (step_4_7_mk_one_ne_zero_iff p y).1 ?_⟩ rwa [ne_eq, ← step_4_7_mk_eq_zero_iff p y s] · -- (⇐) `y/1 ≠ 0` rintro ⟨y, hy⟩ exact ⟨_, (step_4_7_mk_one_ne_zero_iff p y).2 hy⟩
/-- **Proposition 3.6.** Equip `Q` with the order filtration `Q_m = qF_m` induced by the cyclic generator `q = [1]`, and write `P = σ_N^ord(d̃)`. Then `P ∈ Ann_{k[x,ξ]} gr Q` (eq. (3.5)). -/ theorem proposition_3_6 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : orderSymbol k N dt ∈ Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt) := by obtain ⟨⟨hF, -⟩, -⟩ := lemma_3_5 k hform hdeg rw [Module.mem_annihilator] intro m obtain ⟨f, rfl⟩ := (step_3_6_generated k dt).2.2 m rw [smul_comm, step_3_6_kills k hF, smul_zero]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- In (3.1), `N = deg_B d̃`: `d̃ ∈ B_N` and its Bernstein symbol has the coefficient `1` at `ξ₁^N`. -/ theorem bernsteinDegree_eq_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : bernsteinDegree k dt = N := by have hd := isPBWMonicAt_of_isMonicNormalForm k h refine bernsteinDegree_eq_of_principal_ne_zero k hd.1 fun h0 => ?_ have hc : AddMonoidAlgebra.coeff (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) (Finsupp.single (.inr (0 : Fin (n + 1))) N) = 1 := by rw [coeff_presentedPrincipalComponent, ite_eq_left (by simp [monomialWeight, bernsteinWeight])] exact hd.2 rw [h0] at hc exact zero_ne_one hc
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- An element of the form (3.1) is PBW-monic: it lies in `B_N` and the PBW coefficient of `∂₁^N` is `1`. This connects (3.1) to the repository's `WeylPBWMonicBridge.IsPBWMonicAt`, the hypothesis of `UniversalAssembly.CanonicalSupportVanishing`. -/ theorem isPBWMonicAt_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N dt := by obtain ⟨a, ha, rfl⟩ := h have hterm (j : ℕ) (hj : j ∈ Finset.range N) := principal_bernstein_p_pow_mul k (Finset.mem_range.mp hj).le ((bernsteinDegree_le_iff k (a j)).mp (ha j (Finset.mem_range.mp hj)).2) obtain ⟨hpN, hσN⟩ := p_pow_piece_principal k (@bernsteinWeight (n + 1)) rfl N refine ⟨add_mem hpN (Submodule.sum_mem _ fun j hj => (hterm j hj).1), ?_⟩ rw [← coeff_principal_p_pow k (@bernsteinWeight (n + 1)) rfl, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj rw [(hterm j hj).2] have h := coeff_X_pow_mul_single k hjN.le (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) (N - j) (a j)) simp only [MvPolynomial.coeff] at h rw [h] have h2 := coeff_principal_p_pow k (L := N - j) (@bernsteinWeight (n + 1)) rfl (a j) simp only [MvPolynomial.coeff] at h2 rw [h2] by_contra hne have := normalForm_C_no_p k (ha j hjN).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
lemma_4_8 states the lemma as printed: U and V are finite over T, Supp_T V ⊆ Supp_T U, and at every minimal prime 𝔮 of Supp_T U, U_𝔮 and V_𝔮 have finite length over T_𝔮. Lean states it for d̃ in the form (3.1) (Section3.IsMonicNormalForm). The proof uses (3.1) through Proposition 3.6 (P annihilates M; deg_B d̃ = N by Section3.bernsteinDegree_eq_of_isMonicNormalForm) and through its consequence that d̃ ∈ B_N has coefficient 1 at ∂₁^N (Section3.isPBWMonicAt_of_isMonicNormalForm). The proof follows the paragraphs: step_4_8_finite (M finite over T[x₁] because P annihilates M and is monic in ξ₁), step_4_8_support_inclusion (Lemma 4.7 and Vasconcelos' theorem, step_4_8_vasconcelos), step_4_8_localized_isFiniteLength and step_4_8_finite_length.
Review notes
Lemma 4.9 — Filtered-page inequality ℓ(U_q) ≤ ℓ(V_q)
Consider the filtered complex , with decreasing filtration , where is the order filtration. Since has order , for every ; the induced map on , summed over , is exactly the action of on . This is the crudest approximation to and : it correctly identifies as approximations, but it does not yet see elements that vanish to higher order than this naive count predicts. The spectral sequence refines this approximation in successive stages, called pages (see also [Weibel 1994, Construction 5.4.6]): for each , the source and target pages are
Applying to a representative gives a well-defined map , shifting the summand index by ; one checks directly from these definitions that
These pages carry a -module structure. The pages are subquotients of , which a priori carries an action of the transverse Weyl algebra by right multiplication. This action commutes with , since the transverse algebra commutes with , so the maps are compatible with it. Two transverse elements commute only up to a commutator, which lowers the order filtration by one. On the page quotients such lower-order terms vanish, so the induced actions commute, and the commutative ring acts on every and , via the symbols of the transverse generators. The maps and the successor-page maps are -linear. In particular, this means that localizing at and measuring lengths over is a well-defined action for the pages.
Now we fix a minimal prime in and localize at . We localize the -modules and their -linear maps. Exactness of localization preserves the successor-page kernel and cokernel identities. By Lemma 4.8, and have finite length over ; since each and is a quotient of , all later pages have finite length as well.
The maps are surjective by construction.
Let . Since factors through , we have
, so is an increasing chain of submodules of .
This chain exhausts : Before localization, take a homogeneous
class in , represented by . Since by Proposition 4.3, choose one preimage
. Exhaustiveness of the filtration means that must be in for some ; choose with
. Then , so belongs to the denominator of
and its class vanishes. Every class is a finite sum of homogeneous
classes and hence vanishes on a sufficiently late page. The same holds for a
localized class , since its numerator already vanishes on such a page.
As was arbitrary, so is in , and .
Since has finite length, it is a Noetherian -module. By the ascending
chain condition, the increasing chain stabilizes: there exists a finite with
for all . Combined with , stabilization forces .
Finally, means must be the zero map: by the first isomorphism theorem,
, and gives , so
. But is also surjective, so .
At each stage the differential gives an exact sequence
Summing this identity over and reindexing the sums, most terms cancel on both sides, leaving
Lean: Section4.lemma_4_9;
exhaustion AlgebraicAnalysis.exists_uniform_subsingleton_localized
On the other hand, we can also establish a different inequality:
/-- **Lemma 4.9.** For every minimal prime `𝔮` of `Supp_T U`, `ℓ_{T_𝔮}(U_𝔮) ≤ ℓ_{T_𝔮}(V_𝔮)`, for `d̃` in the form (3.1). -/ theorem lemma_4_9 (hform : IsMonicNormalForm k N d) (q : PrimeSpectrum (Tangential k n)) (hqmem : q ∈ Module.support (Tangential k n) (cokernelU k n N d)) (hqmin : ∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ≤ Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by have hd := isPBWMonicAt_of_isMonicNormalForm k hform let S := q.asIdeal.primeCompl let A : ℕ → Type _ := fun r => LocalizedModule S ((complex k n N d hd).SourceTotal r) let B : ℕ → Type _ := fun r => LocalizedModule S ((complex k n N d hd).TargetTotal r) let : ∀ r, AddCommGroup (A r) := fun r => by dsimp only [A]; infer_instance let : ∀ r, Module (Localization S) (A r) := fun r => by dsimp only [A]; infer_instance let : ∀ r, AddCommGroup (B r) := fun r => by dsimp only [B]; infer_instance let : ∀ r, Module (Localization S) (B r) := fun r => by dsimp only [B]; infer_instance -- the pages and their localization; exactness keeps the successor identities obtain ⟨hsucc, htsucc, ⟨eV⟩, ⟨eU⟩⟩ := step_4_9_pages hd let dd : ∀ r, A r →ₗ[Localization S] B r := fun r => localizedMap S (tangentialDrop k n N d hd r) let es : ∀ r, A (r + 1) ≃ₗ[Localization S] LinearMap.ker (dd r) := fun r => (localizedEquiv S (hsucc r).some).trans (localizedKernelEquiv S _) let et : ∀ r, B (r + 1) ≃ₗ[Localization S] B r ⧸ LinearMap.range (dd r) := fun r => (localizedEquiv S (htsucc r).some).trans (localizedCokernelEquiv S _) let eV' : A 1 ≃ₗ[Localization S] LocalizedModule S (kernelV k n N d) := localizedEquiv S eV let eU' : B 1 ≃ₗ[Localization S] LocalizedModule S (cokernelU k n N d) := localizedEquiv S eU -- Lemma 4.8: `(A_1)_𝔮 = V_𝔮` and `(B_1)_𝔮 = U_𝔮` have finite length obtain ⟨-, -, -, hfin⟩ := lemma_4_8 hform obtain ⟨hU, hV⟩ := hfin q hqmem hqmin have hfl := step_4_9_pages_finite_length dd es et (eV'.symm.isFiniteLength hV) (eU'.symm.isFiniteLength hU) -- exhaustion: `(B_R)_𝔮 = 0` let : IsNoetherian (Localization S) (B 1) := (isFiniteLength_iff_isNoetherian_isArtinian.mp (hfl 1 le_rfl).2).1 obtain ⟨R', hR'⟩ := step_4_9_exhaustion hform S -- equation (additivity) and the telescoping sum have hlen := step_4_9_telescope (fun r => if 1 ≤ r then Module.length (Localization S) (A r) else 0) (fun r => if 1 ≤ r then Module.length (Localization S) (B r) else 0) (fun r => by split_ifs with h · exact Module.length_ne_top_iff.mpr (hfl r h).1 · exact ENat.zero_ne_top) (fun r => by split_ifs with h · exact Module.length_ne_top_iff.mpr (hfl r h).2 · exact ENat.zero_ne_top) (fun r hr => by simp only [hr, ite_true, show 1 ≤ r + 1 by omega] exact step_4_9_additivity (dd r) (es r) (et r)) (R' + 1) (by omega) simp only [le_refl, ite_true, show 1 ≤ R' + 1 by omega] at hlen have hBR : Module.length (Localization S) (B (R' + 1)) = 0 := Module.length_eq_zero_iff.mpr hR' -- `ℓ(U_𝔮) + ℓ((A_R)_𝔮) = ℓ(V_𝔮)`, and `ℓ((A_R)_𝔮) ≥ 0` calc Module.length (Localization S) (LocalizedModule S (cokernelU k n N d)) = Module.length (Localization S) (B 1) := eU'.length_eq.symm _ ≤ Module.length (Localization S) (B 1) + Module.length (Localization S) (A (R' + 1)) := le_self_add _ = Module.length (Localization S) (A 1) := by rw [hlen, hBR, add_zero] _ = Module.length (Localization S) (LocalizedModule S (kernelV k n N d)) := eV'.length_eq
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
More Lean proofs and definitions
/-- **The pages and their `T`-structure.** `A_{r+1} = ker d_r` and `B_{r+1} = coker d_r`, `T`-linearly, and `A_1 ≅ V`, `B_1 ≅ U`. The `T`-action on the pages comes from the symbols of the transverse generators, whose commutators lower the filtration (`generator_commutator_lowers`). -/ theorem step_4_9_pages (hd : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N d) : (∀ r, Nonempty ((complex k n N d hd).SourceTotal (r + 1) ≃ₗ[Tangential k n] LinearMap.ker (tangentialDrop k n N d hd r))) ∧ (∀ r, Nonempty ((complex k n N d hd).TargetTotal (r + 1) ≃ₗ[Tangential k n] ((complex k n N d hd).TargetTotal r ⧸ LinearMap.range (tangentialDrop k n N d hd r)))) ∧ Nonempty ((complex k n N d hd).SourceTotal 1 ≃ₗ[Tangential k n] kernelV k n N d) ∧ Nonempty ((complex k n N d hd).TargetTotal 1 ≃ₗ[Tangential k n] cokernelU k n N d) := ⟨fun r => ⟨tangentialSourceSuccEquiv k n N d hd r⟩, fun r => ⟨tangentialTargetSuccEquiv k n N d hd r⟩, ⟨firstSourceGradedKernelEquiv k n N d hd⟩, ⟨firstTargetGradedCokernelEquiv k n N d hd⟩⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Later pages have finite length.** `A_{r+1} ⊆ A_r` and `B_{r+1}` is a quotient of `B_r`, so finite length of `A_1, B_1` passes to all `r ≥ 1`. -/ theorem step_4_9_pages_finite_length (d : ∀ r, A r →ₗ[R] B r) (sourceSucc : ∀ r, A (r + 1) ≃ₗ[R] LinearMap.ker (d r)) (targetSucc : ∀ r, B (r + 1) ≃ₗ[R] B r ⧸ LinearMap.range (d r)) (hA1 : IsFiniteLength R (A 1)) (hB1 : IsFiniteLength R (B 1)) : ∀ r, 1 ≤ r → IsFiniteLength R (A r) ∧ IsFiniteLength R (B r) := by intro r hr induction r, hr using Nat.le_induction with | base => exact ⟨hA1, hB1⟩ | succ r _ ih => exact ⟨(sourceSucc r).symm.isFiniteLength (IsFiniteLength.of_injective ih.1 (Submodule.subtype_injective _)), (targetSucc r).symm.isFiniteLength (IsFiniteLength.of_surjective ih.2 (Submodule.mkQ_surjective _))⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {R : Type*} [Ring R] {A B : ℕ → Type*} [∀ r, AddCommGroup (A r)] [∀ r, Module R (A r)] [∀ r, AddCommGroup (B r)] [∀ r, Module R (B r)]
/-- `f : Q → Q` of the filtered complex is onto: this is `Q = Qx₁` (Proposition 4.3). -/ theorem step_4_9_f_surjective (hform : IsMonicNormalForm k N d) : Function.Surjective (complex k n N d (isPBWMonicAt_of_isMonicNormalForm k hform)).f := by intro q obtain ⟨z, rfl⟩ := Submodule.Quotient.mk_surjective (rightIdealKSubmodule k (presentedCanonicalRightIdeal (k := k) n N d)) q obtain ⟨y, hy⟩ := proposition_4_3 hform (qmk (presentedCanonicalRightIdeal (k := k) n N d) z) obtain ⟨y', hy'⟩ := Submodule.Quotient.mk_surjective (rightIdealKSubmodule k (presentedCanonicalRightIdeal (k := k) n N d)) y rw [← hy'] at hy refine ⟨Submodule.Quotient.mk y', ?_⟩ change rightMulLinearMap k (presentedCanonicalRightIdeal (k := k) n N d) (presentedCoordinate k n) (Submodule.Quotient.mk y') = Submodule.Quotient.mk z rw [rightMulLinearMap_mk] exact hy
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Exhaustion: `(B_R)_𝔮 = 0` for some `R`.** The maps `B_1 → B_r` are onto with increasing kernels `K_r`. A homogeneous class `z ∈ G^p` of `B_1` has a preimage `z = wx₁` (Proposition 4.3) with `w ∈ G^s`, so `z` vanishes in `B_r` once `p − r + 1 ≤ s`; hence `⋃ K_r = (B_1)_𝔮`. As `(B_1)_𝔮` is Noetherian the chain stabilizes at some `K_R = (B_1)_𝔮`, and `(B_R)_𝔮 = 0`. -/ theorem step_4_9_exhaustion (hform : IsMonicNormalForm k N d) (S : Submonoid (Tangential k n)) [IsNoetherian (Localization S) (LocalizedModule S ((complex k n N d (isPBWMonicAt_of_isMonicNormalForm k hform)).TargetTotal 1))] : ∃ R, Subsingleton (LocalizedModule S ((complex k n N d (isPBWMonicAt_of_isMonicNormalForm k hform)).TargetTotal (R + 1))) := let hd := isPBWMonicAt_of_isMonicNormalForm k hform AlgebraicAnalysis.exists_uniform_subsingleton_localized S (fun r => tangentialBoundaryMap k n N d hd r) (tangentialBoundaryMap_ker_mono k n N d hd) (fun z => (complex k n N d hd).totalBoundaryMap_eventually_zero (canonicalOrderFiltration_exhaustive k _) (step_4_9_f_surjective hform) z) (fun r => tangentialBoundaryMap_surjective k n N d hd r)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Equation (additivity).** `0 → A_{r+1} → A_r → B_r → B_{r+1} → 0` splits at `im d_r` into two short exact sequences; additivity of length (Definition 4.6) gives `ℓ(B_r) + ℓ(A_{r+1}) = ℓ(A_r) + ℓ(B_{r+1})`. -/ theorem step_4_9_additivity {R : Type*} [Ring R] {A₀ A₁ B₀ B₁ : Type v} [AddCommGroup A₀] [Module R A₀] [AddCommGroup A₁] [Module R A₁] [AddCommGroup B₀] [Module R B₀] [AddCommGroup B₁] [Module R B₁] (d : A₀ →ₗ[R] B₀) (es : A₁ ≃ₗ[R] LinearMap.ker d) (et : B₁ ≃ₗ[R] B₀ ⧸ LinearMap.range d) : Module.length R B₀ + Module.length R A₁ = Module.length R A₀ + Module.length R B₁ := by have hsource : Module.length R A₀ = Module.length R (LinearMap.ker d) + Module.length R (LinearMap.range d) := (definition_4_6 (T := R) (M := A₀)).2 (LinearMap.ker d).subtype d.rangeRestrict (Submodule.subtype_injective _) (LinearMap.range_eq_top.mp (LinearMap.range_rangeRestrict d)) (by rw [LinearMap.exact_iff, Submodule.range_subtype, LinearMap.ker_rangeRestrict]) have htarget : Module.length R B₀ = Module.length R (LinearMap.range d) + Module.length R (B₀ ⧸ LinearMap.range d) := (definition_4_6 (T := R) (M := B₀)).2 (LinearMap.range d).subtype (LinearMap.range d).mkQ (Submodule.subtype_injective _) (Submodule.mkQ_surjective _) (LinearMap.exact_subtype_mkQ _) rw [es.length_eq, et.length_eq, hsource, htarget] ac_rfl
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {R : Type*} [Ring R] {A B : ℕ → Type*} [∀ r, AddCommGroup (A r)] [∀ r, Module R (A r)] [∀ r, AddCommGroup (B r)] [∀ r, Module R (B r)]
/-- **Summing (additivity) over `r = 1,…,R−1`.** The intermediate terms cancel: `ℓ(B_1) + ℓ(A_R) = ℓ(A_1) + ℓ(B_R)`. -/ theorem step_4_9_telescope (a b : ℕ → ℕ∞) (ha : ∀ r, a r ≠ ⊤) (hb : ∀ r, b r ≠ ⊤) (hadd : ∀ r, 1 ≤ r → b r + a (r + 1) = a r + b (r + 1)) : ∀ R, 1 ≤ R → b 1 + a R = a 1 + b R := by have ha' : ∀ r, a r = ((a r).toNat : ℕ∞) := fun r => (ENat.natCast_toNat (ha r)).symm have hb' : ∀ r, b r = ((b r).toNat : ℕ∞) := fun r => (ENat.natCast_toNat (hb r)).symm have hadd' : ∀ r, 1 ≤ r → (b r).toNat + (a (r + 1)).toNat = (a r).toNat + (b (r + 1)).toNat := fun r hr => by have h := hadd r hr rw [ha' r, ha' (r + 1), hb' r, hb' (r + 1)] at h exact_mod_cast h intro R hR rw [ha' 1, ha' R, hb' 1, hb' R] norm_cast induction R, hR using Nat.le_induction with | base => exact add_comm _ _ | succ R hR ih => have h := hadd' R hR omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {R : Type*} [Ring R] {A B : ℕ → Type*} [∀ r, AddCommGroup (A r)] [∀ r, Module R (A r)] [∀ r, AddCommGroup (B r)] [∀ r, Module R (B r)]
/-- **Lemma 4.8.** `U` and `V` are finite over `T`, `Supp_T V ⊆ Supp_T U`, and at a minimal prime `𝔮` of `Supp_T U`, `U_𝔮` and `V_𝔮` have finite length over `T_𝔮`. Here `d̃` is in the form (3.1), used through `IsPBWMonicAt`. -/ theorem lemma_4_8 (hform : IsMonicNormalForm k N d) : Module.Finite (Tangential k n) (cokernelU k n N d) ∧ Module.Finite (Tangential k n) (kernelV k n N d) ∧ Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) ∧ ∀ q ∈ Module.support (Tangential k n) (cokernelU k n N d), (∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) → IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ∧ IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by obtain ⟨hU, hV⟩ := step_4_8_finite hform have hsupp : Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) := by have h := step_4_8_support_inclusion (R := Tangential k n) (LinearMap.lsmul (Symbol k n) (GrQ k n N d) (symbolCoordinate k n)) rwa [setup_coordinateMap_eq] at h exact ⟨hU, hV, hsupp, fun q hqmem hq => step_4_8_finite_length hsupp q hqmem hq⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Proposition 4.3.** For `d̃` in the form (3.1), right multiplication by `x₁` on `Q = A_n/𝓘` (Equation (3.2)) is surjective, i.e. `Q = Qx₁`. -/ theorem proposition_4_3 [Algebra ℚ k] {N : ℕ} {d : PresentedWeyl k (n + 1)} (hform : Stafford38.Paper.Section3.IsMonicNormalForm k N d) : Function.Surjective (rightMul (Stafford38.Paper.Section3.idealI k N d) (presentedCoordinate k n)) := by obtain ⟨u₀, hu₀, hres⟩ := lemma_4_2 hform exact proposition_4_3_of (transverseCoordinateAlgebra k n) (presentedCoordinate k n) (presentedMomentum k n) u₀ (coordinate_mem_nonnegativePart (Stafford38.Paper.Section2.convention_2_2_relations k n).1) (euler_mem_nonnegativePart _ _) (transverseCoordinateAlgebra_le k n) (presented_coordinate_normal k n) hu₀ hres (closure_transverseCoordinateAlgebra_momentum k n)
Variables and notation
Source ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerSubringSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylEulerRemainderSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreIteratedPairStageSource ↗universe uSource ↗
variable (k : Type u) [Field k] (n : ℕ)Source ↗
variable {k n}/-- An element of the form (3.1) is PBW-monic: it lies in `B_N` and the PBW coefficient of `∂₁^N` is `1`. This connects (3.1) to the repository's `WeylPBWMonicBridge.IsPBWMonicAt`, the hypothesis of `UniversalAssembly.CanonicalSupportVanishing`. -/ theorem isPBWMonicAt_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N dt := by obtain ⟨a, ha, rfl⟩ := h have hterm (j : ℕ) (hj : j ∈ Finset.range N) := principal_bernstein_p_pow_mul k (Finset.mem_range.mp hj).le ((bernsteinDegree_le_iff k (a j)).mp (ha j (Finset.mem_range.mp hj)).2) obtain ⟨hpN, hσN⟩ := p_pow_piece_principal k (@bernsteinWeight (n + 1)) rfl N refine ⟨add_mem hpN (Submodule.sum_mem _ fun j hj => (hterm j hj).1), ?_⟩ rw [← coeff_principal_p_pow k (@bernsteinWeight (n + 1)) rfl, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj rw [(hterm j hj).2] have h := coeff_X_pow_mul_single k hjN.le (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) (N - j) (a j)) simp only [MvPolynomial.coeff] at h rw [h] have h2 := coeff_principal_p_pow k (L := N - j) (@bernsteinWeight (n + 1)) rfl (a j) simp only [MvPolynomial.coeff] at h2 rw [h2] by_contra hne have := normalForm_C_no_p k (ha j hjN).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
lemma_4_9 states the lemma as printed: ℓ_{T_𝔮}(U_𝔮) ≤ ℓ_{T_𝔮}(V_𝔮) for every minimal prime 𝔮 of Supp_T U. Lean states it for d̃ in the form (3.1) (Section3.IsMonicNormalForm). The proof uses (3.1) through its consequence that d̃ ∈ B_N has coefficient 1 at ∂₁^N (Section3.isPBWMonicAt_of_isMonicNormalForm). The proof follows the printed one: the pages A_r, B_r with their T-structure and A_1 ≅ V, B_1 ≅ U (step_4_9_pages), finite length of later pages (step_4_9_pages_finite_length), surjectivity of f from Proposition 4.3 (step_4_9_f_surjective), exhaustion (B_R)_𝔮 = 0 (step_4_9_exhaustion), equation (additivity) (step_4_9_additivity) and the telescoping sum (step_4_9_telescope); Lemma 4.8 gives the finite lengths.
Review notes
Lemma 4.10 — Strict inequality ℓ(U_q) > ℓ(V_q)
Let us consider . By Gabber's Theorem (2.10, [Gabber 1981]), is involutive.
If is a minimal prime of ,
then is itself involutive, i.e. [Hotta 2008, Thm. D.3.4].
Lean:
Section4.lemma_4_10
From the involutivity of follows that no contains : Assume . Since , also the Poisson brackets
Consider the increasing chain of kernels of powers of the localization of right multiplication with . Since is Noetherian, this chain stabilizes. Let at a stage where it stabilizes, so and . Each quotient maps to injectively via , so, as has finite length, so does each quotient and hence itself.
Two elementary facts about :
- : if then (as ); if then , so .
- : "" is trivial. For "", if for some , then , so .
For any endomorphism of a finite-length module , additivity along and gives . We can apply this to : since , we have , so with
On the other hand, the natural surjection has the kernel , giving (we now write for better readability) the exact sequence
Since , . We choose and a prime minimal in [Authors 2026, Tag 00E0]. By the above, , so is a unit in the localization at ; as is annihilated by a power of , . From we get , so lies in the support of in . Since support is closed under passing to larger primes [Authors 2026, Tag 00L2] and , also , i.e. .
Since and is annihilated by , we have . If , then as lies in the maximal ideal of the local ring , Nakayama's lemma [Nakayama 1951], (see also [Authors 2026, Tag 00DV]) would force , a contradiction. Hence , so and .
Therefore . ∎
Section4.lemma_4_10/-- **Lemma 4.10.** For every minimal prime `𝔮` of `Supp_T U`, `ℓ_{T_𝔮}(U_𝔮) > ℓ_{T_𝔮}(V_𝔮)`, for `d̃` in the form (3.1) with `N ≥ 1`. -/ theorem lemma_4_10 (_hN : 0 < N) (hform : IsMonicNormalForm k N d) (q : PrimeSpectrum (Tangential k n)) (hqmem : q ∈ Module.support (Tangential k n) (cokernelU k n N d)) (hqmin : ∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) > Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by -- Lemma 4.8: `V_𝔮` has finite length obtain ⟨-, -, -, hfin⟩ := lemma_4_8 hform have h := step_4_10_localized (R := Tangential k n) (symbolCoordinate k n) q (by rw [setup_coordinateMap_eq]; exact hqmem) (by rw [setup_coordinateMap_eq]; exact (hfin q hqmem hqmin).2) (step_4_10_localized_avoid hform q.asIdeal.primeCompl) rwa [setup_coordinateMap_eq] at h
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
More Lean proofs and definitions
/-- `M = gr Q` is the module `gr(A_n/𝓘)` of Theorem 2.11 for the cyclic right module `A_n/𝓘`, so Gabber's theorem applies to it: `J = √Ann M` is involutive. -/ theorem step_4_10_radical_isInvolutive : PostScalarExtensionPoisson.IsInvolutive (Module.annihilator (Symbol k n) (GrQ k n N d)).radical := (Stafford38.Paper.Section2.isInvolutive_iff k _).1 (Stafford38.Paper.Section2.theorem_2_11_cyclic k (n + 1) (Stafford38.WeylQuotientTransport.presentedCanonicalRightIdeal (k := k) n N d)) omit [Algebra ℚ k] in
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Gabber paragraph.** `J = √Ann M` is involutive (Theorem 2.11), so are its minimal primes `𝔭` [HTT D.3.4]; `P ∈ Ann M ⊆ 𝔭` (Proposition 3.6), and if `x₁ ∈ 𝔭`, the iterated brackets `{x₁, …{x₁, P}}` are the `ξ₁`-derivatives of `P` and the `N`-th is `N! ≠ 0`, so `x₁ ∉ 𝔭`. -/ theorem step_4_10_gabber (hform : IsMonicNormalForm k N d) : ∀ P ∈ (Module.annihilator (Symbol k n) (GrQ k n N d)).minimalPrimes, symbolCoordinate k n ∉ P := by intro P hP let : P.IsPrime := hP.1.1 have hPinv : PostScalarExtensionPoisson.IsInvolutive P := PaperLocalizedInvolutivity.minimalPrimes_isInvolutive_of_radical_isInvolutive_viaLocalization _ step_4_10_radical_isInvolutive P hP exact normalCoordinate_not_mem_of_isInvolutive d (isPBWMonicAt_of_isMonicNormalForm k hform) P hPinv (hP.1.2 (Stafford38.Paper.Section3.proposition_3_6 k hform (Stafford38.Paper.Section3.bernsteinDegree_eq_of_isMonicNormalForm k hform))) omit [Algebra ℚ k] in
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Localization paragraph.** With `S = T∖𝔮` and `k[x,ξ]_𝔮 = S⁻¹k[x,ξ]`, the minimal primes of `Ann(M_𝔮) = S⁻¹Ann(M)` are `S⁻¹𝔭` for minimal primes `𝔭` of `J`, so they do not contain `x₁`. -/ theorem step_4_10_localized_avoid (hform : IsMonicNormalForm k N d) (S : Submonoid (Tangential k n)) : ∀ Q ∈ (Module.annihilator (Localization (Algebra.algebraMapSubmonoid (Symbol k n) S)) (LocalizedModule (Algebra.algebraMapSubmonoid (Symbol k n) S) (GrQ k n N d))).minimalPrimes, algebraMap (Symbol k n) (Localization (Algebra.algebraMapSubmonoid (Symbol k n) S)) (symbolCoordinate k n) ∉ Q := localized_minimalPrime_avoids _ _ (step_4_10_gabber hform) omit [Algebra ℚ k] in
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Localizing at `S = T∖𝔮`.** `M_𝔮 = S⁻¹M` over `T_𝔮` is the coefficient localization `k[x,ξ]_𝔮 = S⁻¹k[x,ξ]` acting on `S⁻¹M`; there `U_𝔮 = M_𝔮/x₁M_𝔮 ≠ 0` (`𝔮 ∈ Supp_T U`) and `V_𝔮 = ker(x₁ on M_𝔮)`, so `step_4_10_core` applies. -/ theorem step_4_10_localized [IsNoetherianRing C] [Module.Finite C E] (x : C) (q : PrimeSpectrum R) (hqmem : q ∈ Module.support R (E ⧸ LinearMap.range ((LinearMap.lsmul C E x).restrictScalars R))) (hkerfinite : IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (LinearMap.ker ((LinearMap.lsmul C E x).restrictScalars R)))) (havoid : ∀ Q ∈ (Module.annihilator (Localization (Algebra.algebraMapSubmonoid C q.asIdeal.primeCompl)) (LocalizedModule (Algebra.algebraMapSubmonoid C q.asIdeal.primeCompl) E)).minimalPrimes, algebraMap C (Localization (Algebra.algebraMapSubmonoid C q.asIdeal.primeCompl)) x ∉ Q) : Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (E ⧸ LinearMap.range ((LinearMap.lsmul C E x).restrictScalars R))) > Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (LinearMap.ker ((LinearMap.lsmul C E x).restrictScalars R))) := by let S : Submonoid R := q.asIdeal.primeCompl let SC : Submonoid C := Algebra.algebraMapSubmonoid C S let fC : Module.End C E := LinearMap.lsmul C E x let fR : Module.End R E := fC.restrictScalars R -- compare the base localization `M_𝔮 = S⁻¹M` with the coefficient localization let : Module (Localization S) (LocalizedModule SC E) := Module.compHom _ (algebraMap (Localization S) (Localization SC)) let : IsScalarTower (Localization S) (Localization SC) (LocalizedModule SC E) := IsScalarTower.of_compHom (Localization S) (Localization SC) (LocalizedModule SC E) let e : LocalizedModule S E ≃ₗ[Localization S] LocalizedModule SC E := localizedModuleComparison S let gR : Module.End (Localization S) (LocalizedModule S E) := localizedMap S fR let y : Localization SC := algebraMap C (Localization SC) x let gC : Module.End (Localization SC) (LocalizedModule SC E) := LinearMap.lsmul (Localization SC) (LocalizedModule SC E) y have heq (z : LocalizedModule S E) : e (gR z) = gC (e z) := by rw [show e (gR z) = LocalizedModule.map SC fC (e z) by exact localizedModuleComparison_natural S fC z] induction e z using LocalizedModule.induction_on with | _ m s => simp [gC, y, fC, LocalizedModule.smul'_mk] let ek : LinearMap.ker gR ≃ₗ[Localization S] LinearMap.ker (gC.restrictScalars (Localization S)) := LinearEquiv.ofBijective (e.toLinearMap.domRestrict (LinearMap.ker gR) |>.codRestrict (LinearMap.ker (gC.restrictScalars (Localization S))) (fun z => by exact LinearMap.mem_ker.mpr (by simpa using (heq z).symm))) (by constructor · intro a b hab exact Subtype.ext (e.injective (congrArg Subtype.val hab)) · intro z refine ⟨⟨e.symm z, ?_⟩, Subtype.ext (e.apply_symm_apply z)⟩ exact LinearMap.mem_ker.mpr (e.injective (by rw [heq] simpa using z.property))) have herange : Submodule.map e.toLinearMap (LinearMap.range gR) = LinearMap.range (gC.restrictScalars (Localization S)) := by ext z constructor · rintro ⟨_, ⟨a, rfl⟩, rfl⟩ exact ⟨e a, by simpa using (heq a).symm⟩ · rintro ⟨a, rfl⟩ refine ⟨gR (e.symm a), ⟨e.symm a, rfl⟩, ?_⟩ simpa using heq (e.symm a) let ec : (LocalizedModule S E ⧸ LinearMap.range gR) ≃ₗ[Localization S] (LocalizedModule SC E ⧸ LinearMap.range (gC.restrictScalars (Localization S))) := Submodule.Quotient.equiv _ _ e herange -- `𝔮 ∈ Supp_T U`, so `U_𝔮 = M_𝔮/x₁M_𝔮 ≠ 0` have hnontrivBase : Nontrivial (LocalizedModule S (E ⧸ LinearMap.range fR)) := by exact Module.mem_support_iff.mp hqmem have hnontrivCokerR : Nontrivial (LocalizedModule S E ⧸ LinearMap.range gR) := not_subsingleton_iff_nontrivial.mp (by intro hs have : Subsingleton (LocalizedModule S (E ⧸ LinearMap.range fR)) := (localizedCokernelEquiv S fR).toEquiv.subsingleton_congr.mpr hs exact not_subsingleton_iff_nontrivial.mpr hnontrivBase this) have hnontrivCokerC : Nontrivial (LocalizedModule SC E ⧸ LinearMap.range (gC.restrictScalars (Localization S))) := not_subsingleton_iff_nontrivial.mp (by intro hs exact not_subsingleton_iff_nontrivial.mpr hnontrivCokerR (ec.toEquiv.subsingleton_congr.mpr hs)) have hrange : LinearMap.range gC = y • (⊤ : Submodule (Localization SC) (LocalizedModule SC E)) := by ext z rw [LinearMap.mem_range, Submodule.mem_smul_pointwise_iff_exists] simp only [Submodule.mem_top, true_and] rfl have hnontrivQuot : Nontrivial (QuotSMulTop y (LocalizedModule SC E)) := by change Nontrivial (LocalizedModule SC E ⧸ y • (⊤ : Submodule (Localization SC) (LocalizedModule SC E))) rw [← hrange] exact hnontrivCokerC have hkbase : IsFiniteLength (Localization S) (LinearMap.ker gR) := (localizedKernelEquiv S fR).isFiniteLength hkerfinite have hkfinite : IsFiniteLength (Localization S) (LinearMap.ker (gC.restrictScalars (Localization S))) := ek.isFiniteLength hkbase -- the strict inequality over `T_𝔮`, from the paragraphs above have hpos := step_4_10_core (R := Localization S) (C := Localization SC) (E := LocalizedModule SC E) y havoid hnontrivQuot hkfinite have h1 : Module.length (Localization S) (LocalizedModule S (E ⧸ LinearMap.range fR)) = Module.length (Localization S) (LocalizedModule SC E ⧸ LinearMap.range (gC.restrictScalars (Localization S))) := (localizedCokernelEquiv S fR).length_eq.trans ec.length_eq have hek : Module.length (Localization S) (LinearMap.ker gR) = Module.length (Localization S) (LinearMap.ker (gC.restrictScalars (Localization S))) := by exact ek.length_eq have h2 : Module.length (Localization S) (LocalizedModule S (LinearMap.ker fR)) = Module.length (Localization S) (LinearMap.ker (gC.restrictScalars (Localization S))) := (localizedKernelEquiv S fR).length_eq.trans hek change Module.length (Localization S) (LocalizedModule S (LinearMap.ker fR)) < Module.length (Localization S) (LocalizedModule S (E ⧸ LinearMap.range fR)) rw [h1, h2] exact hpos
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **`K_∞`.** The chain `ker f ⊆ ker f² ⊆ ⋯` stabilizes (`E` is Noetherian), at `K_∞ = ker f^m = ker f^{m+1}`; the layers `ker f^{i+1}/ker f^i` inject into `V` via `v ↦ x₁^i v`, so `K_∞` has finite length with `V`. -/ theorem step_4_10_stable_kernel [IsNoetherianRing C] [Module.Finite C E] (x : C) (hV : IsFiniteLength R (LinearMap.ker ((LinearMap.lsmul C E x).restrictScalars R))) : ∃ m : ℕ, LinearMap.ker (((LinearMap.lsmul C E x).restrictScalars R) ^ m) = LinearMap.ker (((LinearMap.lsmul C E x).restrictScalars R) ^ (m + 1)) ∧ IsFiniteLength R (LinearMap.ker (((LinearMap.lsmul C E x).restrictScalars R) ^ m)) := by let fC : Module.End C E := LinearMap.lsmul C E x let fR : Module.End R E := fC.restrictScalars R obtain ⟨m, hm⟩ := PrincipalKoszulPositivity.exists_stable_kernel_power_smul (E := E) x have hpow : ∀ j, fR ^ j = (fC ^ j).restrictScalars R := by intro j induction j with | zero => ext z; simp | succ j ih => rw [pow_succ, pow_succ, ih] ext z simp [fR, LinearMap.restrictScalars_apply, Module.End.mul_apply] refine ⟨m, ?_, PrincipalKoszulFiniteTorsion.isFiniteLength_kernel_power fR hV m⟩ change LinearMap.ker (fR ^ m) = LinearMap.ker (fR ^ (m + 1)) rw [hpow, hpow, LinearMap.ker_restrictScalars, LinearMap.ker_restrictScalars] exact congrArg (Submodule.restrictScalars R) hm
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **The two elementary facts.** For `K_∞ = ker f^m = ker f^{m+1}`: `f⁻¹K_∞ = K_∞` and `K_∞ ∩ fE = f(K_∞)`. -/ theorem step_4_10_two_facts (f : Module.End R E) (m : ℕ) (hstable : LinearMap.ker (f ^ m) = LinearMap.ker (f ^ (m + 1))) : (LinearMap.ker (f ^ m)).comap f = LinearMap.ker (f ^ m) ∧ LinearMap.ker (f ^ m) ⊓ LinearMap.range f = (LinearMap.ker (f ^ m)).map f := by have hcomap : (LinearMap.ker (f ^ m)).comap f = LinearMap.ker (f ^ m) := by ext z rw [Submodule.mem_comap] change (f ^ m) (f z) = 0 ↔ (f ^ m) z = 0 rw [← Module.End.mul_apply, (Commute.self_pow f m).eq.symm, Module.End.mul_apply] simpa [Module.End.pow_apply, Function.iterate_succ_apply'] using congrArg (fun K : Submodule R E ↦ z ∈ K) hstable.symm refine ⟨hcomap, le_antisymm ?_ ?_⟩ · rintro w ⟨hwK, v, rfl⟩ have hv : v ∈ (LinearMap.ker (f ^ m)).comap f := hwK rw [hcomap] at hv exact ⟨v, hv, rfl⟩ · rintro w ⟨v, hv, rfl⟩ refine ⟨?_, v, rfl⟩ have : v ∈ (LinearMap.ker (f ^ m)).comap f := by rw [hcomap]; exact hv exact this
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **`ℓ(K_∞/x₁K_∞) = ℓ(V)`.** For an endomorphism `φ` of a finite-length module, additivity along `0 → ker φ → L → φL → 0` and `0 → φL → L → L/φL → 0` gives `ℓ(ker φ) = ℓ(L/φL)`; for `φ = f|_{K_∞}`, `ker φ = ker f` as `ker f ⊆ K_∞`. -/ theorem step_4_10_kernel_length (f : Module.End R E) (K : Submodule R E) (hK : IsFiniteLength R K) (hcomap : K.comap f = K) : Module.length R (K ⧸ LinearMap.range (f.restrict (p := K) (q := K) (fun z hz => show z ∈ K.comap f from hcomap.symm ▸ hz))) = Module.length R (LinearMap.ker f) := by let fK : Module.End R K := f.restrict (fun z hz => show z ∈ K.comap f from hcomap.symm ▸ hz) have hparts := isFiniteLength_iff_isNoetherian_isArtinian.mp hK let : IsNoetherian R K := hparts.1 let : IsArtinian R K := hparts.2 have hkerle : LinearMap.ker f ≤ K := by intro z hz rw [← hcomap] change f z ∈ K rw [LinearMap.mem_ker.mp hz] exact K.zero_mem have hker : Module.length R (LinearMap.ker fK) = Module.length R (LinearMap.ker f) := by have hk : LinearMap.ker fK = (LinearMap.ker f).comap K.subtype := by simp [fK, LinearMap.ker_restrict] rw [hk] exact (Submodule.comapSubtypeEquivOfLe hkerle).length_eq have hsource := (definition_4_6 (T := R) (M := K)).2 (LinearMap.ker fK).subtype fK.rangeRestrict (Submodule.subtype_injective _) (LinearMap.range_eq_top.mp (LinearMap.range_rangeRestrict fK)) (by rw [LinearMap.exact_iff, Submodule.range_subtype, LinearMap.ker_rangeRestrict]) have htarget := (definition_4_6 (T := R) (M := K)).2 (LinearMap.range fK).subtype (LinearMap.range fK).mkQ (Submodule.subtype_injective _) (Submodule.mkQ_surjective _) (LinearMap.exact_subtype_mkQ _) rw [← hker] apply ENat.add_left_injective_of_ne_top (Module.length_ne_top (R := R) (M := LinearMap.range fK)) change Module.length R (K ⧸ LinearMap.range fK) + Module.length R (LinearMap.range fK) = Module.length R (LinearMap.ker fK) + Module.length R (LinearMap.range fK) rw [add_comm (Module.length R (K ⧸ LinearMap.range fK)), ← htarget] exact hsource
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **The split `ℓ(U) = ℓ(V) + ℓ(Ẽ/x₁Ẽ)`.** `E/fE → Ẽ/fẼ` is onto with kernel `(K_∞ + fE)/fE ≅ K_∞/(K_∞ ∩ fE) = K_∞/fK_∞`; additivity and `ℓ(K_∞/fK_∞) = ℓ(V)` give the split. -/ theorem step_4_10_length_split (f : Module.End R E) (K : Submodule R E) (hK : IsFiniteLength R K) (hcomap : K.comap f = K) (hinter : K ⊓ LinearMap.range f = K.map f) : Module.length R (E ⧸ LinearMap.range f) = Module.length R (LinearMap.ker f) + Module.length R (E ⧸ (K ⊔ LinearMap.range f)) := by have hstable : ∀ z ∈ K, f z ∈ K := fun z hz => show z ∈ K.comap f from hcomap.symm ▸ hz let fK : Module.End R K := f.restrict hstable have hpre : (LinearMap.range f).comap K.subtype = LinearMap.range fK := by ext z constructor · intro hz have hz' : (z : E) ∈ K.map f := hinter ▸ ⟨z.2, hz⟩ obtain ⟨w, hw, hwz⟩ := hz' exact ⟨⟨w, hw⟩, Subtype.ext hwz⟩ · rintro ⟨w, rfl⟩ exact ⟨w.val, rfl⟩ let i : (K ⧸ LinearMap.range fK) →ₗ[R] (E ⧸ LinearMap.range f) := (LinearMap.range fK).mapQ (LinearMap.range f) K.subtype hpre.ge let j : (E ⧸ LinearMap.range f) →ₗ[R] (E ⧸ (K ⊔ LinearMap.range f)) := (LinearMap.range f).mapQ (K ⊔ LinearMap.range f) LinearMap.id le_sup_right have hi : Function.Injective i := by rw [← LinearMap.ker_eq_bot] dsimp [i] rw [Submodule.ker_mapQ, hpre, Submodule.mkQ_map_self] have hj : Function.Surjective j := by rw [← LinearMap.range_eq_top] simp [j, Submodule.range_mapQ] have hexact : Function.Exact i j := by rw [LinearMap.exact_iff] simp [i, j, Submodule.range_mapQ, Submodule.ker_mapQ, Submodule.map_sup, Submodule.mkQ_map_self] have hsum := (definition_4_6 (T := R) (M := E ⧸ LinearMap.range f)).2 i j hi hj hexact rw [hsum, step_4_10_kernel_length f K hK hcomap]
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **Nakayama paragraph: `Ẽ/x₁Ẽ ≠ 0`.** Pick `𝔓` in the support of `E/x₁E ≠ 0` and `𝔭' ⊆ 𝔓` minimal in `Supp E`; `x₁ ∉ 𝔭'`, so `K_∞` vanishes at `𝔭'` and `Ẽ_{𝔭'} = E_{𝔭'} ≠ 0`; hence `Ẽ_𝔓 ≠ 0`, and as `x₁ ∈ 𝔓`, Nakayama gives `(Ẽ/x₁Ẽ)_𝔓 ≠ 0`. -/ theorem step_4_10_residual_nonzero [IsNoetherianRing C] [Module.Finite C E] (x : C) (havoid : ∀ p ∈ (Module.annihilator C E).minimalPrimes, x ∉ p) (hU : Nontrivial (QuotSMulTop x E)) (m : ℕ) : Nontrivial (E ⧸ (LinearMap.ker (((LinearMap.lsmul C E x).restrictScalars R) ^ m) ⊔ LinearMap.range ((LinearMap.lsmul C E x).restrictScalars R))) := by let fC : Module.End C E := LinearMap.lsmul C E x let fR : Module.End R E := fC.restrictScalars R -- `𝔓 ∈ Supp(E/x₁E)` and a minimal `𝔭' ⊆ 𝔓` of `Supp E` obtain ⟨P, hPquot⟩ := Module.nonempty_support_of_nontrivial (R := C) (M := QuotSMulTop x E) have hPE : P ∈ Module.support C E := by have h := hPquot rw [Module.support_quotSMulTop] at h exact h.1 obtain ⟨p, hpmin, hpP⟩ := Ideal.exists_minimalPrimes_le (I := Module.annihilator C E) (J := P.asIdeal) (Module.mem_support_iff_of_finite.mp hPE) let p' : PrimeSpectrum C := ⟨p, hpmin.1.1⟩ have hp' : p' ∈ Module.support C E := Module.mem_support_iff_of_finite.mpr hpmin.1.2 have hnonzeroC : Nontrivial (E ⧸ (LinearMap.ker (fC ^ m) ⊔ LinearMap.range fC)) := StableTorsionResidualSupport.residual_nontrivial_of_support x m p' P hp' hpP (havoid p hpmin) hPquot have hpow : ∀ j, fR ^ j = (fC ^ j).restrictScalars R := by intro j induction j with | zero => ext z; simp | succ j ih => rw [pow_succ, pow_succ, ih] ext z simp [fR, LinearMap.restrictScalars_apply, Module.End.mul_apply] have hPR : LinearMap.ker (fR ^ m) ⊔ LinearMap.range fR = (LinearMap.ker (fC ^ m) ⊔ LinearMap.range fC).restrictScalars R := by rw [hpow m] simp only [LinearMap.ker_restrictScalars, Submodule.restrictScalars_sup] rfl apply Submodule.Quotient.nontrivial_iff.mpr intro htop apply (Submodule.Quotient.nontrivial_iff.mp hnonzeroC) apply (Submodule.restrictScalars_eq_top_iff R C E).mp change LinearMap.ker (fR ^ m) ⊔ LinearMap.range fR = ⊤ at htop rw [← hPR, htop]
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- The length comparison over a base ring, assembled from the paragraphs above: `ℓ(U) = ℓ(V) + ℓ(Ẽ/x₁Ẽ) > ℓ(V)`. -/ theorem step_4_10_core [IsNoetherianRing C] [Module.Finite C E] (x : C) (havoid : ∀ p ∈ (Module.annihilator C E).minimalPrimes, x ∉ p) (hU : Nontrivial (QuotSMulTop x E)) (hV : IsFiniteLength R (LinearMap.ker ((LinearMap.lsmul C E x).restrictScalars R))) : Module.length R (E ⧸ LinearMap.range ((LinearMap.lsmul C E x).restrictScalars R)) > Module.length R (LinearMap.ker ((LinearMap.lsmul C E x).restrictScalars R)) := by let f : Module.End R E := (LinearMap.lsmul C E x).restrictScalars R obtain ⟨m, hstable, hK⟩ := step_4_10_stable_kernel x hV obtain ⟨hcomap, hinter⟩ := step_4_10_two_facts f m hstable have hsplit := step_4_10_length_split f _ hK hcomap hinter have hres := step_4_10_residual_nonzero (R := R) x havoid hU m have hpos : 0 < Module.length R (E ⧸ (LinearMap.ker (f ^ m) ⊔ LinearMap.range f)) := Module.length_pos change Module.length R (E ⧸ LinearMap.range f) > Module.length R (LinearMap.ker f) rw [hsplit] have hVtop : Module.length R (LinearMap.ker f) ≠ ⊤ := Module.length_ne_top_iff.mpr hV simpa [add_comm] using ENat.lt_add_left hVtop hpos
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {R C E : Type*} [CommRing R] [CommRing C] [Algebra R C] [AddCommGroup E] [Module C E] [Module R E] [IsScalarTower R C E]
/-- **Lemma 4.8.** `U` and `V` are finite over `T`, `Supp_T V ⊆ Supp_T U`, and at a minimal prime `𝔮` of `Supp_T U`, `U_𝔮` and `V_𝔮` have finite length over `T_𝔮`. Here `d̃` is in the form (3.1), used through `IsPBWMonicAt`. -/ theorem lemma_4_8 (hform : IsMonicNormalForm k N d) : Module.Finite (Tangential k n) (cokernelU k n N d) ∧ Module.Finite (Tangential k n) (kernelV k n N d) ∧ Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) ∧ ∀ q ∈ Module.support (Tangential k n) (cokernelU k n N d), (∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) → IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ∧ IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by obtain ⟨hU, hV⟩ := step_4_8_finite hform have hsupp : Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) := by have h := step_4_8_support_inclusion (R := Tangential k n) (LinearMap.lsmul (Symbol k n) (GrQ k n N d) (symbolCoordinate k n)) rwa [setup_coordinateMap_eq] at h exact ⟨hU, hV, hsupp, fun q hqmem hq => step_4_8_finite_length hsupp q hqmem hq⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- Theorem 2.11 for the cyclic module `M = A_n(K)/I` with `M_m = qF_m`, the case used in the proof of Theorem 1.1 (for `Q_K`): for this filtration `gr M` is `grCyclic` as a `K[x, ξ]`-module. -/ theorem theorem_2_11_cyclic (K : Type u) [Field K] [CharZero K] (n : ℕ) (I : RightIdeal (PresentedWeyl K n)) : IsInvolutive K (Module.annihilator (SymbolRing K n) (grCyclic K I)).radical := by have : Module.Finite (PresentedWeyl K n)ᵐᵒᵖ (RightQuotient I) := Stafford38.Paper.Support.Gabber.rightQuotient_finite I rw [← Stafford38.Paper.Support.Gabber.annihilator_qFilt] exact theorem_2_11 K n (Stafford38.Paper.Support.Gabber.qFilt K I) (qFilt_isGoodFiltration K I)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- Localization proof of the componentwise statement: involutivity of `√J` implies involutivity of each minimal prime over `J`. Unlike the saturation proof in `RadicalMinimalPrimeInvolutivity`, this follows the paper's argument through the localized Poisson structure, its maximal ideal, and contraction by primeness. -/ theorem minimalPrimes_isInvolutive_of_radical_isInvolutive_viaLocalization (J : Ideal (R (k := k) (n := n))) (hJ : IsInvolutive J.radical) : ∀ P ∈ J.minimalPrimes, IsInvolutive P := by intro P hP letI : P.IsPrime := hP.1.1 intro f hf g hg let φ := algebraMap (R (k := k) (n := n)) (AtPrime P) have hrad := radical_map_eq_prime_map_of_minimal J P hP have hfLoc : φ f ∈ P.map φ := Ideal.mem_map_of_mem φ hf have hgLoc : φ g ∈ P.map φ := Ideal.mem_map_of_mem φ hg rw [← hrad] at hfLoc hgLoc have hbrLoc := localizedMap_isInvolutive P J.radical hJ (φ f) hfLoc (φ g) hgLoc rw [localizedPoissonBracket_algebraMap] at hbrLoc rw [hrad] at hbrLoc obtain ⟨s, hs, hmul⟩ := (IsLocalization.algebraMap_mem_map_algebraMap_iff P.primeCompl (AtPrime P) P (poissonBracket f g)).mp hbrLoc have hprime : P.IsPrime := hP.1.1 rcases hprime.mem_or_mem hmul with hsP | hbr · exact (hs hsP).elim · exact hbr
/-- The bracket argument: an involutive prime `P` containing the order symbol of a `ξ₁`-monic `d` of degree `N` omits `x₁`, since otherwise the `N`-th bracket `{x₁, …{x₁, σ(d)}}` is the constant `N! ≠ 0` and lies in `P`. -/ theorem normalCoordinate_not_mem_of_isInvolutive (d : PresentedWeyl k (n + 1)) (hd : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N d) (P : Ideal (SymbolRing k (n + 1))) [P.IsPrime] (hInv : IsInvolutive P) (hf : presentedPrincipalComponent k orderWeight N d ∈ P) : MvPolynomial.X (.inl (0 : Fin (n + 1))) ∉ P := by intro hx let f := presentedPrincipalComponent k orderWeight N d have hderiv := iterPderiv_mem_of_involutive P hInv (0 : Fin (n + 1)) hx f hf N have hfac : MvPolynomial.C (N.factorial : k) ∈ P := by rw [← iterPderiv_eq_factorial_of_homogeneous (.inr (0 : Fin (n + 1))) f (canonical_orderPrincipalComponent_isHomogeneous n N hd) (canonical_orderPrincipalComponent_pureMomentumCoefficient k n N hd)] exact hderiv have hfac0 : (N.factorial : k) ≠ 0 := by exact_mod_cast (Nat.factorial_ne_zero N) have hone : (1 : SymbolRing k (n + 1)) ∈ P := by have hm := P.mul_mem_left (MvPolynomial.C ((N.factorial : k)⁻¹)) hfac rw [← MvPolynomial.C_mul, inv_mul_cancel₀ hfac0] at hm simpa using hm exact (inferInstance : P.IsPrime).ne_top ((Ideal.eq_top_iff_one P).mpr hone)
Variables and notation
Source ↗open Stafford38Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.Characteristic.CanonicalNormalAxisSupportSource ↗open Stafford38.Characteristic.GabberGlobalAssemblySource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗variable {k : Type*} [Field k] [CharZero k] {n N : ℕ}
/-- **Proposition 3.6.** Equip `Q` with the order filtration `Q_m = qF_m` induced by the cyclic generator `q = [1]`, and write `P = σ_N^ord(d̃)`. Then `P ∈ Ann_{k[x,ξ]} gr Q` (eq. (3.5)). -/ theorem proposition_3_6 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : orderSymbol k N dt ∈ Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt) := by obtain ⟨⟨hF, -⟩, -⟩ := lemma_3_5 k hform hdeg rw [Module.mem_annihilator] intro m obtain ⟨f, rfl⟩ := (step_3_6_generated k dt).2.2 m rw [smul_comm, step_3_6_kills k hF, smul_zero]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- In (3.1), `N = deg_B d̃`: `d̃ ∈ B_N` and its Bernstein symbol has the coefficient `1` at `ξ₁^N`. -/ theorem bernsteinDegree_eq_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : bernsteinDegree k dt = N := by have hd := isPBWMonicAt_of_isMonicNormalForm k h refine bernsteinDegree_eq_of_principal_ne_zero k hd.1 fun h0 => ?_ have hc : AddMonoidAlgebra.coeff (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) N dt) (Finsupp.single (.inr (0 : Fin (n + 1))) N) = 1 := by rw [coeff_presentedPrincipalComponent, ite_eq_left (by simp [monomialWeight, bernsteinWeight])] exact hd.2 rw [h0] at hc exact zero_ne_one hc
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- An element of the form (3.1) is PBW-monic: it lies in `B_N` and the PBW coefficient of `∂₁^N` is `1`. This connects (3.1) to the repository's `WeylPBWMonicBridge.IsPBWMonicAt`, the hypothesis of `UniversalAssembly.CanonicalSupportVanishing`. -/ theorem isPBWMonicAt_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N dt := by obtain ⟨a, ha, rfl⟩ := h have hterm (j : ℕ) (hj : j ∈ Finset.range N) := principal_bernstein_p_pow_mul k (Finset.mem_range.mp hj).le ((bernsteinDegree_le_iff k (a j)).mp (ha j (Finset.mem_range.mp hj)).2) obtain ⟨hpN, hσN⟩ := p_pow_piece_principal k (@bernsteinWeight (n + 1)) rfl N refine ⟨add_mem hpN (Submodule.sum_mem _ fun j hj => (hterm j hj).1), ?_⟩ rw [← coeff_principal_p_pow k (@bernsteinWeight (n + 1)) rfl, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj rw [(hterm j hj).2] have h := coeff_X_pow_mul_single k hjN.le (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) (N - j) (a j)) simp only [MvPolynomial.coeff] at h rw [h] have h2 := coeff_principal_p_pow k (L := N - j) (@bernsteinWeight (n + 1)) rfl (a j) simp only [MvPolynomial.coeff] at h2 rw [h2] by_contra hne have := normalForm_C_no_p k (ha j hjN).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.10 (involutive ideal).** `{J, J} ⊆ J`. -/ def IsInvolutive {n : ℕ} (J : Ideal (SymbolRing k n)) : Prop := ∀ f ∈ J, ∀ g ∈ J, poisson k f g ∈ J
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
lemma_4_10 states the lemma as printed: ℓ_{T_𝔮}(U_𝔮) > ℓ_{T_𝔮}(V_𝔮) for every minimal prime 𝔮 of Supp_T U, for d̃ in the form (3.1) (Section3.IsMonicNormalForm) with N ≥ 1. The proof uses (3.1) through Proposition 3.6 (P ∈ Ann M; deg_B d̃ = N by Section3.bernsteinDegree_eq_of_isMonicNormalForm) and through its consequence that d̃ ∈ B_N has coefficient 1 at ∂₁^N (Section3.isPBWMonicAt_of_isMonicNormalForm). The proof follows the paragraphs: Gabber (step_4_10_radical_isInvolutive: J = √Ann M is involutive by Theorem 2.10 for the cyclic module Q, Section2.theorem_2_11_cyclic; step_4_10_gabber: every minimal prime 𝔭 of J is involutive [HTT D.3.4], P ∈ 𝔭, and the bracket argument gives x₁ ∉ 𝔭), localization (step_4_10_localized_avoid, step_4_10_localized), K_∞ (step_4_10_stable_kernel), the two elementary facts, ℓ(K_∞/x₁K_∞) = ℓ(V), the split ℓ(U) = ℓ(V) + ℓ(Ẽ/x₁Ẽ) and the Nakayama paragraph Ẽ/x₁Ẽ ≠ 0, assembled in step_4_10_core. step_4_10_gabber takes the involutivity of minimal primes from PaperLocalizedInvolutivity.minimalPrimes_isInvolutive_of_radical_isInvolutive_viaLocalization and the bracket argument from NoncharacteristicMinimalPrime.normalCoordinate_not_mem_of_isInvolutive.
Review notes
Corollary 4.11 — gr Q = (gr Q)x₁
Section4.corollary_4_11/-- **Corollary 4.11.** Multiplication by `x₁` on `gr Q` is surjective, i.e. `gr Q = (gr Q) x₁`, for `d̃` in the form (3.1) with `N ≥ 1`. -/ theorem corollary_4_11 (hN : 0 < N) (hform : IsMonicNormalForm k N d) : Function.Surjective (fun z : GrQ k n N d => symbolCoordinate k n • z) := by have hU := step_4_11_cokernel_zero hN hform have htop : LinearMap.range (coordinateMap k n N d) = ⊤ := Submodule.Quotient.subsingleton_iff.mp hU intro z obtain ⟨w, hw⟩ := LinearMap.range_eq_top.mp htop z exact ⟨w, (setup_coordinateMap_apply k n N d w).symm.trans hw⟩
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
More Lean proofs and definitions
/-- **Proof of Corollary 4.11.** If `U ≠ 0` it has a minimal support prime `𝔮`; Lemma 4.8 makes the lengths finite, and Lemmas 4.9 and 4.10 contradict one another. Hence `U = M/x₁M = 0`. -/ theorem step_4_11_cokernel_zero (hN : 0 < N) (hform : IsMonicNormalForm k N d) : Subsingleton (cokernelU k n N d) := by obtain ⟨hU, -, -, -⟩ := lemma_4_8 hform by_contra h have : Nontrivial (cokernelU k n N d) := not_subsingleton_iff_nontrivial.mp h obtain ⟨q, hqmem, hqmin⟩ := AlgebraicAnalysis.MinimalSupportExistence.exists_minimal_support_prime (R := Tangential k n) (U := cokernelU k n N d) exact not_lt_of_ge (lemma_4_9 hform q hqmem hqmin) (lemma_4_10 hN hform q hqmem hqmin)
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- Corollary 4.11 in the form `M/x₁M = 0` over `k[x,ξ]` (`QuotSMulTop`), as consumed by the support-avoidance argument of Section 6. -/ theorem corollary_4_11_quotSMulTop (hN : 0 < N) (hform : IsMonicNormalForm k N d) : Subsingleton (QuotSMulTop (symbolCoordinate k n) (GrQ k n N d)) := by have hsurj := corollary_4_11 hN hform change Subsingleton (GrQ k n N d ⧸ symbolCoordinate k n • (⊤ : Submodule (Symbol k n) _)) rw [Submodule.Quotient.subsingleton_iff, eq_top_iff] intro z _ obtain ⟨w, rfl⟩ := hsurj z exact Submodule.smul_mem_pointwise_smul w _ ⊤ trivial
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Lemma 4.8.** `U` and `V` are finite over `T`, `Supp_T V ⊆ Supp_T U`, and at a minimal prime `𝔮` of `Supp_T U`, `U_𝔮` and `V_𝔮` have finite length over `T_𝔮`. Here `d̃` is in the form (3.1), used through `IsPBWMonicAt`. -/ theorem lemma_4_8 (hform : IsMonicNormalForm k N d) : Module.Finite (Tangential k n) (cokernelU k n N d) ∧ Module.Finite (Tangential k n) (kernelV k n N d) ∧ Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) ∧ ∀ q ∈ Module.support (Tangential k n) (cokernelU k n N d), (∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) → IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ∧ IsFiniteLength (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by obtain ⟨hU, hV⟩ := step_4_8_finite hform have hsupp : Module.support (Tangential k n) (kernelV k n N d) ⊆ Module.support (Tangential k n) (cokernelU k n N d) := by have h := step_4_8_support_inclusion (R := Tangential k n) (LinearMap.lsmul (Symbol k n) (GrQ k n N d) (symbolCoordinate k n)) rwa [setup_coordinateMap_eq] at h exact ⟨hU, hV, hsupp, fun q hqmem hq => step_4_8_finite_length hsupp q hqmem hq⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.CanonicalOldTangentialFinitenessSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Lemma 4.9.** For every minimal prime `𝔮` of `Supp_T U`, `ℓ_{T_𝔮}(U_𝔮) ≤ ℓ_{T_𝔮}(V_𝔮)`, for `d̃` in the form (3.1). -/ theorem lemma_4_9 (hform : IsMonicNormalForm k N d) (q : PrimeSpectrum (Tangential k n)) (hqmem : q ∈ Module.support (Tangential k n) (cokernelU k n N d)) (hqmin : ∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) ≤ Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by have hd := isPBWMonicAt_of_isMonicNormalForm k hform let S := q.asIdeal.primeCompl let A : ℕ → Type _ := fun r => LocalizedModule S ((complex k n N d hd).SourceTotal r) let B : ℕ → Type _ := fun r => LocalizedModule S ((complex k n N d hd).TargetTotal r) let : ∀ r, AddCommGroup (A r) := fun r => by dsimp only [A]; infer_instance let : ∀ r, Module (Localization S) (A r) := fun r => by dsimp only [A]; infer_instance let : ∀ r, AddCommGroup (B r) := fun r => by dsimp only [B]; infer_instance let : ∀ r, Module (Localization S) (B r) := fun r => by dsimp only [B]; infer_instance -- the pages and their localization; exactness keeps the successor identities obtain ⟨hsucc, htsucc, ⟨eV⟩, ⟨eU⟩⟩ := step_4_9_pages hd let dd : ∀ r, A r →ₗ[Localization S] B r := fun r => localizedMap S (tangentialDrop k n N d hd r) let es : ∀ r, A (r + 1) ≃ₗ[Localization S] LinearMap.ker (dd r) := fun r => (localizedEquiv S (hsucc r).some).trans (localizedKernelEquiv S _) let et : ∀ r, B (r + 1) ≃ₗ[Localization S] B r ⧸ LinearMap.range (dd r) := fun r => (localizedEquiv S (htsucc r).some).trans (localizedCokernelEquiv S _) let eV' : A 1 ≃ₗ[Localization S] LocalizedModule S (kernelV k n N d) := localizedEquiv S eV let eU' : B 1 ≃ₗ[Localization S] LocalizedModule S (cokernelU k n N d) := localizedEquiv S eU -- Lemma 4.8: `(A_1)_𝔮 = V_𝔮` and `(B_1)_𝔮 = U_𝔮` have finite length obtain ⟨-, -, -, hfin⟩ := lemma_4_8 hform obtain ⟨hU, hV⟩ := hfin q hqmem hqmin have hfl := step_4_9_pages_finite_length dd es et (eV'.symm.isFiniteLength hV) (eU'.symm.isFiniteLength hU) -- exhaustion: `(B_R)_𝔮 = 0` let : IsNoetherian (Localization S) (B 1) := (isFiniteLength_iff_isNoetherian_isArtinian.mp (hfl 1 le_rfl).2).1 obtain ⟨R', hR'⟩ := step_4_9_exhaustion hform S -- equation (additivity) and the telescoping sum have hlen := step_4_9_telescope (fun r => if 1 ≤ r then Module.length (Localization S) (A r) else 0) (fun r => if 1 ≤ r then Module.length (Localization S) (B r) else 0) (fun r => by split_ifs with h · exact Module.length_ne_top_iff.mpr (hfl r h).1 · exact ENat.zero_ne_top) (fun r => by split_ifs with h · exact Module.length_ne_top_iff.mpr (hfl r h).2 · exact ENat.zero_ne_top) (fun r hr => by simp only [hr, ite_true, show 1 ≤ r + 1 by omega] exact step_4_9_additivity (dd r) (es r) (et r)) (R' + 1) (by omega) simp only [le_refl, ite_true, show 1 ≤ R' + 1 by omega] at hlen have hBR : Module.length (Localization S) (B (R' + 1)) = 0 := Module.length_eq_zero_iff.mpr hR' -- `ℓ(U_𝔮) + ℓ((A_R)_𝔮) = ℓ(V_𝔮)`, and `ℓ((A_R)_𝔮) ≥ 0` calc Module.length (Localization S) (LocalizedModule S (cokernelU k n N d)) = Module.length (Localization S) (B 1) := eU'.length_eq.symm _ ≤ Module.length (Localization S) (B 1) + Module.length (Localization S) (A (R' + 1)) := le_self_add _ = Module.length (Localization S) (A 1) := by rw [hlen, hBR, add_zero] _ = Module.length (Localization S) (LocalizedModule S (kernelV k n N d)) := eV'.length_eq
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.FilteredTwoTermPagesSource ↗open Stafford38.Characteristic.CanonicalFilteredTwoTermSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.Characteristic.CanonicalTangentialTotalActionSource ↗open Stafford38.Characteristic.CanonicalGradedTangentialEquivalencesSource ↗open Stafford38.Characteristic.LocalizedKernelCokernelEquivalencesSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylQuotientTransportSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
universe vSource ↗
variable {k : Type*} [Field k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- **Lemma 4.10.** For every minimal prime `𝔮` of `Supp_T U`, `ℓ_{T_𝔮}(U_𝔮) > ℓ_{T_𝔮}(V_𝔮)`, for `d̃` in the form (3.1) with `N ≥ 1`. -/ theorem lemma_4_10 (_hN : 0 < N) (hform : IsMonicNormalForm k N d) (q : PrimeSpectrum (Tangential k n)) (hqmem : q ∈ Module.support (Tangential k n) (cokernelU k n N d)) (hqmin : ∀ p ∈ Module.support (Tangential k n) (cokernelU k n N d), p.asIdeal ≤ q.asIdeal → q.asIdeal ≤ p.asIdeal) : Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (cokernelU k n N d)) > Module.length (Localization q.asIdeal.primeCompl) (LocalizedModule q.asIdeal.primeCompl (kernelV k n N d)) := by -- Lemma 4.8: `V_𝔮` has finite length obtain ⟨-, -, -, hfin⟩ := lemma_4_8 hform have h := step_4_10_localized (R := Tangential k n) (symbolCoordinate k n) q (by rw [setup_coordinateMap_eq]; exact hqmem) (by rw [setup_coordinateMap_eq]; exact (hfin q hqmem hqmin).2) (step_4_10_localized_avoid hform q.asIdeal.primeCompl) rwa [setup_coordinateMap_eq] at h
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.NoncharacteristicMinimalPrimeSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.BaseLocalizationModuleComparisonSource ↗open AlgebraicAnalysis.LocalizedKernelCokernelEquivalencesSource ↗open AlgebraicAnalysis.LocalizedMinimalSupportAvoidanceSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- An element of the form (3.1) is PBW-monic: it lies in `B_N` and the PBW coefficient of `∂₁^N` is `1`. This connects (3.1) to the repository's `WeylPBWMonicBridge.IsPBWMonicAt`, the hypothesis of `UniversalAssembly.CanonicalSupportVanishing`. -/ theorem isPBWMonicAt_of_isMonicNormalForm {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : IsMonicNormalForm k N dt) : IsPBWMonicAt k (.inr (0 : Fin (n + 1))) N dt := by obtain ⟨a, ha, rfl⟩ := h have hterm (j : ℕ) (hj : j ∈ Finset.range N) := principal_bernstein_p_pow_mul k (Finset.mem_range.mp hj).le ((bernsteinDegree_le_iff k (a j)).mp (ha j (Finset.mem_range.mp hj)).2) obtain ⟨hpN, hσN⟩ := p_pow_piece_principal k (@bernsteinWeight (n + 1)) rfl N refine ⟨add_mem hpN (Submodule.sum_mem _ fun j hj => (hterm j hj).1), ?_⟩ rw [← coeff_principal_p_pow k (@bernsteinWeight (n + 1)) rfl, map_add, map_sum, hσN] simp only [MvPolynomial.coeff] rw [AddMonoidAlgebra.coeff_add, Finsupp.add_apply, MvPolynomial.coeff_sum, Finset.sum_eq_zero, add_zero, MvPolynomial.X_pow_eq_monomial, MvPolynomial.coeff_monomial, ite_eq_left rfl] intro j hj have hjN := Finset.mem_range.mp hj rw [(hterm j hj).2] have h := coeff_X_pow_mul_single k hjN.le (presentedPrincipalComponent k (@bernsteinWeight (n + 1)) (N - j) (a j)) simp only [MvPolynomial.coeff] at h rw [h] have h2 := coeff_principal_p_pow k (L := N - j) (@bernsteinWeight (n + 1)) rfl (a j) simp only [MvPolynomial.coeff] at h2 rw [h2] by_contra hne have := normalForm_C_no_p k (ha j hjN).1 _ hne simp at this omega
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylOuterOreMonicSource ↗open Stafford38.WeylPaperRightMonicSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
Comparison notes
corollary_4_11 states the corollary as printed: multiplication by x₁ on gr Q is surjective, for d̃ in the form (3.1) (Section3.IsMonicNormalForm) with N ≥ 1. The proof uses (3.1) through its consequence that d̃ ∈ B_N has coefficient 1 at ∂₁^N (Section3.isPBWMonicAt_of_isMonicNormalForm). The proof (step_4_11_cokernel_zero) is the printed one: if U ≠ 0 it has a minimal support prime, where Lemma 4.8 makes the lengths finite and Lemmas 4.9 and 4.10 contradict each other. corollary_4_11_quotSMulTop is the form M/x₁M = 0, which Proposition 6.1 calls with (3.1).
Review notes
5. Avoiding the hyperplane {x₁ = 0}
Paragraph — Section 5 overview
Avoiding the hyperplane
This section uses mainly vocabulary and concepts from algebraic geometry. We will define the most important quantities used here (for a comprehensive overview of the concepts of algebraic geometry, we refer to textbooks such as [hartshorne]), then state the two main results: Theorem 5.7 shows that for a closed irreducible subvariety not meeting the hyperplane , the covector is an asymptotic conormal direction of . Theorem 5.8 uses it to show that no nonempty closed, fibre-conical, involutive subset of the cotangent space can lie over the complement of the hyperplane and be contained in the zero set of a polynomial with . We prove the first theorem in Subsection 5.2, with the lemmas stated and proved where they are used, and the second in Subsection 5.3.No corresponding declaration.
Comparison notes
Review notes
5.1. Preliminaries and definitions from algebraic geometry
Definition 5.1 — Algebraic and transcendental elements; transcendence bases
- is algebraic over if there exists a nonzero polynomial in one variable over with
- is transcendental over if no such polynomial exists, i.e.\
- The minimal polynomial of an algebraic over is the polynomial of least degree over with and coefficient 1 for the highest degree term.
- is algebraic if every element of is algebraic over
- is finitely generated if for finitely many , and separable if every that is algebraic over has a minimal polynomial without repeated roots - in characteristic zero every algebraic extension is separable.
- is algebraically independent over if no nonzero polynomial over in finitely many elements of vanishes.
- A maximal algebraically independent subset is a transcendence basis of . If is a transcendence basis, then is algebraic over the subfield generated by and .
- Any two transcendence bases of have the same cardinality, called the transcendence degree .
/-- **Definition 5.1(g)–(h).** For a transcendence basis `B` of `L/K`, `L` is algebraic over `K(B)`, and `|B| = trdeg_K L`. -/ theorem definition_5_1_g_h {B : Set L} (hB : IsTranscendenceBasis K ((↑) : B → L)) : Algebra.IsAlgebraic (IntermediateField.adjoin K (Set.range ((↑) : B → L))) L ∧ Cardinal.mk B = Algebra.trdeg K L := ⟨hB.isAlgebraic_field, hB.cardinalMk_eq_trdeg⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
/-- **Definition 5.1, closing sentence.** The elements of `L` that are algebraic over `K` form a subfield, and every finite set generating `L` over `K` contains a transcendence basis ([Lang, Thm. VIII.1.1]). -/ theorem definition_5_1_closing : (∃ E : IntermediateField K L, ∀ z : L, z ∈ E ↔ IsAlgebraic K z) ∧ ∀ S : Set L, S.Finite → IntermediateField.adjoin K S = ⊤ → ∃ B ⊆ S, IsTranscendenceBasis K ((↑) : B → L) := by refine ⟨⟨algebraicClosure K L, fun z => mem_algebraicClosure_iff⟩, fun S _ hS => ?_⟩ have : Algebra.IsAlgebraic (Algebra.adjoin K S) L := by rw [← IntermediateField.isAlgebraic_adjoin_iff_top, hS] exact ⟨fun z => ⟨Polynomial.X - Polynomial.C ⟨z, trivial⟩, Polynomial.X_sub_C_ne_zero _, by simp [Polynomial.aeval_def]⟩⟩ exact exists_isTranscendenceBasis_subset S
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
Comparison notes
(a)–(f) are Mathlib's notions with the printed content: IsAlgebraic, Transcendental, minpoly, Algebra.IsAlgebraic, finitely generated extensions, separability (that it is automatic in characteristic zero is Mathlib's and is not restated) and AlgebraicIndependent, which Mathlib phrases as injectivity of evaluation on polynomials in the elements of B. (g)–(h) are definition_5_1_g_h: for a transcendence basis B (IsTranscendenceBasis, a maximal algebraically independent set), L is algebraic over K(B) and |B| = trdeg_K L. The closing sentence is definition_5_1_closing: the elements algebraic over K form a subfield, and every finite set generating L over K contains a transcendence basis [Lang, Thm. VIII.1.1]. The paper cites these facts; Lean proves them from Mathlib. Lemma 5.11 uses the second half of the closing sentence and (g).
Review notes
Definition 5.2 — Derivations and Kähler differentials of a field extension
- A -derivation of is a -linear map with .
- Let be the free -vector space with basis symbols
, . The space of differentials
is the quotient of by the subspace spanned by
and denotes the class of .
- A -derivation induces the -linear map , , since the linear map on kills the relations. Conversely, a -linear map gives the -derivation . Hence a basis of has dual derivations with .
/-- **Definition 5.2(b).** `Ω_{κ/K}`, the quotient of the free module `𝒢` by the printed relations, is Mathlib's `Ω[κ⁄K]`, with the class of `⟨a⟩` going to `da`. -/ theorem definition_5_2_b : ∃ e : ((L →₀ L) ⧸ differentialRelations K L) ≃ₗ[L] Ω[L⁄K], ∀ a : L, e (Submodule.Quotient.mk (Finsupp.single a 1)) = KaehlerDifferential.D K L a := by refine ⟨(Submodule.quotEquivOfEq _ _ differentialRelations_eq_kerTotal).trans (KaehlerDifferential.quotKerTotalEquiv K L), fun a ↦ ?_⟩ simp [KaehlerDifferential.quotKerTotalEquiv]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
/-- **Definition 5.2(c), first sentences.** A `K`-derivation `D` of `κ` induces the `κ`-linear map `Ω_{κ/K} → κ`, `da ↦ D(a)`, and conversely a `κ`-linear map `φ : Ω_{κ/K} → κ` gives the `K`-derivation `a ↦ φ(da)`; the two constructions are inverse (`KaehlerDifferential.linearMapEquivDerivation`). -/ theorem definition_5_2_c_correspondence : ∃ e : (Ω[L⁄K] →ₗ[L] L) ≃ₗ[L] Derivation K L L, (∀ (φ : Ω[L⁄K] →ₗ[L] L) (a : L), e φ a = φ (KaehlerDifferential.D K L a)) ∧ ∀ (D : Derivation K L L) (a : L), e.symm D (KaehlerDifferential.D K L a) = D a := ⟨KaehlerDifferential.linearMapEquivDerivation K L, fun _ _ => rfl, fun D a => D.liftKaehlerDifferential_comp_D a⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
More Lean proofs and definitions
/-- **Definition 5.2(c).** A basis `dβ_1, …, dβ_d` of `Ω_{κ/K}` has dual derivations `D_k` with `D_k(β_l) = δ_{kl}`, namely `D_k = (dβ_k)^* ∘ d`. -/ theorem definition_5_2_c {ι : Type*} [DecidableEq ι] (β : ι → L) (b : Module.Basis ι L Ω[L⁄K]) (hb : ∀ l, b l = KaehlerDifferential.D K L (β l)) : ∃ D : ι → Derivation K L L, ∀ k' l, D k' (β l) = if k' = l then 1 else 0 := by -- `dβ_l` is the class of `⟨β_l⟩` (Definition 5.2(b)); `D_k` is the derivation -- of the coordinate functional `(dβ_k)^*` on `Ω_{κ/K}`. obtain ⟨e, he⟩ := definition_5_2_b (K := K) (L := L) obtain ⟨c, hc, -⟩ := definition_5_2_c_correspondence (K := K) (L := L) refine ⟨fun k' ↦ c (b.coord k'), fun k' l ↦ ?_⟩ have hb' : ∀ l, b l = e (Submodule.Quotient.mk (Finsupp.single (β l) 1)) := fun l ↦ (hb l).trans (he _).symm rw [hc, ← he, ← hb', Module.Basis.coord_apply, Module.Basis.repr_self, Finsupp.single_apply] by_cases h : k' = l · subst h; simp · simp [h, Ne.symm h]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
/-- The relations of Definition 5.2(b) in the free `κ`-vector space `𝒢` on symbols `⟨a⟩ = Finsupp.single a 1`: `⟨a+b⟩ - ⟨a⟩ - ⟨b⟩`, `⟨ab⟩ - a⟨b⟩ - b⟨a⟩`, `⟨c⟩` (`c ∈ K`). -/ def differentialRelations (K κ : Type*) [Field K] [Field κ] [Algebra K κ] : Submodule κ (κ →₀ κ) := Submodule.span κ (((Set.range fun x : κ × κ ↦ Finsupp.single (x.1 + x.2) 1 - Finsupp.single x.1 1 - Finsupp.single x.2 1) ∪ Set.range fun x : κ × κ ↦ Finsupp.single (x.1 * x.2) 1 - x.1 • Finsupp.single x.2 1 - x.2 • Finsupp.single x.1 1) ∪ Set.range fun c : K ↦ Finsupp.single (algebraMap K κ c) 1)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
/-- The printed relations span Mathlib's `KaehlerDifferential.kerTotal`. -/ theorem differentialRelations_eq_kerTotal : differentialRelations K L = KaehlerDifferential.kerTotal K L := by apply le_antisymm · rw [differentialRelations, Submodule.span_le] rintro _ ((⟨⟨a, b⟩, rfl⟩ | ⟨⟨a, b⟩, rfl⟩) | ⟨c, rfl⟩) · have e : Finsupp.single (a + b) (1 : L) - Finsupp.single a 1 - Finsupp.single b 1 = -(Finsupp.single a 1 + Finsupp.single b 1 - Finsupp.single (a + b) 1) := by abel rw [SetLike.mem_coe] beta_reduce rw [e] exact Submodule.neg_mem _ (Submodule.subset_span (Or.inl (Or.inl ⟨(a, b), rfl⟩))) · have e : Finsupp.single (a * b) (1 : L) - a • Finsupp.single b 1 - b • Finsupp.single a 1 = -(Finsupp.single b a + Finsupp.single a b - Finsupp.single (a * b) 1) := by simp only [Finsupp.smul_single, smul_eq_mul, mul_one] abel rw [SetLike.mem_coe] beta_reduce rw [e] exact Submodule.neg_mem _ (Submodule.subset_span (Or.inl (Or.inr ⟨(a, b), rfl⟩))) · exact Submodule.subset_span (Or.inr ⟨c, rfl⟩) · rw [KaehlerDifferential.kerTotal, Submodule.span_le] rintro _ ((⟨⟨a, b⟩, rfl⟩ | ⟨⟨a, b⟩, rfl⟩) | ⟨c, rfl⟩) · have e : Finsupp.single a (1 : L) + Finsupp.single b 1 - Finsupp.single (a + b) 1 = -(Finsupp.single (a + b) 1 - Finsupp.single a 1 - Finsupp.single b 1) := by abel rw [SetLike.mem_coe] beta_reduce rw [e] exact Submodule.neg_mem _ (Submodule.subset_span (Or.inl (Or.inl ⟨(a, b), rfl⟩))) · have e : Finsupp.single b a + Finsupp.single a b - Finsupp.single (a * b) (1 : L) = -(Finsupp.single (a * b) 1 - a • Finsupp.single b 1 - b • Finsupp.single a 1) := by simp only [Finsupp.smul_single, smul_eq_mul, mul_one] abel rw [SetLike.mem_coe] beta_reduce rw [e] exact Submodule.neg_mem _ (Submodule.subset_span (Or.inl (Or.inr ⟨(a, b), rfl⟩))) · exact Submodule.subset_span (Or.inr ⟨c, rfl⟩)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
Comparison notes
(a) is Mathlib's Derivation. (b): definition_5_2_b identifies the printed quotient of the free module 𝒢 by the relations ⟨a+b⟩ − ⟨a⟩ − ⟨b⟩, ⟨ab⟩ − a⟨b⟩ − b⟨a⟩, ⟨c⟩ (differentialRelations) with Mathlib's Ω[κ⁄K], ⟨a⟩ ↦ da. (c): definition_5_2_c_correspondence states the first two sentences: D ↦ (da ↦ D(a)) and φ ↦ (a ↦ φ(da)) are mutually inverse κ-linear bijections between K-derivations of κ and κ-linear maps Ω_{κ/K} → κ. Lean takes this correspondence from Mathlib (KaehlerDifferential.linearMapEquivDerivation) instead of checking that ⟨a⟩ ↦ D(a) kills the relations on 𝒢. definition_5_2_c derives from it the dual derivations D_k(β_l) = δ_kl of a basis dβ_l, which are used after Lemma 5.11.
Review notes
Definition 5.3 — Valuation rings, discrete valuation rings and Laurent series
- A subring is a valuation ring of if for every we have or . It is a local ring with maximal ideal .
- A discrete valuation ring (DVR) is a Noetherian valuation ring . Then is generated by some , and every is with and , and .
- For a field we write for the ring of formal power series, and for its fraction field (formal Laurent series).
/-- **Definition 5.3.** Let `F` be a field. (a) A subring `W ⊆ F` is a valuation ring of `F` iff `a ∈ W` or `a⁻¹ ∈ W` for every `a ∈ F^×`; it is a local ring with maximal ideal `𝔪_W = W ∖ W^×`. (b) A valuation ring `V` is a discrete valuation ring iff it is Noetherian and `V ≠ F`; then `𝔪_V` is generated by some `ϖ ∈ V`, and every `f ∈ F^×` is `f = ϖ^k u` with `u ∈ V^×` and a unique `k = ord_V(f) ∈ ℤ`. (c) For a field `κ`, `κ((t))` (`LaurentSeries κ`) is the fraction field of `κ[[t]]` (`PowerSeries κ`). -/ theorem definition_5_3 {F : Type u} [Field F] : (∀ S : Subring F, (∀ a : F, a ≠ 0 → a ∈ S ∨ a⁻¹ ∈ S) ↔ ∃ W : ValuationSubring F, W.toSubring = S) ∧ (∀ W : ValuationSubring F, IsLocalRing W.toSubring ∧ ∀ w : W.toSubring, w ∈ maximalIdeal W.toSubring ↔ ¬ IsUnit w) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring ↔ IsNoetherianRing V.toSubring ∧ V ≠ ⊤) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring → ∃ ϖ : V.toSubring, maximalIdeal V.toSubring = Ideal.span {ϖ} ∧ ∀ f : F, f ≠ 0 → ∃! k : ℤ, ∃ u : V.toSubringˣ, f = (ϖ : F) ^ k * ((u : V.toSubring) : F)) ∧ ∀ (κ : Type u) [Field κ], IsFractionRing (PowerSeries κ) (LaurentSeries κ) := by refine ⟨fun S => ⟨fun h => ?_, ?_⟩, fun W => ⟨inferInstance, fun w => ?_⟩, fun V => ⟨fun hV => ?_, fun ⟨hN, hV⟩ => ?_⟩, fun V hV => ?_, fun κ _ => inferInstance⟩ · refine ⟨⟨S, fun a => ?_⟩, rfl⟩ by_cases ha : a = 0 · exact Or.inl (ha ▸ S.zero_mem) · exact h a ha · rintro ⟨W, rfl⟩ a _ exact W.mem_or_inv_mem a · rw [mem_maximalIdeal, mem_nonunits_iff] · refine ⟨inferInstance, fun hVtop => IsDiscreteValuationRing.not_isField V.toSubring ?_⟩ exact (step_5_3_isField_iff V).2 hVtop · have hfield : ¬ IsField V.toSubring := fun h => hV ((step_5_3_isField_iff V).1 h) have := step_5_3_valuationRing V exact ((IsDiscreteValuationRing.TFAE V.toSubring hfield).out 1 2).2 this · obtain ⟨ϖ, hϖ⟩ := IsDiscreteValuationRing.exists_irreducible V.toSubring refine ⟨ϖ, hϖ.maximalIdeal_eq, fun f hf => ?_⟩ obtain ⟨k, u, hku⟩ := step_5_3_ord_exists V hϖ hf exact ⟨k, ⟨u, hku⟩, fun l ⟨u', hl⟩ => step_5_3_ord_unique V hϖ (hl.symm.trans hku)⟩
More Lean proofs and definitions
theorem step_5_3_valuationRing {F : Type*} [Field F] (V : ValuationSubring F) : ValuationRing V.toSubring := inferInstanceAs (ValuationRing V)
/-- A valuation ring equal to `F` is a field. -/ theorem step_5_3_isField_iff {F : Type*} [Field F] (V : ValuationSubring F) : IsField V.toSubring ↔ V = ⊤ := by constructor · intro hV rw [eq_top_iff] intro f _ rcases V.mem_or_inv_mem f with h | h · exact h · by_cases hf : f = 0 · rw [hf]; exact V.zero_mem · obtain ⟨g, hg⟩ := hV.mul_inv_cancel (a := (⟨f⁻¹, h⟩ : V.toSubring)) (fun h0 => inv_ne_zero hf (congrArg Subtype.val h0)) have hg' : f⁻¹ * (g : F) = 1 := congrArg Subtype.val hg have : (g : F) = f := by rw [← one_mul (g : F), ← mul_inv_cancel₀ hf, mul_assoc, hg', mul_one] exact this ▸ g.2 · rintro rfl refine ⟨⟨0, 1, fun h => zero_ne_one (congrArg Subtype.val h)⟩, fun a b => mul_comm a b, fun {a} ha => ?_⟩ refine ⟨⟨(a : F)⁻¹, ValuationSubring.mem_top _⟩, Subtype.ext ?_⟩ exact mul_inv_cancel₀ fun h => ha (Subtype.ext h)
/-- Existence of `f = ϖ^k u` in a discrete valuation ring `V ⊆ F`. -/ theorem step_5_3_ord_exists {F : Type*} [Field F] (V : ValuationSubring F) [IsDiscreteValuationRing V.toSubring] {ϖ : V.toSubring} (hϖ : Irreducible ϖ) {f : F} (hf : f ≠ 0) : ∃ (k : ℤ) (u : V.toSubringˣ), f = (ϖ : F) ^ k * ((u : V.toSubring) : F) := by rcases V.mem_or_inv_mem f with h | h · obtain ⟨m, u, hu⟩ := IsDiscreteValuationRing.eq_unit_mul_pow_irreducible (x := (⟨f, h⟩ : V.toSubring)) (fun h0 => hf (congrArg Subtype.val h0)) hϖ refine ⟨m, u, ?_⟩ have := congrArg Subtype.val hu simp only [Subring.coe_mul, SubmonoidClass.coe_pow] at this rw [zpow_natCast, mul_comm] exact this · obtain ⟨m, u, hu⟩ := IsDiscreteValuationRing.eq_unit_mul_pow_irreducible (x := (⟨f⁻¹, h⟩ : V.toSubring)) (fun h0 => inv_ne_zero hf (congrArg Subtype.val h0)) hϖ refine ⟨-(m : ℤ), u⁻¹, ?_⟩ have := congrArg Subtype.val hu simp only [Subring.coe_mul, SubmonoidClass.coe_pow] at this have hinv : ((u : V.toSubring) : F) * ((u⁻¹ : V.toSubringˣ) : V.toSubring) = 1 := by rw [← Subring.coe_mul, Units.mul_inv]; rfl have hϖ0 : (ϖ : F) ≠ 0 := fun h0 => hϖ.ne_zero (Subtype.ext h0) rw [zpow_neg, zpow_natCast] calc f = (f⁻¹)⁻¹ := (inv_inv f).symm _ = ((u : V.toSubring) : F)⁻¹ * ((ϖ : F) ^ m)⁻¹ := by rw [this, mul_inv] _ = _ := by rw [mul_comm, eq_inv_of_mul_eq_one_right hinv]
/-- Uniqueness of `ord_V`: `ϖ^k u = ϖ^l u'` with units `u, u'` forces `k = l`. -/ theorem step_5_3_ord_unique {F : Type*} [Field F] (V : ValuationSubring F) {ϖ : V.toSubring} (hϖ : Irreducible ϖ) {k l : ℤ} {u u' : V.toSubringˣ} (h : (ϖ : F) ^ k * ((u : V.toSubring) : F) = (ϖ : F) ^ l * ((u' : V.toSubring) : F)) : k = l := by have hϖ0 : (ϖ : F) ≠ 0 := fun h0 => hϖ.ne_zero (Subtype.ext h0) have hu : ∀ w : V.toSubringˣ, ((w : V.toSubring) : F) ≠ 0 := fun w h0 => w.ne_zero (Subtype.ext h0) -- `ϖ^m` with `m ≥ 1` is not the image of a unit have key : ∀ (m : ℕ) (w : V.toSubringˣ), (ϖ : F) ^ (m + 1) ≠ ((w : V.toSubring) : F) := by intro m w hw have : ϖ ^ (m + 1) = (w : V.toSubring) := Subtype.ext (by simpa using hw) exact hϖ.not_isUnit ((isUnit_pow_iff (Nat.succ_ne_zero m)).mp (this ▸ w.isUnit)) rcases lt_trichotomy k l with hkl | hkl | hkl · exfalso obtain ⟨m, hm⟩ : ∃ m : ℕ, l - k = (m + 1 : ℕ) := ⟨(l - k - 1).toNat, by omega⟩ apply key m (u * u'⁻¹) have : (ϖ : F) ^ (l - k) = ((u : V.toSubring) : F) * ((u'⁻¹ : V.toSubringˣ) : V.toSubring) := by have hinv : ((u' : V.toSubring) : F) * ((u'⁻¹ : V.toSubringˣ) : V.toSubring) = 1 := by rw [← Subring.coe_mul, Units.mul_inv]; rfl rw [zpow_sub₀ hϖ0, div_eq_iff (zpow_ne_zero _ hϖ0)] calc (ϖ : F) ^ l = (ϖ : F) ^ l * ((u' : V.toSubring) : F) * ((u'⁻¹ : V.toSubringˣ) : V.toSubring) := by rw [mul_assoc, hinv, mul_one] _ = _ := by rw [← h]; ring rw [hm, zpow_natCast] at this rw [this, Units.val_mul] rfl · exact hkl · exfalso obtain ⟨m, hm⟩ : ∃ m : ℕ, k - l = (m + 1 : ℕ) := ⟨(k - l - 1).toNat, by omega⟩ apply key m (u' * u⁻¹) have : (ϖ : F) ^ (k - l) = ((u' : V.toSubring) : F) * ((u⁻¹ : V.toSubringˣ) : V.toSubring) := by have hinv : ((u : V.toSubring) : F) * ((u⁻¹ : V.toSubringˣ) : V.toSubring) = 1 := by rw [← Subring.coe_mul, Units.mul_inv]; rfl rw [zpow_sub₀ hϖ0, div_eq_iff (zpow_ne_zero _ hϖ0)] calc (ϖ : F) ^ k = (ϖ : F) ^ k * ((u : V.toSubring) : F) * ((u⁻¹ : V.toSubringˣ) : V.toSubring) := by rw [mul_assoc, hinv, mul_one] _ = _ := by rw [h]; ring rw [hm, zpow_natCast] at this rw [this, Units.val_mul] rfl
Comparison notes
definition_5_3 states the printed definitions as properties of Mathlib's ValuationSubring, IsDiscreteValuationRing, PowerSeries and LaurentSeries, for a field F. (a) A subring S ⊆ F has a ∈ S or a⁻¹ ∈ S for every a ∈ F^× exactly when it is a valuation subring, and a valuation ring W is local with 𝔪_W = W ∖ W^×. (b) A valuation ring V is a discrete valuation ring exactly when it is Noetherian and V ≠ F; then 𝔪_V = (ϖ) for some ϖ ∈ V, and every f ∈ F^× is ϖ^k u with u ∈ V^× and a unique k = ord_V(f) ∈ ℤ (step_5_3_ord_exists, step_5_3_ord_unique). (c) κ((t)) is the fraction field of κ[[t]], for every field κ. The consequences stated in (a) and (b) are cited in the paper [Matsumura, §10–§11] and proved in Lean from Mathlib. Lemmas 5.9 and 5.10 use (a) and the characterization of discrete valuation rings in (b) through definition_5_3; no declaration refers to the rest of (b) or to (c).
Review notes
Definition 5.4 — Irreducibility, smooth locus, conormal space and conormal cone
- A closed set is irreducible if it is not a union of two proper closed subsets, equivalently if is prime [hartshorne, Cor I.1.4].
- For irreducible closed , the smooth locus is the set of at which the Jacobian matrix of generators of has rank , i.e. the points where has a well-defined tangent space of the expected dimension (i.e. without kinks etc.). It is open and dense in . [hartshorne, Ch. I.5]
- For the conormal space is
- denotes the conormal cone.
/-- **Definition 5.4(a), "equivalently if `I(Y)` is prime"** ([Hartshorne, Cor. I.1.4]). As there, `Y` is nonempty (the empty set has `I(∅) = K[x]`, which is not prime). -/ theorem definition_5_4_a {Y : Set (Fin n → k)} (hne : Y.Nonempty) (hY : IsZariskiClosed Y) : IsIrreducibleClosed Y ↔ (MvPolynomial.vanishingIdeal k Y).IsPrime := by constructor · rintro ⟨-, hirr⟩ refine ⟨fun htop => ?_, fun {f g} hfg => ?_⟩ · obtain ⟨y, hy⟩ := hne have h1 : (1 : MvPolynomial (Fin n) k) ∈ MvPolynomial.vanishingIdeal k Y := htop ▸ Submodule.mem_top have := MvPolynomial.mem_vanishingIdeal_iff.mp h1 y hy simp at this · by_contra h push Not at h obtain ⟨hf, hg⟩ := h obtain ⟨J, rfl⟩ := hY apply hirr have hproper : ∀ e : MvPolynomial (Fin n) k, e ∉ MvPolynomial.vanishingIdeal k (MvPolynomial.zeroLocus k J) → MvPolynomial.zeroLocus k (J ⊔ Ideal.span {e}) ⊂ MvPolynomial.zeroLocus k J := by intro e he refine ⟨MvPolynomial.zeroLocus_anti_mono le_sup_left, fun hsub => he ?_⟩ rw [MvPolynomial.mem_vanishingIdeal_iff] intro y hy exact ((mem_zeroLocus_sup_span_singleton J e y).mp (hsub hy)).2 refine ⟨_, _, ⟨_, rfl⟩, ⟨_, rfl⟩, hproper f hf, hproper g hg, ?_⟩ ext y simp only [Set.mem_union, mem_zeroLocus_sup_span_singleton] constructor · intro hy have h0 := MvPolynomial.mem_vanishingIdeal_iff.mp hfg y hy rw [map_mul] at h0 rcases mul_eq_zero.mp h0 with h0 | h0 · exact Or.inl ⟨hy, h0⟩ · exact Or.inr ⟨hy, h0⟩ · rintro (⟨hy, -⟩ | ⟨hy, -⟩) <;> exact hy · intro hP refine ⟨hY, ?_⟩ rintro ⟨Y₁, Y₂, hY₁, hY₂, h₁, h₂, hU⟩ -- a proper closed `Y_i` is cut out inside `Y` by some `f_i ∈ I(Y_i) ∖ I(Y)` have hsep : ∀ Z : Set (Fin n → k), IsZariskiClosed Z → Z ⊂ Y → ∃ f ∈ MvPolynomial.vanishingIdeal k Z, f ∉ MvPolynomial.vanishingIdeal k Y := by intro Z hZ hZY obtain ⟨y, hyY, hyZ⟩ := Set.exists_of_ssubset hZY rw [← zeroLocus_vanishingIdeal_of_isZariskiClosed hZ, MvPolynomial.mem_zeroLocus_iff] at hyZ push Not at hyZ obtain ⟨f, hf, hfy⟩ := hyZ exact ⟨f, hf, fun hfY => hfy (MvPolynomial.mem_vanishingIdeal_iff.mp hfY y hyY)⟩ obtain ⟨f₁, hf₁, hf₁Y⟩ := hsep Y₁ hY₁ h₁ obtain ⟨f₂, hf₂, hf₂Y⟩ := hsep Y₂ hY₂ h₂ have hprod : f₁ * f₂ ∈ MvPolynomial.vanishingIdeal k Y := by rw [MvPolynomial.mem_vanishingIdeal_iff] intro y hy rw [hU] at hy rw [map_mul] rcases hy with hy | hy · rw [MvPolynomial.mem_vanishingIdeal_iff.mp hf₁ y hy, zero_mul] · rw [MvPolynomial.mem_vanishingIdeal_iff.mp hf₂ y hy, mul_zero] rcases hP.mem_or_mem hprod with h | h · exact hf₁Y h · exact hf₂Y h
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- **Definition 5.4(b), the Jacobian criterion.** `y` lies in Mathlib's smooth locus of `K[x]/𝔭` iff the Jacobian of `𝔭` has rank `n - dim Y` at `y ∈ Y`. The converse direction picks equations `f_1, …, f_c ∈ 𝔭` with independent differentials at `y` and applies `smoothAffinePoint_of_jacobian`. -/ theorem definition_5_4_b_iff [CharZero k] (P : PrimeSpectrum (MvPolynomial (Fin n) k)) (y : Fin n → k) : SmoothAffinePoint P.asIdeal y ↔ JacobianSmoothPoint P y := by classical refine ⟨definition_5_4_b P, fun h ↦ ?_⟩ have hle := componentDimension_le P rw [jacobianSmoothPoint_iff_trdeg] at h obtain ⟨hy, hr⟩ := h obtain ⟨s, hsub, hspan, hli⟩ := exists_linearIndependent k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1) have : Fintype s := (hli.setFinite).fintype have hcard : Fintype.card s = Module.finrank k (Submodule.span k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1)) := by rw [← hspan, ← finrank_span_eq_card hli, Subtype.range_coe] have hg : ∀ v : s, ∃ g : MvPolynomial (Fin n) k, g ∈ P.asIdeal ∧ differentialAt y g = v := fun v ↦ by obtain ⟨g, hg⟩ := hsub v.2 exact ⟨g.1, g.2, hg⟩ choose g hgP hgd using hg have hli' : LinearIndependent k fun v : s ↦ differentialAt y (g v) := by convert hli using 1 funext v exact hgd v exact smoothAffinePoint_of_jacobian P hy g hgP hli' (by change Fintype.card s + componentDimension P = n omega)
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
variable {k : Type u} [Field k] {n : ℕ}Source ↗
variable [IsAlgClosed k]
More Lean proofs and definitions
/-- **Definition 5.4(a).** A closed set `Y` is irreducible if it is not a union of two proper closed subsets. -/ def IsIrreducibleClosed (Y : Set (Fin n → k)) : Prop := IsZariskiClosed Y ∧ ¬ ∃ Y₁ Y₂ : Set (Fin n → k), IsZariskiClosed Y₁ ∧ IsZariskiClosed Y₂ ∧ Y₁ ⊂ Y ∧ Y₂ ⊂ Y ∧ Y = Y₁ ∪ Y₂
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- A Zariski-closed subset of `X = K^n`: `Y = V_K(J)` for an ideal `J ⊆ K[x]`. -/ def IsZariskiClosed (Y : Set (Fin n → k)) : Prop := ∃ J : Ideal (MvPolynomial (Fin n) k), Y = MvPolynomial.zeroLocus k J
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- For nonempty irreducible closed `Y`, the prime `I(Y)` as a point of `Spec K[x]`; Definitions 5.4(b)–(d) and 5.5 are stated in terms of it. -/ def idealPoint (Y : Set (Fin n → k)) (hne : Y.Nonempty) (hY : IsIrreducibleClosed Y) : PrimeSpectrum (MvPolynomial (Fin n) k) := ⟨MvPolynomial.vanishingIdeal k Y, (definition_5_4_a hne hY.1).1 hY⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- A closed set is the zero set of its vanishing ideal: `V_K(I(Y)) = Y`. -/ theorem zeroLocus_vanishingIdeal_of_isZariskiClosed {Y : Set (Fin n → k)} (hY : IsZariskiClosed Y) : MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) = Y := by obtain ⟨J, rfl⟩ := hY exact le_antisymm (MvPolynomial.zeroLocus_anti_mono (MvPolynomial.le_vanishingIdeal_zeroLocus J)) (MvPolynomial.zeroLocus_vanishingIdeal_le _)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- **Definition 5.4(b), the Jacobian smooth locus.** `y ∈ Y_sm` iff `y ∈ Y = V(𝔭)` and the Jacobian matrix `(∂_i g_j(y))` of generators `g_1, …, g_s` of `I(Y) = 𝔭` has rank `n - dim Y`, written `rank + dim Y = n` with `dim Y` the Krull dimension of `K[x]/𝔭`. The rank does not depend on the generators (`jacobian_rank_eq_finrank_differentialSpan`, `jacobianSmoothPoint_iff_span`). -/ def JacobianSmoothPoint (P : PrimeSpectrum (MvPolynomial (Fin n) k)) (y : Fin n → k) : Prop := y ∈ MvPolynomial.zeroLocus k P.asIdeal ∧ ∃ (s : ℕ) (g : Fin s → MvPolynomial (Fin n) k), Ideal.span (Set.range g) = P.asIdeal ∧ ((Matrix.of fun j i ↦ differentialAt y (g j) i).rank : WithBot ℕ∞) + ringKrullDim (MvPolynomial (Fin n) k ⧸ P.asIdeal) = n
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
variable {k : Type u} [Field k] {n : ℕ}
/-- **Definition 5.4(b), Lean smooth points are Jacobian smooth.** A point of Mathlib's smooth locus of `K[x]/𝔭` (`SmoothAffinePoint`) lies in the paper's `Y_sm`: the Jacobian of `𝔭` has rank `n - dim Y` there. The converse is `definition_5_4_b_iff`. -/ theorem definition_5_4_b [CharZero k] (P : PrimeSpectrum (MvPolynomial (Fin n) k)) {y : Fin n → k} (hy : SmoothAffinePoint P.asIdeal y) : JacobianSmoothPoint P y := by rw [jacobianSmoothPoint_iff_trdeg] refine ⟨smoothAffinePoint_mem_zeroLocus P.asIdeal hy, ?_⟩ have h1 := finrank_differentialSpan_add_finrank_zariskiTangentSpace P.asIdeal y have h2 := finrank_zariskiTangentSpace_of_smoothAffinePoint P hy have h3 := (component_kaehler_finrank_eq_trdeg (k := k) (AffineComponentCoordinateSplit.componentCoordinate P) (ExactDivisorialVisibleFrameExistence.componentCoordinate_adjoin_eq_top P)).2 rw [componentDimension] omega
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
variable {k : Type u} [Field k] {n : ℕ}Source ↗
variable [IsAlgClosed k]
/-- **Definition 5.4(b), transcendence-degree form.** -/ theorem jacobianSmoothPoint_iff_trdeg (P : PrimeSpectrum (MvPolynomial (Fin n) k)) (y : Fin n → k) : JacobianSmoothPoint P y ↔ y ∈ MvPolynomial.zeroLocus k P.asIdeal ∧ Module.finrank k (Submodule.span k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1)) = n - componentDimension P := by have hle := componentDimension_le P rw [jacobianSmoothPoint_iff_span, componentDimension_eq_ringKrullDim] apply and_congr_right intro _ constructor · intro h have h' : Module.finrank k (Submodule.span k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1)) + componentDimension P = n := by exact_mod_cast h omega · intro h have h' : Module.finrank k (Submodule.span k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1)) + componentDimension P = n := by omega exact_mod_cast h'
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
variable {k : Type u} [Field k] {n : ℕ}
/-- **Definition 5.4(c).** `N*_y Y = span_K {df_y : f ∈ I}`, with covectors in coordinates. -/ theorem definition_5_4_c (I : Ideal (MvPolynomial (Fin n) k)) (y ξ : Fin n → k) : coordinateCovector ξ ∈ affineConormalSpace y I ↔ ξ ∈ Submodule.span k (Set.range fun f : I => differentialAt y f.1) := by have hspan : equationCovectorSpan y I = (Submodule.span k (Set.range fun f : I => differentialAt y f.1)).map (coordinateCovectorLinear (k := k) (n := n)) := by rw [Submodule.map_span, ← Set.range_comp] rfl rw [affineConormalSpace_eq_equationCovectorSpan, hspan] constructor · rintro ⟨η, hη, hηξ⟩ have : η = ξ := coordinateCovector_injective hηξ exact this ▸ hη · intro hξ exact ⟨ξ, hξ, rfl⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- **Definition 5.4(d), as printed.** The conormal cone `N(Y) = ⋃_{y ∈ Y_sm} N*_y Y` over the Jacobian smooth locus, with `N*_y Y = span_K {df_y : f ∈ I(Y)}`. -/ def conormalCone (P : PrimeSpectrum (MvPolynomial (Fin n) k)) : Set (Fin n → k) := {ξ | ∃ y, JacobianSmoothPoint P y ∧ ξ ∈ Submodule.span k (Set.range fun f : P.asIdeal ↦ differentialAt y f.1)}
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗variable {k : Type u} [Field k] [IsAlgClosed k] {n : ℕ}
/-- **Definition 5.4(d).** `N(Y) = ⋃_{y ∈ Y_sm} N*_y Y`, in the paper's form of `N*_y Y`. -/ theorem definition_5_4_d [IsAlgClosed k] (I : Ideal (MvPolynomial (Fin n) k)) (ξ : Fin n → k) : ξ ∈ smoothConormalFibreProjection I ↔ ∃ y, SmoothAffinePoint I y ∧ ξ ∈ Submodule.span k (Set.range fun f : I => differentialAt y f.1) := by constructor · rintro ⟨y, hy, hξ⟩ exact ⟨y, hy, (definition_5_4_c I y ξ).1 hξ⟩ · rintro ⟨y, hy, hξ⟩ exact ⟨y, hy, (definition_5_4_c I y ξ).2 hξ⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- **`N(Y)` as printed is the Lean conormal cone**: the Jacobian smooth locus is Mathlib's smooth locus (`definition_5_4_b_iff`). -/ theorem conormalCone_paper_eq [CharZero k] (P : PrimeSpectrum (MvPolynomial (Fin n) k)) : conormalCone P = smoothConormalFibreProjection P.asIdeal := by ext ξ rw [definition_5_4_d] simp only [conormalCone, Set.mem_ofPred_eq, definition_5_4_b_iff] omit [IsAlgClosed k] in
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗variable {k : Type u} [Field k] [IsAlgClosed k] {n : ℕ}
/-- **Krull dimension equals transcendence degree** for a finitely generated domain over a field. -/ theorem ringKrullDim_eq_trdeg (A : Type u) [CommRing A] [IsDomain A] [Algebra K A] [Algebra.FiniteType K A] : ringKrullDim A = ((Algebra.trdeg K A).toNat : WithBot ℕ∞) := le_antisymm (ringKrullDim_le_trdeg A) (trdeg_le_ringKrullDim A)
/-- **Jacobian smooth points are Mathlib smooth points.** If `f_1, …, f_c ∈ 𝔭` have independent differentials at `y ∈ V(𝔭)` and `c + trdeg_K Frac(K[x]/𝔭) = n`, then `y` is a smooth point of `K[x]/𝔭`. -/ theorem smoothAffinePoint_of_jacobian [IsAlgClosed k] [CharZero k] (𝔓 : PrimeSpectrum (MvPolynomial (Fin n) k)) {y : Fin n → k} (hy : y ∈ MvPolynomial.zeroLocus k 𝔓.asIdeal) {ι : Type*} [Fintype ι] (f : ι → MvPolynomial (Fin n) k) (hf : ∀ i, f i ∈ 𝔓.asIdeal) (hli : LinearIndependent k fun i ↦ differentialAt y (f i)) (hc : Fintype.card ι + (Algebra.trdeg k (FractionRing (MvPolynomial (Fin n) k ⧸ 𝔓.asIdeal))).toNat = n) : SmoothAffinePoint 𝔓.asIdeal y := by have hloc := map_le_map_span_of_jacobian 𝔓 hy f hf hli hc have hyI : ∀ p ∈ 𝔓.asIdeal, MvPolynomial.aeval y p = 0 := fun p hp ↦ (MvPolynomial.mem_zeroLocus_iff.1 hy) p hp let e : (MvPolynomial (Fin n) k ⧸ 𝔓.asIdeal) →ₐ[k] k := Ideal.Quotient.liftₐ _ (aeval y) hyI have heval : ∀ p, e (Ideal.Quotient.mk 𝔓.asIdeal p) = MvPolynomial.aeval y p := fun p ↦ by simp [e] refine ⟨e, heval, ?_⟩ have : (RingHom.ker e.toRingHom).IsPrime := RingHom.ker_isPrime e exact formallySmooth_localization_of_jacobian 𝔓.asIdeal (Ideal.Quotient.mkₐ k 𝔓.asIdeal) Ideal.Quotient.mk_surjective (fun p ↦ Ideal.Quotient.eq_zero_iff_mem) y e heval f hf hli hloc
def SmoothAffinePoint (I : Ideal (MvPolynomial (Fin n) k)) (y : Fin n → k) : Prop := ∃ e : (MvPolynomial (Fin n) k ⧸ I) →ₐ[k] k, (∀ p, e (Ideal.Quotient.mk I p) = MvPolynomial.aeval y p) ∧ let hprime : (RingHom.ker e.toRingHom).IsPrime := RingHom.ker_isPrime e (⟨RingHom.ker e.toRingHom, hprime⟩ : PrimeSpectrum (MvPolynomial (Fin n) k ⧸ I)) ∈ Algebra.smoothLocus k (MvPolynomial (Fin n) k ⧸ I)
/-- The embedded conormal space, defined as the annihilator of the embedded Zariski tangent space. -/ def affineConormalSpace (y : Fin n → k) (I : Ideal (MvPolynomial (Fin n) k)) : Submodule k (Module.Dual k (AffineTangentVector k n)) := (zariskiTangentSpace y I).dualAnnihilator
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.BaseRelativePoissonSource ↗open Stafford38.Characteristic.BaseZeroSectionSource ↗open Stafford38.Characteristic.ReducedSupportIdealSource ↗open Stafford38.Characteristic.ZeroSectionContainmentSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.PointwiseConormalContainmentSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable {k : Type*} [Field k] {n : ℕ}
Comparison notes
(a) IsIrreducibleClosed is the printed definition: Y is closed (IsZariskiClosed, Y = V_K(J) for an ideal J) and not the union of two proper closed subsets. definition_5_4_a proves the cited equivalence with primality of I(Y) [Hartshorne, Cor. I.1.4] for nonempty closed Y. The paper does not require Y nonempty, and the equivalence needs it: the empty set satisfies the printed definition, but I(∅) = K[x] is not prime. For nonempty irreducible closed Y, idealPoint Y is I(Y) as a point of Spec K[x]; (b)–(d) are stated at this point, written 𝔭 below. (b) JacobianSmoothPoint is the printed smooth locus: y ∈ Y and the Jacobian of generators of 𝔭 has rank n − dim Y, dim Y the Krull dimension of K[x]/𝔭, independent of the generators. definition_5_4_b_iff proves that it is Mathlib's smooth locus of K[x]/𝔭 (SmoothAffinePoint, used by the proofs), i.e. the Jacobian criterion; it uses dim Y = trdeg_K F (Support.ringKrullDim_eq_trdeg), standard dimension theory that the paper does not prove and the repository proves (Noether normalization and going up, and the drop of transcendence degree along a chain of primes). The cited density of Y_sm is not formalized. (c) definition_5_4_c: N*_y Y = span{df_y : f ∈ I(Y)} in coordinates. (d) conormalCone is the printed N(Y) over the Jacobian smooth locus; definition_5_4_d and conormalCone_paper_eq identify it with the cone of the proofs.
Review notes
Definition 5.5 — Projective conormal directions Dir(Y)
/-- **Definition 5.5.** The printed `Dir(Y)` is the Lean direction closure used in Theorem 5.7(ii). -/ theorem definition_5_5 [CharZero k] (P : PrimeSpectrum (MvPolynomial (Fin n) k)) : conormalDirections P = projectiveHomogeneousClosure (projectivizedDirectionSet (smoothConormalDirectionSet P.asIdeal)) := by rw [projectivizedDirectionSet_smoothConormalDirectionSet, conormalDirections, conormalCone_paper_eq]
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗variable {k : Type u} [Field k] [IsAlgClosed k] {n : ℕ}
More Lean proofs and definitions
/-- **Definition 5.5, as printed.** `Dir(Y)` is the Zariski closure in `ℙ((K^n)^*)` of `{[ξ] : 0 ≠ ξ ∈ N(Y)}`; projective Zariski closed sets are the common zero sets of homogeneous polynomials. -/ def conormalDirections (P : PrimeSpectrum (MvPolynomial (Fin n) k)) : Set (Projectivization k (Fin n → k)) := projectiveHomogeneousClosure (projectivizedDirectionSet (conormalCone P))
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗variable {k : Type u} [Field k] [IsAlgClosed k] {n : ℕ}
def projectiveHomogeneousClosure (T : Set (Projectivization k (Fin n → k))) : Set (Projectivization k (Fin n → k)) := {z | ∀ P : MvPolynomial (Fin n) k, ∀ d : ℕ, P.IsHomogeneous d → T ⊆ projectiveHomogeneousZeroLocus P → z ∈ projectiveHomogeneousZeroLocus P}
Variables and notation
Source ↗open scoped LinearAlgebra.ProjectivizationSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ}
def projectivizedDirectionSet (T : Set (Fin n → k)) : Set (Projectivization k (Fin n → k)) := {z | ∃ v, v ∈ T ∧ ∃ hv : v ≠ 0, z = Projectivization.mk k v hv}
Variables and notation
Source ↗open scoped LinearAlgebra.ProjectivizationSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ}
Comparison notes
conormalDirections is Dir(Y) as printed: the closure in ℙ((K^n)*) of {[ξ] : 0 ≠ ξ ∈ N(Y)} for the printed cone, the projective closed sets being the common zero sets of homogeneous polynomials (projectiveHomogeneousClosure). definition_5_5 identifies it with the direction closure of the proofs. Dir(Y) enters only Theorem 5.7(ii); neither the paper nor Lean uses it afterwards, so definition_5_5 is proved but not used downstream.
Review notes
Definition 5.6 — Fibre-conical and involutive sets
We say it is involutive if is.
/-- **Definition 5.6.** `W` is fibre-conical iff it is stable under the fibrewise scaling `(y, ξ) ↦ (y, tξ)`, `t ∈ K^×`: the scaling maps `W ∩ T*_y X` into `W ∩ T*_y X` for every fibre `T*_y X = π⁻¹(y)` (`Section2.fibre`, Definition 2.5). -/ theorem definition_5_6 [IsAlgClosed k] [CharZero k] (W : Set (PhaseVar n → k)) : Stafford38.Geometry.FibreConicalVanishingIdeal.IsFibreConical W ↔ ∀ y ξ : Fin n → k, Sum.elim y ξ ∈ W ∩ Stafford38.Paper.Section2.fibre y → ∀ t : k, t ≠ 0 → Sum.elim y (t • ξ) ∈ W ∩ Stafford38.Paper.Section2.fibre y := by constructor · intro hW y ξ hq t ht exact ⟨by simpa [Pi.smul_def] using hW _ hq.1 t ht, rfl⟩ · intro hW q hq t ht have hsplit : q = Sum.elim (fun i => q (.inl i)) (fun i => q (.inr i)) := by funext i; cases i <;> rfl rw [hsplit] at hq simpa [Pi.smul_def] using (hW _ _ ⟨hq, rfl⟩ t ht).1
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- Stability under every nonzero scalar dilation in each cotangent fibre. -/ def IsFibreConical (W : Set (PhaseVar n → k)) : Prop := ∀ q ∈ W, ∀ a : k, a ≠ 0 → Sum.elim (fun i => q (.inl i)) (fun i => a * q (.inr i)) ∈ W
/-- **Definition 2.10 (involutive closed set).** A closed set `W ⊆ K^{2n}` of closed points is involutive if `I(W)` is. -/ def IsInvolutiveSet {n : ℕ} (W : Set (PhaseVar n → k)) : Prop := IsInvolutive k (vanishingIdealOf k W)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
definition_5_6: the repository's IsFibreConical W holds iff W is stable under (y, ξ) ↦ (y, tξ), t ∈ K^×, on every fibre T*_y X (Section2.fibre). IsInvolutiveSet W means that the vanishing ideal I(W) is involutive, as in the final sentence.
Review notes
Theorem 5.7 — Asymptotic conormal direction over an algebraically closed field
- Every polynomial vanishing on satisfies .
- .
Lean ():
Section5.theorem_5_7_i_printed,
Section5.theorem_5_7_ii_printed
/-- **Theorem 5.7(i).** Let `Y ⊆ X = K^n` be nonempty, closed and irreducible (Definition 5.4(a)) with `Y ∩ H = ∅`, `H = {x_1 = 0}`, and let `N(Y) = ⋃_{y ∈ Y_sm} N*_y Y` be the conormal cone over the smooth locus. Every polynomial `P ∈ K[ξ_1, …, ξ_n]` vanishing on `N(Y)` satisfies `P(1, 0, …, 0) = 0`. Derived from `theorem_5_7_i_prime` with `𝔭 = I(Y)`. -/ theorem theorem_5_7_i_printed {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (Y : Set (Fin m → k)) (hne : Y.Nonempty) (hY : IsIrreducibleClosed Y) (hH : Y ∩ {y | y ⟨0, hm⟩ = 0} = ∅) (P : MvPolynomial (Fin m) k) (hP : ∀ ξ ∈ conormalCone (idealPoint Y hne hY), MvPolynomial.eval ξ P = 0) : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P = 0 := by refine theorem_5_7_i_prime hm (idealPoint Y hne hY) (fun y hy h0 => ?_) P hP change y ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) at hy rw [zeroLocus_vanishingIdeal_of_isZariskiClosed hY.1] at hy exact (Set.eq_empty_iff_forall_notMem.mp hH) y ⟨hy, h0⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- **Theorem 5.7(ii).** For `Y` as in (i), `[dx_1] ∈ Dir(Y)`. Derived from `theorem_5_7_ii_prime` with `𝔭 = I(Y)`. -/ theorem theorem_5_7_ii_printed {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (Y : Set (Fin m → k)) (hne : Y.Nonempty) (hY : IsIrreducibleClosed Y) (hH : Y ∩ {y | y ⟨0, hm⟩ = 0} = ∅) : Projectivization.mk k (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) (by intro h; have := congrFun h ⟨0, hm⟩; simp at this) ∈ conormalDirections (idealPoint Y hne hY) := by refine theorem_5_7_ii_prime hm (idealPoint Y hne hY) (fun y hy h0 => ?_) change y ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) at hy rw [zeroLocus_vanishingIdeal_of_isZariskiClosed hY.1] at hy exact (Set.eq_empty_iff_forall_notMem.mp hH) y ⟨hy, h0⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- **Theorem 5.7(i), along the paper's proof.** If `Y = V(𝔭)` is nonempty, irreducible and avoids `H = {x_1 = 0}`, every polynomial vanishing on the conormal cone `N(Y)` over the smooth locus vanishes at `(1, 0, …, 0)`. -/ theorem theorem_5_7_i_paper {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := by classical intro Pol hPol by_cases halg : IsAlgebraic k (componentCoordinate I ⟨0, hm⟩) · -- First paragraph: `x̄_1` algebraic, so `x_1 − c ∈ 𝔭` with `c ≠ 0`, and -- `(1, 0, …, 0) = dx_1` is conormal at a smooth point of `Y`, so lies in `N(Y)`. obtain ⟨-, -, -, hN⟩ := step_coordinate_algebraic hm I havoid halg exact hPol _ hN -- `x̄_1` transcendental. `Y ∩ H = ∅` makes `x̄_1` a unit of `R`. have hunit := exists_coordinate_inverse_of_avoidance I.asIdeal ⟨0, hm⟩ havoid -- Lemmas 5.9–5.10: the discrete valuation ring `V` of `F` at infinity, the -- column `q_i = x̄_i / x̄_j` and `κ = V/𝔪_V`, read in `V̂ ≅ κ[[t]]`. obtain ⟨w, hfin⟩ := Stafford38.Geometry.GeneralDivisorialVisibleFrame.generalDivisorialVisibleFrameWithKaehlerFinite hm I hunit halg let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField I) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : IsLocalRing V := W.place.isDiscrete.toIsLocalRing letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra I ⟨0, hm⟩ W letI : CharZero κ := charZero_of_injective_algebraMap (FaithfulSMul.algebraMap_injective k κ) -- Paragraph "`V̂ ≅ κ[[t]]`": `F ↪ κ((t))`, and the residue is the constant term. obtain ⟨-, -, -, -, hres, -, -, -⟩ := step_completion hm I w let q : Fin (m + 1) → PowerSeries κ := fun a ↦ retainedToCompletedPowerSeries W (C.q a) let axis : Fin (m + 1) := Fin.succ ⟨0, hm⟩ -- Paragraph "`a = ord q_0`, `b = ord q_1`": `0 < a < b`, `j ∉ {0, 1}`. obtain ⟨a, b, u₀, u₁, ha, hab, hq0, hu₀, hq1, -, hqj, -, -⟩ := step_orders hm I w hunit have hq0ne : q 0 ≠ 0 := by intro h apply hu₀ have h' : (PowerSeries.X : PowerSeries κ) ^ a * u₀ = 0 := hq0.symm.trans h rw [mul_eq_zero] at h' rcases h' with h' | h' · exact absurd h' (pow_ne_zero a PowerSeries.X_ne_zero) · rw [h', map_zero] -- Paragraph defining `y`, equation (generic): `f(y) = 0` for `f ∈ 𝔭`. have hbase : ∀ f ∈ I.asIdeal, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) (dehomogenizedPoint (laurentColumn q)) f = 0 := fun f hf ↦ (step_generic_point hm I w f).2 hf -- Lemma 5.11: `d(q_{i_l}(0))`, `i_l ∉ {0, 1, j}`, form a basis of `Ω_{κ/K}`. obtain ⟨-, -, -, d, rows, hinj, -, hli, hspan, hd⟩ := witness_lemma_5_11 hm I w hunit have hresRows : (fun l ↦ KaehlerDifferential.D k κ (PowerSeries.constantCoeff (q (rows l)))) = fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l))) := by funext l rw [hres] rw [← hresRows] at hli hspan let rowsE : Fin d ↪ Fin (m + 1) := ⟨rows, hinj⟩ -- Paragraph after Lemma 5.11: dual derivations `D_k` and the matrix `Z(t)`. obtain ⟨D, -, -, hminor, hZj, hZ0, hZ1⟩ := step_dual_derivations (k := k) hm q rowsE hli hspan C.chart hqj a b u₀ u₁ hq0 hq1 let Z := coefficientwiseTangentMatrix q D -- Paragraph defining `λ` (unique), `w = q' − Zλ`, `c ≤ a − 1` and `τ = t^{-c} w`. obtain ⟨lambda, c, tau, -, -, -, -, -, -, -, hfactor, hprim, htaurows, -, htau1, htauj⟩ := step_lambda_w_tau hm q Z rowsE C.chart a b u₀ u₁ hqj hZj ha hab hq0 hu₀ hZ0 hq1 hZ1 hminor -- Paragraph "`M(0)` has full column rank" for `M = (q, Z, τ)`: a unit minor -- `M_ι`, independent columns over `κ((t))`, vanishing `1`-coordinates of `M(0)`. obtain ⟨hinjM, -, hliM, haxis⟩ := step_M_minor hm q Z tau rowsE C.chart b u₁ hqj hZj htauj htaurows hminor hprim (ha.trans hab) hq1 hZ1 htau1 -- Lemma 5.12(a): `dim Θ_y = dim_F Ω_{F/K} ≤ d' + 1`, the bound by Lemma 5.10(c), -- the second fundamental exact sequence for `V → κ` and Nakayama. obtain ⟨⟨hdim, -, hfieldRank, -⟩, -, -⟩ := witness_lemma_5_12_5_13_paper hm I w hfin have hbound : Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I.asIdeal) ≤ d + 1 := by rw [hd] exact hdim.trans_le hfieldRank -- Paragraph "the columns of `M` span `𝒞`" (Lemmas 5.12(a), 5.12(b)). obtain ⟨-, hspanC⟩ := step_M_spans_cone I.asIdeal q hq0ne hbase D Z rfl lambda c tau hfactor hliM hbound -- Paragraph lifting `u = (0, 1, 0, …, 0)`: `uM(0) = 0`, an invertible `M_ι`, -- and `ℓ_υ = u_υ`, `ℓ_ι = −u_υ M_υ M_ι⁻¹` with `ℓM = 0`, `ℓ(0) = u`. have hu : rowMul (axisRow (k := κ) axis) (residueMatrix (formalTangentMatrix q Z tau)) = 0 := by funext col simp only [rowMul, axisRow, residueMatrix, ite_mul, one_mul, zero_mul, Finset.sum_ite_eq', Finset.mem_univ, ite_true, Pi.zero_apply] exact haxis col obtain ⟨rowsι, -, -, -, -, hell, hellres⟩ := step_lift_ell_minor (formalTangentMatrix q Z tau) hinjM (axisRow (k := κ) axis) hu -- Paragraph "`ℓ` annihilates `𝒞`" (Lemma 5.13): `ℓ̃ ∈ span {df_y}`. obtain ⟨hconormal, htail⟩ := step_ell_conormal hm I.asIdeal q hq0ne _ hspanC _ hell hellres -- Final paragraph: `P(ℓ̃(0)) = 0` with `ℓ̃(0) = (1, 0, …, 0)`. have hzero := step_polynomial_limit I Pol hPol _ hbase _ hconormal rw [htail, eval₂_extensionAxis_eq_algebraMap_eval_axis (K := κ) hm Pol] at hzero exact (FaithfulSMul.algebraMap_injective k κ) (hzero.trans (map_zero _).symm)
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.FiniteGradientResidueExtensionSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralConstantCoordinateAxisSource ↗open Stafford38.Geometry.GeneralCoordinateAvoidanceSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗universe u
More Lean proofs and definitions
/-- Theorem 5.7(i) for `Y = V(𝔭)` with `𝔭 = I(Y)` prime: every polynomial `P ∈ K[ξ_1, …, ξ_n]` vanishing on the conormal cone `N(Y)` over the Jacobian smooth locus satisfies `P(1, 0, …, 0) = 0`. `theorem_5_7_i_printed` is the printed form. -/ theorem theorem_5_7_i_prime {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) (Q : MvPolynomial (Fin m) k) (hQ : ∀ ξ ∈ conormalCone I, MvPolynomial.eval ξ Q = 0) : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) Q = 0 := by have hQ' : Q ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal) := by rw [MvPolynomial.mem_vanishingIdeal_iff, ← conormalCone_paper_eq I] simpa using hQ simpa using (MvPolynomial.mem_zeroLocus_iff.1 (theorem_5_7_i hm I havoid)) Q hQ'
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- Theorem 5.7(ii) for `Y = V(𝔭)` with `𝔭` prime: `[dx_1] ∈ Dir(Y)`. `theorem_5_7_ii_printed` is the printed form. -/ theorem theorem_5_7_ii_prime {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : Projectivization.mk k (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) (by intro h; have := congrFun h ⟨0, hm⟩; simp at this) ∈ conormalDirections I := by rw [definition_5_5] exact theorem_5_7_ii hm I havoid
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- A closed set is the zero set of its vanishing ideal: `V_K(I(Y)) = Y`. -/ theorem zeroLocus_vanishingIdeal_of_isZariskiClosed {Y : Set (Fin n → k)} (hY : IsZariskiClosed Y) : MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) = Y := by obtain ⟨J, rfl⟩ := hY exact le_antisymm (MvPolynomial.zeroLocus_anti_mono (MvPolynomial.le_vanishingIdeal_zeroLocus J)) (MvPolynomial.zeroLocus_vanishingIdeal_le _)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}
/-- **Theorem 5.7(i).** If `Y = V(𝔭)` is nonempty, irreducible and avoids `H = {x_1 = 0}`, every polynomial vanishing on the conormal cone `N(Y)` over the smooth locus vanishes at `(1, 0, …, 0)`. Proved paragraph by paragraph in `theorem_5_7_i_paper`. -/ theorem theorem_5_7_i {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := theorem_5_7_i_paper hm I havoid
/-- **Theorem 5.7(ii).** `[dx_1] ∈ Dir(Y)`: by `step_dir_of_fibre`, every homogeneous `G` vanishing on the classes `[ξ]`, `0 ≠ ξ ∈ N(Y)`, vanishes on `N(Y)`, hence at `(1, 0, …, 0)` by (i). -/ theorem theorem_5_7_ii {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : Projectivization.mk k (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) (by intro h; have := congrFun h ⟨0, hm⟩; simp at this) ∈ projectiveHomogeneousClosure (projectivizedDirectionSet (smoothConormalDirectionSet I.asIdeal)) := step_dir_of_fibre I.asIdeal _ _ (theorem_5_7_i hm I havoid)
/-- **Proof of Theorem 5.7(ii) from (i).** Let every polynomial vanishing on `N(Y)` vanish at `ξ ≠ 0`, and let `G` be homogeneous of degree `d` and vanish on all classes `[ξ']` with `0 ≠ ξ' ∈ N(Y)`. By homogeneity `G` vanishes on every nonzero `ξ' ∈ N(Y)`, and at `0` if `d > 0`, hence on `N(Y)`; so `G(ξ) = 0`, and `G` vanishes at every representative `a ξ` of `[ξ]`. If `d = 0`, `G` is a constant vanishing at a class of `N(Y)`, so `G = 0`; that such a class exists follows from the hypothesis itself, applied to a coordinate `ξ_i ≠ 0`. -/ theorem step_dir_of_fibre {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ} (I : Ideal (MvPolynomial (Fin n) k)) (ξ : Fin n → k) (hξ : ξ ≠ 0) (hvan : ξ ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I))) : Projectivization.mk k ξ hξ ∈ projectiveHomogeneousClosure (projectivizedDirectionSet (smoothConormalDirectionSet I)) := by intro G d hG hT -- `G` vanishes on every nonzero `ξ' ∈ N(Y)`. have hnonzero : ∀ v ∈ smoothConormalFibreProjection I, v ≠ 0 → MvPolynomial.eval v G = 0 := by intro v hv hv0 rcases hv with ⟨y, hy, hc⟩ exact hT ⟨v, ⟨y, hy, hv0, hc⟩, hv0, rfl⟩ v hv0 rfl have hξG : MvPolynomial.eval ξ G = 0 := by rcases Nat.eq_zero_or_pos d with hd | hd · -- `d = 0`: `G` is constant, and some nonzero `ξ' ∈ N(Y)` exists. subst hd have hconst : ∀ x : Fin n → k, MvPolynomial.eval x G = MvPolynomial.eval 0 G := by intro x have h := eval_smul_of_isHomogeneous G 0 hG 0 x rw [zero_smul, pow_zero, one_mul] at h exact h.symm obtain ⟨i, hi⟩ : ∃ i, ξ i ≠ 0 := by simpa [funext_iff] using hξ obtain ⟨v, hv, hv0⟩ : ∃ v ∈ smoothConormalFibreProjection I, v ≠ 0 := by by_contra hnone push Not at hnone apply hi have hXi : MvPolynomial.X i ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I) := by rw [MvPolynomial.mem_vanishingIdeal_iff] intro v hv rw [hnone v hv] simp simpa using (MvPolynomial.mem_zeroLocus_iff.mp hvan) _ hXi rw [hconst, ← hconst v] exact hnonzero v hv hv0 · -- `d > 0`: `G(0) = 0` too, so `G` vanishes on `N(Y)`. have hGvan : G ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I) := by rw [MvPolynomial.mem_vanishingIdeal_iff] intro v hv by_cases hv0 : v = 0 · subst hv0 have h := eval_smul_of_isHomogeneous G d hG 0 0 rw [smul_zero, zero_pow hd.ne', zero_mul] at h exact h · exact hnonzero v hv hv0 exact (MvPolynomial.mem_zeroLocus_iff.mp hvan) _ hGvan -- Every representative `a ξ` of `[ξ]`. intro v hv hz obtain ⟨a, ha⟩ := (Projectivization.mk_eq_mk_iff k ξ v hξ hv).mp hz have hscaled := eval_smul_of_isHomogeneous G d hG (a : k) v have ha' : (a : k) • v = ξ := by rw [← Units.smul_def] exact ha rw [ha', hξG] at hscaled exact (mul_eq_zero.mp hscaled.symm).resolve_left (pow_ne_zero d (Units.ne_zero a))
theorem coordinate_axis_mem_smooth_fibre_closure {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := by exact Stafford38.Geometry.SameWitness.coordinate_axis_mem_smooth_fibre_closure hm I havoid
Variables and notation
Source ↗open Stafford38.Geometry.GeneralAsymptoticLaurentAxisSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗universe u
Comparison notes
theorem_5_7_i_printed and theorem_5_7_ii_printed state (i) and (ii) as printed, over an algebraically closed field K of characteristic zero with n ≥ 1: Y ⊆ K^n is nonempty, closed and irreducible (IsIrreducibleClosed, Definition 5.4(a)) with Y ∩ H = ∅, H = {x₁ = 0}, and N(Y) and Dir(Y) are the printed cone over the Jacobian smooth locus and the printed set of directions, taken at I(Y) (idealPoint). They are derived from the prime-ideal forms theorem_5_7_i_prime and theorem_5_7_ii_prime with 𝔭 = I(Y), using V_K(I(Y)) = Y for closed Y (zeroLocus_vanishingIdeal_of_isZariskiClosed). The prime-ideal forms follow from theorem_5_7_i and theorem_5_7_ii, the same statements for Mathlib's smooth locus, by Definitions 5.4(b) and 5.5. theorem_5_7_i is theorem_5_7_i_paper, the printed proof (see the proof card); theorem_5_7_ii deduces (ii) from (i) as in the last paragraph of the proof (step_dir_of_fibre). (ii) is proved but not used downstream, in the paper or in Lean; Theorem 5.8 uses (i). The earlier route's GeneralAsymptoticConormal.coordinate_axis_mem_smooth_fibre_closure proves (i) by a second, chart-based route.
Review notes
Theorem 5.8 — Involutive exclusion
Lean ():
Section5.theorem_5_8
/-- **Theorem 5.8.** Let `K` be algebraically closed of characteristic zero and `W ⊆ T*X` Zariski closed, fibre-conical and involutive. If `P ∈ K[ξ_1, …, ξ_n]` with `P(1, 0, …, 0) ≠ 0` vanishes on `W` and `W ∩ π⁻¹(H) = ∅`, then `W = ∅`. Proved along Subsection 5.3. -/ theorem theorem_5_8 (hm : 0 < m) (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (P : MvPolynomial (Fin m) k) (haxis : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P ≠ 0) (hPW : ∀ q ∈ W, MvPolynomial.eval (fun i => q (.inr i)) P = 0) (hH : W ∩ {q | q (.inl ⟨0, hm⟩) = 0} = ∅) : W = ∅ := by -- Suppose `W ≠ ∅`. by_contra hne have hW : W.Nonempty := Set.nonempty_iff_ne_empty.2 hne -- `B` is radical with a minimal prime `𝔮`; `Y_0 = V_K(𝔮)`. obtain ⟨hrad, -, -, 𝔮, h𝔮⟩ := step_minimal_primes W hclosed hcone hW obtain ⟨hprime, -, -, -, havoid⟩ := step_component hm W hclosed hcone hH 𝔮 h𝔮 -- The separator `s`, then `P` vanishes on `N(Y_0)`. obtain ⟨s, hs𝔮, hs⟩ := step_separator W 𝔮 hrad h𝔮 have hPN := step_P_vanishes_on_conormal_cone W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs P hPW -- Theorem 5.7(i) contradicts `P(1, 0, …, 0) ≠ 0`. exact haxis (step_theorem_5_7_applies hm 𝔮 hprime havoid P (step_P_vanishes_on_N 𝔮 hprime P hPN))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- **Theorem 5.7(i).** Let `Y ⊆ X = K^n` be nonempty, closed and irreducible (Definition 5.4(a)) with `Y ∩ H = ∅`, `H = {x_1 = 0}`, and let `N(Y) = ⋃_{y ∈ Y_sm} N*_y Y` be the conormal cone over the smooth locus. Every polynomial `P ∈ K[ξ_1, …, ξ_n]` vanishing on `N(Y)` satisfies `P(1, 0, …, 0) = 0`. Derived from `theorem_5_7_i_prime` with `𝔭 = I(Y)`. -/ theorem theorem_5_7_i_printed {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (Y : Set (Fin m → k)) (hne : Y.Nonempty) (hY : IsIrreducibleClosed Y) (hH : Y ∩ {y | y ⟨0, hm⟩ = 0} = ∅) (P : MvPolynomial (Fin m) k) (hP : ∀ ξ ∈ conormalCone (idealPoint Y hne hY), MvPolynomial.eval ξ P = 0) : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P = 0 := by refine theorem_5_7_i_prime hm (idealPoint Y hne hY) (fun y hy h0 => ?_) P hP change y ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) at hy rw [zeroLocus_vanishingIdeal_of_isZariskiClosed hY.1] at hy exact (Set.eq_empty_iff_forall_notMem.mp hH) y ⟨hy, h0⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirectionstheorem exists_zero_base_coordinate_of_isFibreConical {k : Type*} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (W : Set (PhaseVar m → k)) (hnonempty : W.Nonempty) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hW : FibreConicalVanishingIdeal.IsFibreConical W) (hpoisson : ∀ f ∈ MvPolynomial.vanishingIdeal k W, ∀ g ∈ MvPolynomial.vanishingIdeal k W, poissonBracket f g ∈ MvPolynomial.vanishingIdeal k W) (P : MvPolynomial (Fin m) k) (hP : fibreLift P ∈ MvPolynomial.vanishingIdeal k W) (haxis : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P ≠ 0) : ∃ q ∈ W, q (.inl ⟨0, hm⟩) = 0 := by exact exists_zero_base_coordinate_of_isFibreConical_using Stafford38.Geometry.GeneralAsymptoticConormal.coordinate_axis_mem_smooth_fibre_closure hm W hnonempty hclosed hW hpoisson P hP haxis
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.BaseRelativePoissonSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.GeneralComponentConormalContainmentSource ↗open Stafford38.Geometry.GeneralCoisotropicExclusionSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗variable {k : Type*} [Field k] [IsAlgClosed k] [CharZero k] {n : ℕ}
More Lean proofs and definitions
/-- For a prime `𝔭`, `Y = V_K(𝔭)` is nonempty, closed and irreducible (Definition 5.4(a)), with `I(Y) = 𝔭` (Nullstellensatz). -/ theorem irreducibleClosed_zeroLocus_of_isPrime {k : Type u} [Field k] [IsAlgClosed k] {m : ℕ} (𝔭 : Ideal (MvPolynomial (Fin m) k)) [h𝔭 : 𝔭.IsPrime] : ∃ (hne : (MvPolynomial.zeroLocus k 𝔭).Nonempty) (hY : IsIrreducibleClosed (MvPolynomial.zeroLocus k 𝔭)), idealPoint (MvPolynomial.zeroLocus k 𝔭) hne hY = ⟨𝔭, h𝔭⟩ := by have hI : MvPolynomial.vanishingIdeal k (MvPolynomial.zeroLocus k 𝔭) = 𝔭 := MvPolynomial.IsPrime.vanishingIdeal_zeroLocus 𝔭 have hne : (MvPolynomial.zeroLocus k 𝔭).Nonempty := by rw [Set.nonempty_iff_ne_empty] intro h0 rw [h0, MvPolynomial.vanishingIdeal_empty] at hI exact h𝔭.ne_top hI.symm have hY : IsIrreducibleClosed (MvPolynomial.zeroLocus k 𝔭) := (definition_5_4_a hne ⟨𝔭, rfl⟩).2 (by rw [hI]; exact h𝔭) exact ⟨hne, hY, PrimeSpectrum.ext hI⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirectionsComparison notes
theorem_5_8 states the theorem as printed for K algebraically closed of characteristic zero and n ≥ 1: W ⊆ T*X the zero set of an ideal of K[x, ξ], fibre-conical (IsFibreConical, Definition 5.6 by definition_5_6) and involutive (I(W) involutive, Definition 5.6), P ∈ K[ξ] with P(1, 0, …, 0) ≠ 0 vanishing on W, and W ∩ π⁻¹(H) = ∅; then W = ∅. The Lean proof of theorem_5_8 follows the 17 printed steps of Subsection 5.3, one step declaration each, and closes by calling Theorem 5.7(i) as printed (theorem_5_7_i_printed, through step_theorem_5_7_applies) for Y₀ = V_K(𝔮), which is nonempty, closed and irreducible with I(Y₀) = 𝔮 by the Nullstellensatz and Definition 5.4(a) (irreducibleClosed_zeroLocus_of_isPrime). One extra declaration, step_P_vanishes_on_N, passes from the cone over Mathlib's smooth locus to the printed cone over the Jacobian smooth locus (Definition 5.4(b)). The earlier route's GeneralCoisotropicSets.exists_zero_base_coordinate_of_isFibreConical proves the statement by the set-level route.
Review notes
5.2. Proof of Theorem 5.7
Proof of 5.7 — Proof of the asymptotic conormal theorem
Proof of Theorem 5.7
If is nonempty, closed and irreducible, then is a prime
ideal and . We will write , for the field of fractions, and for the image of in .
If is algebraic over , there exists a nonzero polynomial over
in one variable with (Def. 5.1), therefore . Since
is algebraically closed, with
, and as is prime, some factor lies
in . Thus on , and since and , we have .
In this case lies in and has the differential
at every point , so , and its coordinate vector in the dual basis . Every polynomial vanishing on therefore vanishes at , and
. So we may assume from now on that
is transcendental over .
The rest of the proof follows this rough logic. We construct a discrete valuation ring of at infinity in which vanishes, and the corresponding generic point of . The affine cone of over the projective tangent plane at has a basis of power series in that remains a basis at , and whose reduction has vanishing -coordinate. Hence the row vector , which reads off the -coordinate, satisfies , and we lift it to row vectors annihilating , with . By construction the affine part consists of conormal covectors to at , and . A polynomial vanishing on therefore vanishes at , and so .
We set .
Let be the valuation ring of Lemma 5.9. We define the column vector in homogeneous coordinates by
- for all , and ;
- is algebraic over ;
- is a finitely generated -module.
Put . This is a Noetherian local domain. It is normal, since a localization of an integrally closed domain is integrally closed [Atiyah 1969, Prop 5.12], and its dimension equals . A Noetherian local normal domain of dimension one is a discrete valuation ring [Matsumura 1986, Thm. 11.2], so is a discrete valuation ring.
It remains to check the listed properties.
- : we have , and the domain embeds into its localization .
- : since and .
- for all : since .
- : since and . , so by Theorem 2.11, i.e. for some , . Modulo this gives , so . If , then , contradicting .
(b) Since and , we have , and the residue map restricts to the quotient map . As is integral over and , the ring is integral over the image of in , which is . So every element of , and hence of its fraction field , is algebraic over .
(c) is a finitely generated -algebra, so is a finite -module, and since is a localization of [hartshorne, Prop II.8.2A]. ∎
By Lemma 5.10 there exists a discrete valuation ring with for all , and . Let be a generator of , then every nonzero can be written as with and a unit of - we say is the valuation of in . We denote the residual field with .
Let be the completion of . Since and is of characteristic zero, there exists a coefficient field of containing and mapping isomorphically onto [Matsumura 1986, Thm. 28.3]; choosing one gives us an isomorphism under which corresponds to the formal variable . Since the natural map is injective ( [Atiyah 1969, Corr. 10.20]) and extends to an embedding , we can identify elements of with Laurent series. The residue of is then just the constant term of its series; in particular
Let and . since are elements of , so that in
Dividing by , we define in affine coordinates. Its coordinates are the images of under . Since ring homomorphisms commute with polynomial evaluation and this map is injective, for every polynomial we have
Let for a polynomial . We define to be the tangent space of at the generic point ,
Since and , the field is generated over by the finitely many with , so some subset of them is a transcendence basis of [Lang 2012, Thm. VIII.1.1]. As and are algebraic (Lemma 5.10(b)), so is ; hence is also a transcendence basis of . Since has characteristic zero, a transcendence basis of is a differential basis, i.e. its differentials form a -basis of [Matsumura 1986, Thm. 26.5]. ∎
We write for all . The derivative with respect to is then . This is a -derivation of . By Lemma 5.11 there exist such that form a basis of . Let be the dual -derivations of this basis (Definition 5.2(c)). We apply coefficientwise to and define . Since , the reduction is the identity on the rows , and since is constant, . Rows and of have order at least and , since does not differentiate with respect to t.
Let be the submatrix of consisting of the rows
and all columns. Its reduction is the identity
matrix, so is invertible over . Hence there exists a unique vector
with
. We define , so
that for .
Its th entry is . Here
has order exactly (since the coefficient
of is nonzero in characteristic zero),
while has order at least , because every entry of
row of has order at least (since
has order ). Hence . Let be the minimal order
among the entries of ; then , and is a vector
over with and . In row , both
and have order at least , so , so
. Finally and
(as ), so .
We now show that span all of : Let , an matrix over . Its reduction has full column rank: in row only is nonzero (); in rows the columns form an identity block while vanishes there; and , i.e. we have independent reduced column vectors. This also translates to the column vectors of themselves: Since some minor of is nonzero, the corresponding minor of is a unit of ; hence the columns of are also linearly independent over , so they span a -dimensional subspace there. Furthermore, every column of has a vanishing entry for the -coordinate: (as ), the entries have order at least , hence , and . Thus the span of is contained in and also has dimension .
To show that the columns of really span all of , we must still show that a) the dimensions match and b) the columns actually lie in . The following Lemma will help us with that.
- ; hence .
- For every -derivation of the column lies in ; moreover .
The second fundamental exact sequence for is
(b) Let and . By (5.1), , and vanishes on , so
∎
The column vectors , and are of the form , where
is a -derivation of (the extended
coefficientwise, and ). By Lemma 5.12(b) they lie in . Since
and , is a
-linear combination, hence also lies in . By Lemma 5.12(a), , so the independent columns of must span .
Let be the row vector reading off the -coordinate; since , we know that . We can lift to a row vector with and : Since has full column rank , there is a set of row indices for which the square submatrix of is invertible; then is invertible over , since its determinant has a nonzero constant term. Let be the complementary set of row indices, and define by
Since the columns of span , annihilates , so by Lemma 5.13 its affine part lies in , and since , .
Finally, let vanish on . We will show : Since , we can write
By statement (i), every polynomial vanishing on vanishes at . We use this to show . Recall that is the Zariski closure of the set of classes with , and that a point of lies in the closure of a set if and only if every homogeneous polynomial vanishing on also vanishes at that point. So let be a homogeneous polynomial vanishing on all classes with . Since also by homogeneity, vanishes on every nonzero (and at if ), hence on ; if , is a constant that vanishes somewhere (such classes exist: if , the first coordinate function vanishes on but not at , contrary to (i)), so . By (i), , i.e. vanishes at . As was arbitrary, . ∎
Lean: Section5.theorem_5_7_i_paper;
Section5.theorem_5_7_i;
Section5.lemma_5_9;
Section5.lemma_5_10;
Section5.lemma_5_11;
Section5.lemma_5_12_a_bound;
Section5.lemma_5_12_a_cone;
Section5.lemma_5_12_b;
Section5.lemma_5_13
/-- **Theorem 5.7(i).** If `Y = V(𝔭)` is nonempty, irreducible and avoids `H = {x_1 = 0}`, every polynomial vanishing on the conormal cone `N(Y)` over the smooth locus vanishes at `(1, 0, …, 0)`. Proved paragraph by paragraph in `theorem_5_7_i_paper`. -/ theorem theorem_5_7_i {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := theorem_5_7_i_paper hm I havoid
/-- **Theorem 5.7(i), along the paper's proof.** If `Y = V(𝔭)` is nonempty, irreducible and avoids `H = {x_1 = 0}`, every polynomial vanishing on the conormal cone `N(Y)` over the smooth locus vanishes at `(1, 0, …, 0)`. -/ theorem theorem_5_7_i_paper {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := by classical intro Pol hPol by_cases halg : IsAlgebraic k (componentCoordinate I ⟨0, hm⟩) · -- First paragraph: `x̄_1` algebraic, so `x_1 − c ∈ 𝔭` with `c ≠ 0`, and -- `(1, 0, …, 0) = dx_1` is conormal at a smooth point of `Y`, so lies in `N(Y)`. obtain ⟨-, -, -, hN⟩ := step_coordinate_algebraic hm I havoid halg exact hPol _ hN -- `x̄_1` transcendental. `Y ∩ H = ∅` makes `x̄_1` a unit of `R`. have hunit := exists_coordinate_inverse_of_avoidance I.asIdeal ⟨0, hm⟩ havoid -- Lemmas 5.9–5.10: the discrete valuation ring `V` of `F` at infinity, the -- column `q_i = x̄_i / x̄_j` and `κ = V/𝔪_V`, read in `V̂ ≅ κ[[t]]`. obtain ⟨w, hfin⟩ := Stafford38.Geometry.GeneralDivisorialVisibleFrame.generalDivisorialVisibleFrameWithKaehlerFinite hm I hunit halg let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField I) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : IsLocalRing V := W.place.isDiscrete.toIsLocalRing letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra I ⟨0, hm⟩ W letI : CharZero κ := charZero_of_injective_algebraMap (FaithfulSMul.algebraMap_injective k κ) -- Paragraph "`V̂ ≅ κ[[t]]`": `F ↪ κ((t))`, and the residue is the constant term. obtain ⟨-, -, -, -, hres, -, -, -⟩ := step_completion hm I w let q : Fin (m + 1) → PowerSeries κ := fun a ↦ retainedToCompletedPowerSeries W (C.q a) let axis : Fin (m + 1) := Fin.succ ⟨0, hm⟩ -- Paragraph "`a = ord q_0`, `b = ord q_1`": `0 < a < b`, `j ∉ {0, 1}`. obtain ⟨a, b, u₀, u₁, ha, hab, hq0, hu₀, hq1, -, hqj, -, -⟩ := step_orders hm I w hunit have hq0ne : q 0 ≠ 0 := by intro h apply hu₀ have h' : (PowerSeries.X : PowerSeries κ) ^ a * u₀ = 0 := hq0.symm.trans h rw [mul_eq_zero] at h' rcases h' with h' | h' · exact absurd h' (pow_ne_zero a PowerSeries.X_ne_zero) · rw [h', map_zero] -- Paragraph defining `y`, equation (generic): `f(y) = 0` for `f ∈ 𝔭`. have hbase : ∀ f ∈ I.asIdeal, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) (dehomogenizedPoint (laurentColumn q)) f = 0 := fun f hf ↦ (step_generic_point hm I w f).2 hf -- Lemma 5.11: `d(q_{i_l}(0))`, `i_l ∉ {0, 1, j}`, form a basis of `Ω_{κ/K}`. obtain ⟨-, -, -, d, rows, hinj, -, hli, hspan, hd⟩ := witness_lemma_5_11 hm I w hunit have hresRows : (fun l ↦ KaehlerDifferential.D k κ (PowerSeries.constantCoeff (q (rows l)))) = fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l))) := by funext l rw [hres] rw [← hresRows] at hli hspan let rowsE : Fin d ↪ Fin (m + 1) := ⟨rows, hinj⟩ -- Paragraph after Lemma 5.11: dual derivations `D_k` and the matrix `Z(t)`. obtain ⟨D, -, -, hminor, hZj, hZ0, hZ1⟩ := step_dual_derivations (k := k) hm q rowsE hli hspan C.chart hqj a b u₀ u₁ hq0 hq1 let Z := coefficientwiseTangentMatrix q D -- Paragraph defining `λ` (unique), `w = q' − Zλ`, `c ≤ a − 1` and `τ = t^{-c} w`. obtain ⟨lambda, c, tau, -, -, -, -, -, -, -, hfactor, hprim, htaurows, -, htau1, htauj⟩ := step_lambda_w_tau hm q Z rowsE C.chart a b u₀ u₁ hqj hZj ha hab hq0 hu₀ hZ0 hq1 hZ1 hminor -- Paragraph "`M(0)` has full column rank" for `M = (q, Z, τ)`: a unit minor -- `M_ι`, independent columns over `κ((t))`, vanishing `1`-coordinates of `M(0)`. obtain ⟨hinjM, -, hliM, haxis⟩ := step_M_minor hm q Z tau rowsE C.chart b u₁ hqj hZj htauj htaurows hminor hprim (ha.trans hab) hq1 hZ1 htau1 -- Lemma 5.12(a): `dim Θ_y = dim_F Ω_{F/K} ≤ d' + 1`, the bound by Lemma 5.10(c), -- the second fundamental exact sequence for `V → κ` and Nakayama. obtain ⟨⟨hdim, -, hfieldRank, -⟩, -, -⟩ := witness_lemma_5_12_5_13_paper hm I w hfin have hbound : Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I.asIdeal) ≤ d + 1 := by rw [hd] exact hdim.trans_le hfieldRank -- Paragraph "the columns of `M` span `𝒞`" (Lemmas 5.12(a), 5.12(b)). obtain ⟨-, hspanC⟩ := step_M_spans_cone I.asIdeal q hq0ne hbase D Z rfl lambda c tau hfactor hliM hbound -- Paragraph lifting `u = (0, 1, 0, …, 0)`: `uM(0) = 0`, an invertible `M_ι`, -- and `ℓ_υ = u_υ`, `ℓ_ι = −u_υ M_υ M_ι⁻¹` with `ℓM = 0`, `ℓ(0) = u`. have hu : rowMul (axisRow (k := κ) axis) (residueMatrix (formalTangentMatrix q Z tau)) = 0 := by funext col simp only [rowMul, axisRow, residueMatrix, ite_mul, one_mul, zero_mul, Finset.sum_ite_eq', Finset.mem_univ, ite_true, Pi.zero_apply] exact haxis col obtain ⟨rowsι, -, -, -, -, hell, hellres⟩ := step_lift_ell_minor (formalTangentMatrix q Z tau) hinjM (axisRow (k := κ) axis) hu -- Paragraph "`ℓ` annihilates `𝒞`" (Lemma 5.13): `ℓ̃ ∈ span {df_y}`. obtain ⟨hconormal, htail⟩ := step_ell_conormal hm I.asIdeal q hq0ne _ hspanC _ hell hellres -- Final paragraph: `P(ℓ̃(0)) = 0` with `ℓ̃(0) = (1, 0, …, 0)`. have hzero := step_polynomial_limit I Pol hPol _ hbase _ hconormal rw [htail, eval₂_extensionAxis_eq_algebraMap_eval_axis (K := κ) hm Pol] at hzero exact (FaithfulSMul.algebraMap_injective k κ) (hzero.trans (map_zero _).symm)
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.FiniteGradientResidueExtensionSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralConstantCoordinateAxisSource ↗open Stafford38.Geometry.GeneralCoordinateAvoidanceSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗universe u
More Lean proofs and definitions
/-- **First paragraph of the proof of Theorem 5.7 (`x̄_1` algebraic).** A nonzero `g` with `g(x̄_1) = 0` gives `g(x_1) ∈ 𝔭`; over the algebraically closed `K`, `g(x_1) = λ ∏ (x_1 − c_i)` with `λ ∈ K^×`, and since `𝔭` is prime some factor `x_1 − c` lies in `𝔭`. Thus `x_1 = c` on `Y`, and `Y ≠ ∅`, `Y ∩ H = ∅` give `c ≠ 0`. The equation `f = x_1 − c` has `df_y = dx_1` at every smooth point `y`, and `Y_sm ≠ ∅`, so `(1, 0, …, 0) ∈ N(Y)`. -/ theorem step_coordinate_algebraic {k : Type u} [Field k] [IsAlgClosed k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) (halg : IsAlgebraic k (componentCoordinate I ⟨0, hm⟩)) : ∃ c : k, MvPolynomial.X ⟨0, hm⟩ - MvPolynomial.C c ∈ I.asIdeal ∧ c ≠ 0 ∧ (fun i : Fin m ↦ if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ smoothConormalFibreProjection I.asIdeal := by classical let i₀ : Fin m := ⟨0, hm⟩ let x₁ : MvPolynomial (Fin m) k := MvPolynomial.X i₀ -- A nonzero `g` with `g(x̄_1) = 0`, hence `g(x_1) ∈ 𝔭`. obtain ⟨g, hg0, hgx⟩ := halg have hgmem : Polynomial.aeval x₁ g ∈ I.asIdeal := by have hcomp : componentCoordinate I i₀ = ((IsScalarTower.toAlgHom k (MvPolynomial (Fin m) k ⧸ I.asIdeal) (FractionRing (MvPolynomial (Fin m) k ⧸ I.asIdeal))).comp (Ideal.Quotient.mkₐ k I.asIdeal)) x₁ := rfl rw [hcomp, Polynomial.aeval_algHom_apply] at hgx rw [← Ideal.Quotient.eq_zero_iff_mem] apply IsFractionRing.injective (MvPolynomial (Fin m) k ⧸ I.asIdeal) (FractionRing (MvPolynomial (Fin m) k ⧸ I.asIdeal)) rw [map_zero, ← hgx] rfl -- `K` algebraically closed: `g(x_1) = λ ∏ (x_1 − c_i)`, `λ ∈ K^×`. have hfac := Polynomial.C_leadingCoeff_mul_prod_multiset_X_sub_C (IsAlgClosed.card_roots_eq_natDegree (p := g)) have hlc : g.leadingCoeff ≠ 0 := Polynomial.leadingCoeff_ne_zero.mpr hg0 rw [← hfac, map_mul, Polynomial.aeval_C, map_multiset_prod, Multiset.map_map] at hgmem -- `𝔭` is prime, so some factor `x_1 − c` lies in `𝔭`. rcases I.isPrime.mem_or_mem hgmem with hC | hprod · exact absurd (I.isPrime.ne_top ((Ideal.eq_top_iff_one _).mpr (by have := Ideal.mul_mem_left _ (MvPolynomial.C g.leadingCoeff⁻¹) hC simpa [MvPolynomial.algebraMap_eq, ← MvPolynomial.C_mul, inv_mul_cancel₀ hlc] using this))) (by simp) obtain ⟨_, hmem, hc⟩ := (I.isPrime.multiset_prod_mem_iff_exists_mem _).mp hprod obtain ⟨c, -, rfl⟩ := Multiset.mem_map.mp hmem have hc' : x₁ - MvPolynomial.C c ∈ I.asIdeal := by simpa [x₁, Polynomial.aeval_X, MvPolynomial.algebraMap_eq] using hc -- `x_1 = c` on `Y`; `Y ≠ ∅` and `Y ∩ H = ∅` give `c ≠ 0`. obtain ⟨y, hy⟩ := exists_mem_zeroLocus_of_isPrime I have hyc : y i₀ = c := by have := hy _ hc' simpa [x₁, sub_eq_zero] using this have hc0 : c ≠ 0 := hyc ▸ havoid y hy -- `f = x_1 − c ∈ 𝔭` has `df_y = dx_1` at a smooth point `y`, so `dx_1 ∈ N*_y Y ⊆ N(Y)`. obtain ⟨y', hy'⟩ := exists_smoothAffinePoint I refine ⟨c, hc', hc0, (definition_5_4_d I.asIdeal _).2 ⟨y', hy', ?_⟩⟩ have hdf : differentialAt y' (x₁ - MvPolynomial.C c) = fun i : Fin m ↦ if i = i₀ then (1 : k) else 0 := by funext i by_cases hi : i = i₀ · subst hi; simp [differentialAt, x₁] · simp [differentialAt, x₁, hi, MvPolynomial.pderiv_X, Ne.symm hi] rw [← hdf] exact Submodule.subset_span ⟨⟨_, hc'⟩, rfl⟩
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.CoisotropicTranslation/-- The smooth locus `Y_sm` of a nonempty `Y = V(𝔭)` is nonempty (it is dense in `Y`, Definition 5.4(b)). -/ theorem exists_smoothAffinePoint {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ} (I : PrimeSpectrum (MvPolynomial (Fin n) k)) : ∃ y : Fin n → k, SmoothAffinePoint I.asIdeal y := by classical by_contra hnone push Not at hnone obtain ⟨y, hy⟩ := exists_mem_zeroLocus_of_isPrime I have hmem : (Sum.elim y 0 : Fin n ⊕ Fin n → k) ∈ equationConormalClosure I.asIdeal := by intro f hf refine hf _ ⟨fun g hg ↦ by simpa using hy g hg, ?_⟩ have h0 : coordinateCovector (fun i : Fin n ↦ (Sum.elim y 0 : Fin n ⊕ Fin n → k) (.inr i)) = 0 := by apply LinearMap.ext intro v simp [coordinateCovector_apply] rw [h0] exact Submodule.zero_mem _ rw [← equationConormalClosure_smoothAffine_eq I.asIdeal I.isPrime] at hmem have hempty : restrictedEquationConormalLocus I.asIdeal {y | SmoothAffinePoint I.asIdeal y} = ∅ := by ext q simp only [Set.mem_empty_iff_false, iff_false] exact fun hq ↦ hnone _ hq.2 have h1 : (1 : MvPolynomial (Fin n ⊕ Fin n) k) ∈ MvPolynomial.vanishingIdeal k (restrictedEquationConormalLocus I.asIdeal {y | SmoothAffinePoint I.asIdeal y}) := by rw [hempty, MvPolynomial.mem_vanishingIdeal_iff] intro x hx exact absurd hx (Set.notMem_empty x) simpa using hmem 1 h1
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.CoisotropicTranslation/-- `Y = V(𝔭)` is nonempty for a prime `𝔭` (Nullstellensatz). -/ theorem exists_mem_zeroLocus_of_isPrime {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ} (I : PrimeSpectrum (MvPolynomial (Fin n) k)) : ∃ y : Fin n → k, y ∈ MvPolynomial.zeroLocus k I.asIdeal := by obtain ⟨M, hMmax, hPM⟩ := Ideal.exists_le_maximal I.asIdeal I.isPrime.ne_top obtain ⟨y, hMy⟩ := MvPolynomial.isMaximal_iff_eq_vanishingIdeal_singleton.mp hMmax refine ⟨y, fun f hf ↦ ?_⟩ have hfM : f ∈ M := hPM hf rw [hMy, MvPolynomial.mem_vanishingIdeal_singleton_iff] at hfM exact hfM
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
open Stafford38.Geometry.AffineConormalClosureSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗open Stafford38.Geometry.CoisotropicTranslation/-- The general divisorial producer with its actual Lane-C residue algebraicity retained, together with Lemma 5.10(c): `Ω_{V/k}` is a finitely generated `V`-module for the place `V` of the witness (used for Nakayama's lemma in the proof of Lemma 5.12(a)). The algebraicity is for the same normalized column and the same visible frame, under the ground algebra structure induced by the retained coefficient map. -/ theorem generalDivisorialVisibleFrameWithKaehlerFinite {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) (htrans : Transcendental k (componentCoordinate P ⟨0, hm⟩)) : ∃ w : GeneralDivisorialVisibleFrameWitness hm P, letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k]) := by let i : Fin m := ⟨0, hm⟩ let K := ComponentFractionField P obtain ⟨E, V, hEV, hVdvr, hxV, htransE, hxm, hEfin, hkaehler, halgAll, j, hj, hqV⟩ := Stafford38.Geometry.LaneC.divisorialVisibleFrameExistenceWithChart k K m (componentCoordinate P) i (componentCoordinate_adjoin_eq_top P) htrans letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing letI : Algebra E V.toSubring := (Stafford38.Geometry.LaneC.coeffHom E V hEV).toAlgebra letI : Algebra k V.toSubring := (Stafford38.Geometry.LaneC.groundHom E V hEV).toAlgebra letI : Algebra V.toSubring K := V.toSubring.subtype.toAlgebra letI : IsScalarTower k V.toSubring K := IsScalarTower.of_algebraMap_eq fun c => by change algebraMap k K c = (algebraMap k E c : K) exact IsScalarTower.algebraMap_apply k E K c letI : Module.Finite V.toSubring (Ω[V.toSubring⁄k]) := hkaehler let W : Data k K (componentCoordinate P i) := retainedDataOfValuation E V hEV hVdvr (componentCoordinate P i) hxV htransE hxm hEfin letI : Algebra (CoordinateZeroLocalRing W.coefficientField) K := W.ambientAlgebra letI : IsScalarTower W.coefficientField (CoordinateZeroLocalRing W.coefficientField) K := W.coefficientTower -- The column `q_a = x̄_a / x̄_j` with the chart `j` of Lemma 5.9. obtain ⟨chart, qraw, scale, hscale, hchartRaw, hqraw⟩ : ∃ (chart : Fin (m + 1)) (q : Fin (m + 1) → V) (scale : K), scale ≠ 0 ∧ q chart = 1 ∧ ∀ a, (q a : K) = scale * componentProjectivePoint P a := by simp only [componentProjectivePoint_eq_finCases] refine ⟨j, fun a ↦ ⟨_, hqV a⟩, (Fin.cases 1 (componentCoordinate P) j : K)⁻¹, inv_ne_zero hj, Subtype.ext (div_self hj), fun a ↦ ?_⟩ change (Fin.cases 1 (componentCoordinate P) a : K) / _ = _ rw [div_eq_inv_mul] let q : Fin (m + 1) → V.toSubring := fun a => ⟨qraw a, (qraw a).property⟩ have hchart : q chart = 1 := by apply Subtype.ext exact congrArg Subtype.val hchartRaw have hq : ∀ a, (q a : K) = scale * componentProjectivePoint P a := by intro a exact hqraw a have hq0 : q 0 ≠ 0 := by intro hzero apply hscale have hz : ((q 0 : V.toSubring) : K) = 0 := congrArg (fun z : V.toSubring => (z : K)) hzero have h := hq 0 rw [hz] at h simpa only [componentProjectivePoint_eq_finCases, Fin.cases_zero, mul_one] using h.symm let xV : V.toSubring := ⟨componentCoordinate P i, hxV⟩ have hratioV : q (Fin.succ i) = q 0 * xV := by apply Subtype.ext change (q (Fin.succ i) : K) = (q 0 : K) * componentCoordinate P i rw [hq, hq] simp only [componentProjectivePoint_eq_finCases, Fin.cases_zero, Fin.cases_succ, mul_one] let coeff : k →+* V := (relativeCoefficientMap W.coefficientField W.place).comp (algebraMap k W.coefficientField) have hcoeff : W.place.valuation.toSubring.subtype.comp coeff = algebraMap k K := by ext c change ((relativeCoefficientMap W.coefficientField W.place (algebraMap k W.coefficientField c) : V) : K) = algebraMap k K c calc ((relativeCoefficientMap W.coefficientField W.place (algebraMap k W.coefficientField c) : V) : K) = algebraMap W.coefficientField K (algebraMap k W.coefficientField c) := DFunLike.congr_fun (relativeCoefficientMap_commutes W.coefficientField W.place) (algebraMap k W.coefficientField c) _ = algebraMap k K c := IsScalarTower.algebraMap_apply k W.coefficientField K c have hLaneCoeff : W.place.valuation.toSubring.subtype.comp (Stafford38.Geometry.LaneC.groundHom E V hEV) = algebraMap k K := by ext c change ((algebraMap k E c : E) : K) = algebraMap k K c exact IsScalarTower.algebraMap_apply k E K c have hCoeffEq : coeff = Stafford38.Geometry.LaneC.groundHom E V hEV := by apply RingHom.ext intro c apply Subtype.ext change ((coeff c : V) : K) = ((Stafford38.Geometry.LaneC.groundHom E V hEV c : V) : K) calc ((coeff c : V) : K) = algebraMap k K c := by have hc := DFunLike.congr_fun hcoeff c change ((coeff c : V) : K) = algebraMap k K c at hc exact hc _ = ((Stafford38.Geometry.LaneC.groundHom E V hEV c : V) : K) := by have hc := DFunLike.congr_fun hLaneCoeff c change ((Stafford38.Geometry.LaneC.groundHom E V hEV c : V) : K) = algebraMap k K c at hc exact hc.symm have hq0nonunit : ¬ IsUnit (q 0) := by obtain ⟨g, hg⟩ := hunit let F := K letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra letI : IsScalarTower W.coefficientField (CoordinateZeroLocalRing W.coefficientField) F := W.coefficientTower let phi : MvPolynomial (Fin m) k →+* F := (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) F).comp (Ideal.Quotient.mk P.asIdeal) have hphiC : phi.comp MvPolynomial.C = algebraMap k F := by ext c exact IsScalarTower.algebraMap_apply k (MvPolynomial (Fin m) k ⧸ P.asIdeal) F c have hphiX : ∀ j, phi (MvPolynomial.X j) = componentCoordinate P j := by intro j rfl have hpoly : phi g = MvPolynomial.eval₂ (algebraMap k F) (fun j ↦ componentCoordinate P j) g := by rw [MvPolynomial.map_mvPolynomial_eq_eval₂ phi g] change MvPolynomial.eval₂Hom (phi.comp MvPolynomial.C) (fun j ↦ phi (MvPolynomial.X j)) g = MvPolynomial.eval₂Hom (algebraMap k F) (fun j ↦ componentCoordinate P j) g apply MvPolynomial.eval₂Hom_congr hphiC · funext j exact hphiX j · rfl have hinverse : componentCoordinate P i * MvPolynomial.eval₂ (algebraMap k F) (fun j ↦ componentCoordinate P j) g = 1 := by have hzero : phi (MvPolynomial.X i * g - 1) = 0 := by have hmk : Ideal.Quotient.mk P.asIdeal (MvPolynomial.X i * g - 1) = 0 := Ideal.Quotient.eq_zero_iff_mem.mpr hg simpa [phi] using (congrArg (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) F) hmk) rw [map_sub, map_mul, map_one, sub_eq_zero, hphiX, hpoly] at hzero exact hzero exact normalized_denominator_nonunit_of_polynomial_inverse (V := W.place.valuation) (coeff := coeff) hcoeff (x := fun j ↦ componentCoordinate P j) (qzero := q 0) (q := fun j ↦ q (Fin.succ j)) (scale := scale) (i := i) (parameter := W.place.parameter) (g := g) (by simpa only [componentProjectivePoint_eq_finCases, Fin.cases_zero, mul_one] using hq 0) (by intro j; simpa only [componentProjectivePoint_eq_finCases, Fin.cases_succ] using hq (Fin.succ j)) W.parameter_eq_coordinate W.place.parameter_nonunit hinverse let A_lane : Algebra k (ResidueField V) := ((residue V).comp (Stafford38.Geometry.LaneC.groundHom E V hEV)).toAlgebra let A_retained : Algebra k (ResidueField V) := retainedResidueGroundAlgebra P i W have hresMapEq : (residue V).comp (Stafford38.Geometry.LaneC.groundHom E V hEV) = (residue V).comp (retainedComponentCoefficientMap P i W) := by change (residue V).comp (Stafford38.Geometry.LaneC.groundHom E V hEV) = (residue V).comp coeff rw [← hCoeffEq] have hresAlgEq : A_lane = A_retained := by change ((residue V).comp (Stafford38.Geometry.LaneC.groundHom E V hEV)).toAlgebra = ((residue V).comp (retainedComponentCoefficientMap P i W)).toAlgebra exact congrArg RingHom.toAlgebra hresMapEq have halgLane : letI : Algebra k (ResidueField V) := A_lane Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun j : Fin m ↦ residue V (q (Fin.succ j))) : IntermediateField k (ResidueField V)) (ResidueField V) := by letI : Algebra k (ResidueField V) := A_lane have hqFinCases : ∀ a, (q a : K) = scale * Fin.cases 1 (fun i ↦ componentCoordinate P i) a := by intro a have hpoint : componentProjectivePoint P a = Fin.cases 1 (fun i ↦ componentCoordinate P i) a := congrFun (componentProjectivePoint_eq_finCases P) a calc (q a : K) = scale * componentProjectivePoint P a := hq a _ = scale * Fin.cases 1 (fun i ↦ componentCoordinate P i) a := by rw [hpoint] exact halgAll scale q hqFinCases ⟨chart, hchart⟩ hq0nonunit have halg := algebraic_adjoin_transfer_of_algebra_eq A_lane A_retained hresAlgEq (Set.range fun j : Fin m ↦ residue V (q (Fin.succ j))) halgLane have hchart_ne : chart ≠ 0 := by intro hzero apply hq0nonunit rw [← hzero, hchart] exact isUnit_one obtain ⟨j₀, rfl⟩ := Fin.exists_succ_eq_of_ne_zero hchart_ne let t := Stafford38.Geometry.CompletedDVRPowerSeries.chosenUniformizer V.toSubring have ht : Irreducible t := Stafford38.Geometry.CompletedDVRPowerSeries.chosenUniformizer_irreducible V.toSubring have hq0max : q 0 ∈ maximalIdeal V.toSubring := by apply (IsLocalRing.mem_maximalIdeal (q 0)).2 exact mem_nonunits_iff.mpr hq0nonunit have hxV_ne : xV ≠ 0 := by intro hx0 apply htransE have hx0K : componentCoordinate P i = 0 := congrArg Subtype.val hx0 exact hx0K ▸ isAlgebraic_zero obtain ⟨a, e, b, u₀, ur, u₁, ha, he, hb, hab, hq0factor, hparameterFactor, hu₁, hq1factor⟩ := exists_uniformizer_strict_orderGap t ht (q 0) xV (q (Fin.succ i)) hq0 hxV_ne hq0max hxm hratioV let Q : Fin m → V.toSubring := fun j => q (Fin.succ j) have hQj₀ : Q j₀ = 1 := hchart have hq0frame : q 0 = t ^ a * (u₀ : V.toSubring) := by simpa [mul_comm] using hq0factor have hq1frame : q (Fin.succ i) = t ^ (a + e) * (u₁ : V.toSubring) := by simpa [hb, mul_comm] using hq1factor obtain ⟨D, hD0, hD1, hDt, hDu, hDw, hDQ, hDa, hDe, hDj, hDW⟩ := exists_visibleDivisorFrame_of_kaehler_image (k := k) (F := K) (q 0) (q (Fin.succ i)) t (u₀ : V.toSubring) (u₁ : V.toSubring) Q a e j₀ ht.maximalIdeal_eq ht.ne_zero u₀.isUnit (Nat.one_le_iff_ne_zero.mpr ha.ne') (Nat.one_le_iff_ne_zero.mpr he.ne') hq0frame hq1frame hQj₀ halgLane have hretMap : @algebraMap k (ResidueField V) _ _ A_retained = (residue V).comp (retainedComponentCoefficientMap P i W) := by exact RingHom.algebraMap_toAlgebra _ let C : GeneralDivisorialVisibleFrameColumn hm P := { W := W chart := Fin.succ j₀ q := q scale := scale scale_ne := hscale chart_one := hchart q0_ne := hq0 q_commonScale := hq q_parameter := by apply Subtype.ext have hv := congrArg Subtype.val hratioV change (q (Fin.succ ⟨0, hm⟩) : K) = (q 0 : K) * componentCoordinate P ⟨0, hm⟩ at hv calc (q (Fin.succ ⟨0, hm⟩) : K) = (q 0 : K) * componentCoordinate P ⟨0, hm⟩ := hv _ = (q 0 : K) * (W.place.parameter : K) := by rw [W.parameter_eq_coordinate] groundCoeff := Stafford38.Geometry.LaneC.groundHom E V hEV groundCoeff_commutes := hLaneCoeff groundCoeff_eq_retained := hCoeffEq.symm groundTower := by let V' := W.place.valuation.toSubring letI : Algebra k V' := (Stafford38.Geometry.LaneC.groundHom E V hEV).toAlgebra letI : Algebra V' (ComponentFractionField P) := V'.subtype.toAlgebra exact IsScalarTower.of_algebraMap_eq fun c => by exact (DFunLike.congr_fun hLaneCoeff c).symm } let core : GeneralDivisorialVisibleFrameCore hm P C := { D := D D_Q0 := hD0 D_Q1 := hD1 D_Q := fun j => by rw [hDQ] D_maximalIdeal := by rw [hDt] exact Stafford38.Geometry.CompletedDVRPowerSeries.maximalIdeal_eq_span_chosenUniformizer V.toSubring D_uniformizer := hDt D_w_unit := by rw [hDw] exact u₁.isUnit } let S : GeneralDivisorialVisibleFrameSourceImage hm P C core := ⟨hDW⟩ let F : GeneralDivisorialVisibleFrameDifferential hm P C := ⟨core, S⟩ exact ⟨{ column := C differential := F retainedGroundMap := RingHom.algebraMap_toAlgebra _ halg := halg }, hkaehler⟩
Variables and notation
Source ↗open IsLocalRing PolynomialSource ↗open Stafford38Source ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ComponentProjectiveClosureNormalizationSource ↗open Stafford38.Geometry.ComponentProjectiveOrderSource ↗open Stafford38.Geometry.CompletedDVRPowerSeriesSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.ExactVisibleDivisorFrameInterfaceSource ↗open Stafford38.Geometry.KaehlerDVRVisibilitySource ↗open Stafford38.Geometry.ProjectiveDivisorOrderGapSource ↗open Stafford38.Geometry.ProjectiveValuationNormalizationSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedDVRSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.DivisorTangentLatticeSource ↗universe u
/-- The place at infinity of the paper's proof of Theorem 5.7: the valuation ring and chart of Lemma 5.9, the discrete valuation ring `V = A_{𝔭_D}` of Lemma 5.10, and a coefficient field `E` over which `κ` is finite (the input to the coefficient field of the completion, Matsumura Thm. 28.3). -/ theorem divisorialVisibleFrameExistenceWithChart (k K : Type u) [Field k] [CharZero k] [Field K] [Algebra k K] (r : ℕ) (y : Fin r → K) (i : Fin r) (hgen : IntermediateField.adjoin k (Set.range y) = ⊤) (hx : Transcendental k (y i)) : ∃ (E : IntermediateField k K) (V : ValuationSubring K) (hEV : ∀ z : E, (z : K) ∈ V.toSubring) (hVdvr : IsDiscreteValuationRing V.toSubring) (hxV : y i ∈ V.toSubring), letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra letI : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra Transcendental E (y i) ∧ (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ∧ Module.Finite E (ResidueField V.toSubring) ∧ Module.Finite V.toSubring (Ω[V.toSubring⁄k]) ∧ (∀ (s : K) (q : Fin (r + 1) → V.toSubring), (∀ a, (q a : K) = s * Fin.cases 1 y a) → (∃ j, q j = 1) → ¬ IsUnit (q 0) → Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun j : Fin r ↦ residue V.toSubring (q (Fin.succ j))) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) ∧ -- The chart `j` of Lemma 5.9, with `q_a = x̄_a / x̄_j ∈ V` (Lemma 5.10(a)). ∃ j, (Fin.cases 1 y j : K) ≠ 0 ∧ ∀ a, (Fin.cases 1 y a : K) / Fin.cases 1 y j ∈ V.toSubring := by -- Lemma 5.9. obtain ⟨W, j, hkW, -, hxW, hj, hqW⟩ := Stafford38.Paper.Section5.lemma_5_9 y i hx -- Lemma 5.10. obtain ⟨L⟩ := Stafford38.Paper.Section5.lemma_5_10 y i hgen hx W j hkW hxW hj hqW let A := L.A let p := L.p let V : ValuationSubring K := L.V have hVdvr : IsDiscreteValuationRing V.toSubring := L.isDVR letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing have hAV : ∀ a : A, (a : K) ∈ V.toSubring := fun a => ofField_mem_algebraMap_v A p L.place a have hpV : ∀ a : A, (⟨a, hAV a⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ↔ a ∈ p := fun a => ofField_mem_maximalIdeal_iff_v A p L.place a have hsurj : ∀ z : ResidueField V.toSubring, ∃ (a b : A), b ∉ p ∧ z = residue V.toSubring ⟨a, hAV a⟩ / residue V.toSubring ⟨b, hAV b⟩ := fun z => ofField_residue_surjective_v A p L.place z have hy0 : y i ≠ 0 := fun h => hx (h ▸ isAlgebraic_zero) have hxV : y i ∈ V.toSubring := hAV ⟨y i, L.x_mem⟩ have hxm : (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := (hpV ⟨y i, L.x_mem⟩).2 L.x_mem_p obtain ⟨E, hEV, hEfin⟩ := stage5_of_verified A p V hAV hpV hsurj letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra have htrans : Transcendental E (y i) := stage6_transcendental_of_mem_maximalIdeal E V hEV (y i) hy0 hxV hxm have hk : ∀ c : k, algebraMap k K c ∈ V.toSubring := fun c => hAV (algebraMap k A c) let cod : Algebra k V.toSubring := ((algebraMap k K).codRestrict _ hk).toAlgebra let ground : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra have hcod_ground : cod = ground := by apply Algebra.algebra_ext intro c apply Subtype.ext rfl letI : Algebra k V.toSubring := ground letI : IsScalarTower k V.toSubring K := IsScalarTower.of_algebraMap_eq fun c => by change algebraMap k K c = algebraMap k K c rfl have hkaehler : Module.Finite V.toSubring (Ω[V.toSubring⁄k]) := by have h := L.kaehler_finite change (letI : Algebra k V.toSubring := cod; Module.Finite V.toSubring (Ω[V.toSubring⁄k])) at h rw [hcod_ground] at h exact h -- The normalized column `q_a = x̄_a / x̄_j` of Lemma 5.10. let s : K := (Fin.cases 1 y j : K)⁻¹ have hjs : (Fin.cases 1 y j : K) * s = 1 := mul_inv_cancel₀ hj let c : Fin (r + 1) → V.toSubring := fun a => ⟨(Fin.cases 1 y a : K) / Fin.cases 1 y j, hAV ⟨_, L.q_mem a⟩⟩ have hc : ∀ a, (c a : K) = (Fin.cases 1 y a : K) * s := fun a => div_eq_mul_inv _ _ have hcj : c j = 1 := Subtype.ext (div_self hj) have halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring) := by have h := L.residue_algebraic change (letI : Algebra k V.toSubring := cod; Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) at h rw [hcod_ground] at h exact h refine ⟨E, V, hEV, hVdvr, hxV, htrans, hxm, hEfin, hkaehler, ?_, j, hj, fun a ↦ (c a).2⟩ intro s' q hq hq1 hq0 by_cases hj0 : j = 0 · obtain ⟨j', hj'⟩ := hq1 exfalso have hsone : s = 1 := by simpa [hj0] using hjs have hprod : ∀ a, q a = q 0 * c a := by intro a apply Subtype.ext change (q a : K) = (q 0 : K) * (c a : K) rw [hq a, hq 0, hc a] simp [hsone] have hq0res : residue V.toSubring (q 0) = 0 := by by_contra hnz exact hq0 ((residue_ne_zero_iff_isUnit _).1 hnz) have hqjres : residue V.toSubring (q j') = 0 := by rw [hprod j', map_mul, hq0res, zero_mul] have hqjunit : IsUnit (q j') := by rw [hj']; exact isUnit_one exact ((residue_ne_zero_iff_isUnit _).2 hqjunit) hqjres · exact DivisorialVisibleFrameStage4.stage4'_isAlgebraic_of_normalized_column V y j hj0 s c hc hcj halg s' q hq hq1 hq0
/-- **Paragraph "`V̂ ≅ κ[[t]]`".** For the discrete valuation ring `V` of the proof, with uniformizer `ϖ` (`𝔪_V = (ϖ)`), the map `V → V̂ ≅ κ[[t]]` given by a coefficient field containing `K`: is injective; sends `c ∈ K` to the constant series `c`; sends `ϖ` to `t`; has constant term equal to the residue `res : V → κ`; and extends to an embedding `F ↪ κ((t))` which is the identity on `K`. -/ theorem step_completion {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) : let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W maximalIdeal V = Ideal.span {chosenUniformizer V} ∧ Function.Injective (retainedToCompletedPowerSeries W) ∧ (∀ c : k, retainedToCompletedPowerSeries W (C.groundCoeff c) = PowerSeries.C (algebraMap k κ c)) ∧ retainedToCompletedPowerSeries W (chosenUniformizer V) = PowerSeries.X ∧ (∀ v : V, PowerSeries.constantCoeff (retainedToCompletedPowerSeries W v) = residue V v) ∧ Function.Injective (retainedLaurentLift P ⟨0, hm⟩ W) ∧ (∀ v : V, retainedLaurentLift P ⟨0, hm⟩ W (algebraMap V (ComponentFractionField P) v) = algebraMap (PowerSeries κ) (LaurentSeries κ) (retainedToCompletedPowerSeries W v)) ∧ (retainedLaurentLift P ⟨0, hm⟩ W).comp (algebraMap k (ComponentFractionField P)) = groundLaurentMap (k := k) (K := κ) := by intro C W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra intro V letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra intro κ letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W have hground : ∀ c : k, retainedToCompletedPowerSeries W (C.groundCoeff c) = PowerSeries.C (algebraMap k κ c) := by intro c rw [C.groundCoeff_eq_retained] exact retainedToCompletedPowerSeries_ground P ⟨0, hm⟩ W c exact ⟨maximalIdeal_eq_span_chosenUniformizer V, retainedToCompletedPowerSeries_injective W, hground, retainedToCompletedPowerSeries_chosenUniformizer (k := k) W, constantCoeff_retainedToCompleted_eq_residue (k := k) W, RingHom.injective _, retainedLaurentLift_algebraMap P ⟨0, hm⟩ W, retainedLaurentLift_comp_algebraMap P ⟨0, hm⟩ W⟩
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureNormalizationSource ↗open Stafford38.Geometry.CompletedDVRPowerSeriesSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperResidueGroundAlgebraTransportSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗universe u
/-- **Paragraph "`a = ord q_0`, `b = ord q_1`".** For the column `q_i = x̄_i / x̄_j` of the discrete valuation ring of Lemmas 5.9–5.10, read in `κ[[t]]`: `q_0 = t^a u_0` and `q_1 = t^b u_1` with `u_0(0) ≠ 0 ≠ u_1(0)` and `0 < a < b` (`a > 0` by `step_a_pos`, from Lemma 5.10(a)); hence `q_j = 1` forces `j ∉ {0, 1}`. The hypothesis `hunit` (`x̄_1` is a unit of `R`) is the consequence of `Y ∩ H = ∅`. -/ theorem step_orders {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) : let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V let q : Fin (m + 1) → PowerSeries κ := fun a ↦ retainedToCompletedPowerSeries W (C.q a) ∃ (a b : ℕ) (u₀ u₁ : PowerSeries κ), 0 < a ∧ a < b ∧ q 0 = (PowerSeries.X : PowerSeries κ) ^ a * u₀ ∧ PowerSeries.constantCoeff u₀ ≠ 0 ∧ q (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ b * u₁ ∧ PowerSeries.constantCoeff u₁ ≠ 0 ∧ q C.chart = 1 ∧ C.chart ≠ 0 ∧ C.chart ≠ Fin.succ ⟨0, hm⟩ := by classical intro C W V κ q letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra -- `a > 0` (`step_a_pos`, by Lemma 5.10(a)) and `b - a = ord x̄_1 > 0`. have hq₀max : C.q 0 ∈ maximalIdeal V := by rw [mem_maximalIdeal, mem_nonunits_iff] exact step_a_pos hm P w hunit have hparammax : W.place.parameter ∈ maximalIdeal V := by rw [mem_maximalIdeal, mem_nonunits_iff] exact W.place.parameter_nonunit obtain ⟨a, b', u₀, u₁', ha, hab', hq0, hu₀, hq1'⟩ := completed_order_gap (retainedToCompletedPowerSeries W) (Stafford38.Geometry.CompletedDVRPowerSeries.chosenUniformizer V) (retainedToCompletedPowerSeries_chosenUniformizer W) (Stafford38.Geometry.CompletedDVRPowerSeries.chosenUniformizer_irreducible V) (C.q 0) W.place.parameter (C.q (Fin.succ ⟨0, hm⟩)) C.q0_ne W.place.parameter_ne hq₀max hparammax C.q_parameter have hqj : q C.chart = 1 := (congrArg (retainedToCompletedPowerSeries W) C.chart_one).trans (map_one _) -- `q_1 = q_0 x̄_1 ≠ 0`, so `b = ord q_1` is finite. have hq1ne : q (Fin.succ ⟨0, hm⟩) ≠ 0 := by intro h have hV : C.q (Fin.succ ⟨0, hm⟩) = 0 := retainedToCompletedPowerSeries_injective W (h.trans (map_zero _).symm) rw [C.q_parameter] at hV exact mul_ne_zero C.q0_ne W.place.parameter_ne hV have hu₁ : PowerSeries.constantCoeff (q (Fin.succ ⟨0, hm⟩)).divXPowOrder ≠ 0 := fun h ↦ hq1ne (PowerSeries.constantCoeff_divXPowOrder_eq_zero_iff.mp h) have hq1'' : q (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ b' * u₁' := hq1' have hb : b' ≤ (q (Fin.succ ⟨0, hm⟩)).order.toNat := by by_contra hlt apply hu₁ rw [PowerSeries.constantCoeff_divXPowOrder] generalize (q (Fin.succ ⟨0, hm⟩)).order.toNat = n at hlt ⊢ rw [hq1'', PowerSeries.coeff_X_pow_mul'] exact ite_eq_right_iff.mpr fun h ↦ absurd h (by omega) have hconst : ∀ i, q i = (PowerSeries.X : PowerSeries κ) ^ a * u₀ → i ≠ C.chart := by intro i hi hij have hi1 : q i = 1 := by rw [hij]; exact hqj have h1 := congrArg PowerSeries.constantCoeff (hi1.symm.trans hi) simp [ha.ne'] at h1 have hconst1 : C.chart ≠ Fin.succ ⟨0, hm⟩ := by intro hij have hi1 : q (Fin.succ ⟨0, hm⟩) = 1 := by rw [← hij]; exact hqj have h1 := congrArg PowerSeries.constantCoeff (hi1.symm.trans hq1'') simp [(ha.trans hab').ne'] at h1 exact ⟨a, (q (Fin.succ ⟨0, hm⟩)).order.toNat, u₀, (q (Fin.succ ⟨0, hm⟩)).divXPowOrder, ha, hab'.trans_le hb, hq0, hu₀, PowerSeries.X_pow_order_mul_divXPowOrder.symm, hu₁, hqj, fun h ↦ hconst 0 hq0 h.symm, hconst1⟩
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Paragraph "`a = ord q_0`", `a > 0`.** If `a = 0`, then `q_0` is a unit of `V`, hence all `x̄_i = q_i/q_0` lie in `V`, so `R ⊆ V` since also `K ⊆ V`, contradicting Lemma 5.10(a) (`lemma_5_10_a_not_R_le`). So `q_0 ∉ V^×`. -/ theorem step_a_pos {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra ¬ IsUnit (w.column.q 0) := by let C := w.column let W := C.W let K := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) K := W.ambientAlgebra let V := W.place.valuation letI : IsDiscreteValuationRing V.toSubring := W.place.isDiscrete let x : Fin m → K := fun j ↦ componentCoordinate P j -- `1 = ḡ x̄_1` with `ḡ ∈ R = K[x̄_1, …, x̄_n]`. obtain ⟨g, hg⟩ := hunit let phi : MvPolynomial (Fin m) k →+* K := (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) K).comp (Ideal.Quotient.mk P.asIdeal) have hphiC : phi.comp MvPolynomial.C = algebraMap k K := by ext c exact IsScalarTower.algebraMap_apply k (MvPolynomial (Fin m) k ⧸ P.asIdeal) K c have hpoly : phi g = MvPolynomial.eval₂ (algebraMap k K) x g := by rw [MvPolynomial.map_mvPolynomial_eq_eval₂ phi g] change MvPolynomial.eval₂Hom (phi.comp MvPolynomial.C) (fun j ↦ phi (MvPolynomial.X j)) g = MvPolynomial.eval₂Hom (algebraMap k K) x g exact MvPolynomial.eval₂Hom_congr hphiC rfl rfl have hinverse : x ⟨0, hm⟩ * MvPolynomial.eval₂ (algebraMap k K) x g = 1 := by have hzero : phi (MvPolynomial.X ⟨0, hm⟩ * g - 1) = 0 := by have hmk : Ideal.Quotient.mk P.asIdeal (MvPolynomial.X ⟨0, hm⟩ * g - 1) = 0 := Ideal.Quotient.eq_zero_iff_mem.mpr hg simpa [phi] using congrArg (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) K) hmk rw [map_sub, map_mul, map_one, sub_eq_zero, hpoly] at hzero exact hzero have hgR : MvPolynomial.eval₂ (algebraMap k K) x g ∈ Algebra.adjoin k (Set.range x) := by rw [← MvPolynomial.aeval_def, Algebra.adjoin_range_eq_range_aeval] exact ⟨g, rfl⟩ -- `x̄_1 = ϖ ∈ 𝔪_V`. have hxV : x ⟨0, hm⟩ ∈ V.toSubring := by change componentCoordinate P ⟨0, hm⟩ ∈ V.toSubring rw [← W.parameter_eq_coordinate] exact W.place.parameter.2 have hxm : (⟨x ⟨0, hm⟩, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := by have : (⟨x ⟨0, hm⟩, hxV⟩ : V.toSubring) = W.place.parameter := Subtype.ext W.parameter_eq_coordinate.symm rw [this, mem_maximalIdeal, mem_nonunits_iff] exact W.place.parameter_nonunit have hR := lemma_5_10_a_not_R_le x ⟨0, hm⟩ V hxV hxm ⟨_, hgR, hinverse⟩ rintro ⟨u, hu⟩ apply hR -- `q_0 ∈ V^×` puts every `x̄_i = q_i/q_0` in `V`. have hq0 : (C.q 0 : K) = C.scale := by simpa [componentProjectivePoint] using C.q_commonScale 0 have hxjV : ∀ j, x j ∈ V.toSubring := by intro j have hqj : (C.q (Fin.succ j) : K) = C.scale * x j := by simpa [componentProjectivePoint] using C.q_commonScale (Fin.succ j) have hscale : ((u : V.toSubring) : K) = C.scale := by rw [hu]; exact hq0 have hu1 : ((u : V.toSubring) : K) * ((↑(u⁻¹) : V.toSubring) : K) = 1 := by rw [← Subring.coe_mul, Units.mul_inv] rfl have : x j = (C.q (Fin.succ j) : K) * ((↑(u⁻¹) : V.toSubring) : K) := by rw [hqj, ← hscale, mul_comm ((u : V.toSubring) : K), mul_assoc, hu1, mul_one] rw [this] exact V.toSubring.mul_mem (C.q _).2 (↑(u⁻¹) : V.toSubring).2 intro z hz induction hz using Algebra.adjoin_induction with | mem z hz => obtain ⟨j, rfl⟩ := hz exact hxjV j | algebraMap c => rw [← C.groundCoeff_commutes] exact (C.groundCoeff c).2 | add a b _ _ ha hb => exact V.toSubring.add_mem ha hb | mul a b _ _ ha hb => exact V.toSubring.mul_mem ha hb
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Paragraph defining `y` (equation (generic)).** The affine point `y = (q_1/q_0, …, q_n/q_0) ∈ κ((t))^n` is the image of `(x̄_1, …, x̄_n)` under `F ↪ κ((t))`, so `f(y) = 0 ↔ f ∈ 𝔭` for every `f ∈ K[x]`. -/ theorem step_generic_point {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) : let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W let q : Fin (m + 1) → PowerSeries κ := fun a ↦ retainedToCompletedPowerSeries W (C.q a) ∀ f : MvPolynomial (Fin m) k, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) (dehomogenizedPoint (laurentColumn q)) f = 0 ↔ f ∈ P.asIdeal := by classical intro C W V κ q letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W have hground := retainedLaurentLift_comp_algebraMap P ⟨0, hm⟩ W have hpoint := retainedLaurentLift_componentCoordinate P ⟨0, hm⟩ W C.q C.scale C.q_commonScale C.q0_ne have hker := component_generic_kernel_after_point_transport P (retainedLaurentLift P ⟨0, hm⟩ W) (RingHom.injective _) hground _ hpoint intro f rw [← MvPolynomial.eval_map] exact hker f
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Lemma 5.11 for the valuation ring of the formal proof.** The residues `res q_i` of its normalized column satisfy `res q_0 = res q_1 = 0`, `res q_j = 1`, and some `res q_i` with `i ∉ {0, 1, j}` have differentials forming a basis of `Ω_{κ/K}`. -/ theorem witness_lemma_5_11 {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) : let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W residue V (C.q 0) = 0 ∧ residue V (C.q (Fin.succ ⟨0, hm⟩)) = 0 ∧ residue V (C.q C.chart) = 1 ∧ ∃ (d' : ℕ) (rows : Fin d' → Fin (m + 1)), Function.Injective rows ∧ (∀ l, rows l ≠ 0 ∧ rows l ≠ Fin.succ ⟨0, hm⟩ ∧ rows l ≠ C.chart) ∧ LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l)))) ∧ Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l)))) = ⊤ ∧ d' = Module.finrank κ (Ω[κ⁄k]) := by classical let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W letI : IsLocalRing V := W.place.isDiscrete.toIsLocalRing obtain ⟨a, b, u₀, u₁, ha, hab, hq0, -, hq1⟩ := order_gap_of_actual_witness hm P w hunit have hres := constantCoeff_retainedToCompleted_eq_residue (k := k) W have hr0 : residue V (C.q 0) = 0 := by rw [← hres, hq0] simp [ha.ne'] have hr1 : residue V (C.q (Fin.succ ⟨0, hm⟩)) = 0 := by rw [← hres, hq1] simp [(ha.trans hab).ne'] have hrj : residue V (C.q C.chart) = 1 := by rw [C.chart_one, map_one] have halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a ↦ residue V (C.q a)) : IntermediateField k κ) κ := by have h := w.halg have hle : IntermediateField.adjoin k (Set.range fun j : Fin m ↦ residue V (C.q (Fin.succ j))) ≤ IntermediateField.adjoin k (Set.range fun a ↦ residue V (C.q a)) := IntermediateField.adjoin.mono _ _ _ (by rintro _ ⟨j, rfl⟩; exact ⟨_, rfl⟩) exact ⟨fun z ↦ IsAlgebraic.tower_top_of_subalgebra_le (IntermediateField.toSubalgebra_le_toSubalgebra.mpr hle) (h.isAlgebraic z)⟩ exact ⟨hr0, hr1, hrj, lemma_5_11 hm (fun a ↦ residue V (C.q a)) C.chart hr0 hr1 hrj halg⟩
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Paragraph after Lemma 5.11 (dual derivations and `Z`).** If the differentials `d(q_{i_l}(0))` form a basis of `Ω_{κ/K}`, there are dual `K`-derivations `D_k` with `D_k(q_{i_l}(0)) = δ_{kl}` (Definition 5.2(c)). The matrix `Z_{ik} = Σ_r D_k(c_{ir}) t^r` then has `Z(0)` equal to the identity on the rows `i_l`, so its selected minor has constant coefficient `1`; `Z_j = 0`; and rows `0` and `1` are divisible by `t^a` and `t^b`. -/ theorem step_dual_derivations {m d : ℕ} (hm : 0 < m) (q : Fin (m + 1) → PowerSeries κ) (rows : Fin d ↪ Fin (m + 1)) (hli : LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (PowerSeries.constantCoeff (q (rows l))))) (hspan : Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (PowerSeries.constantCoeff (q (rows l)))) = ⊤) (j : Fin (m + 1)) (hqj : q j = 1) (a b : ℕ) (u₀ u₁ : PowerSeries κ) (hq0 : q 0 = (PowerSeries.X : PowerSeries κ) ^ a * u₀) (hq1 : q (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ b * u₁) : ∃ D : Fin d → Derivation k κ κ, (∀ k' l, D k' (PowerSeries.constantCoeff (q (rows l))) = if k' = l then 1 else 0) ∧ (∀ l k', PowerSeries.constantCoeff (coefficientwiseTangentMatrix q D (rows l) k') = if l = k' then 1 else 0) ∧ PowerSeries.constantCoeff (selectedMinor (coefficientwiseTangentMatrix q D) rows).det = 1 ∧ (∀ k', coefficientwiseTangentMatrix q D j k' = 0) ∧ (∀ k', ∃ w : PowerSeries κ, coefficientwiseTangentMatrix q D 0 k' = (PowerSeries.X : PowerSeries κ) ^ a * w) ∧ (∀ k', ∃ w : PowerSeries κ, coefficientwiseTangentMatrix q D (Fin.succ ⟨0, hm⟩) k' = (PowerSeries.X : PowerSeries κ) ^ b * w) := by classical let B : Module.Basis (Fin d) κ (Ω[κ⁄k]) := Module.Basis.mk hli hspan.ge -- Definition 5.2(c): the dual derivations `D_k` of the basis `dβ_l`. obtain ⟨D, hdual⟩ := definition_5_2_c (K := k) (L := κ) (fun l ↦ PowerSeries.constantCoeff (q (rows l))) B (fun l ↦ Module.Basis.mk_apply _ _ _) have hentry : ∀ l k', PowerSeries.constantCoeff (coefficientwiseTangentMatrix q D (rows l) k') = if l = k' then 1 else 0 := by intro l k' rw [constantCoeff_coefficientwiseTangentMatrix, hdual] by_cases h : k' = l · subst h; simp · simp [h, Ne.symm h] refine ⟨D, hdual, hentry, ?_, ?_, ?_, ?_⟩ · rw [constantCoeff_selectedMinor_det] have hone : selectedMinor (residueMatrix (coefficientwiseTangentMatrix q D)) rows = 1 := by ext l k' rw [Matrix.one_apply] exact hentry l k' rw [hone, Matrix.det_one] · intro k' rw [coefficientwiseTangentMatrix] simp only [hqj] exact (coefficientwiseDerivation (D k')).map_one_eq_zero · intro k' refine ⟨coefficientwiseDerivation (D k') u₀, ?_⟩ rw [coefficientwiseTangentMatrix, hq0] exact coefficientwiseDerivation_X_pow_mul (D k') a u₀ · intro k' refine ⟨coefficientwiseDerivation (D k') u₁, ?_⟩ rw [coefficientwiseTangentMatrix, hq1] exact coefficientwiseDerivation_X_pow_mul (D k') b u₁ omit [CharZero κ] in
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Paragraph defining `λ`, `w` and `τ`.** `Z̃(0)` is the identity, so `Z̃` is invertible over `κ[[t]]` and there is a unique `λ` with `Z̃ λ = (q'_{i_1}, …, q'_{i_{d'}})`. For `w = q' − Zλ`: `w_{i_l} = 0`; `ord w_0 = a − 1` exactly; `c`, the minimal order of the entries of `w`, satisfies `c ≤ a − 1`; `τ = t^{-c} w` lies in `κ[[t]]^{n+1}` (stated as `w = t^c τ`) with `τ(0) ≠ 0`, `τ_{i_l} = 0`, `ord τ_1 ≥ b − a > 0` (so `τ_1(0) = 0`) and `τ_j = 0`. -/ theorem step_lambda_w_tau {m d : ℕ} (hm : 0 < m) (q : Fin (m + 1) → PowerSeries κ) (Z : Matrix (Fin (m + 1)) (Fin d) (PowerSeries κ)) (rows : Fin d ↪ Fin (m + 1)) (j : Fin (m + 1)) (a b : ℕ) (u₀ u₁ : PowerSeries κ) (hqj : q j = 1) (hZj : ∀ k', Z j k' = 0) (ha : 0 < a) (hab : a < b) (hq0 : q 0 = (PowerSeries.X : PowerSeries κ) ^ a * u₀) (hu₀ : PowerSeries.constantCoeff u₀ ≠ 0) (hZ0 : ∀ k', ∃ w : PowerSeries κ, Z 0 k' = (PowerSeries.X : PowerSeries κ) ^ a * w) (hq1 : q (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ b * u₁) (hZ1 : ∀ k', ∃ w : PowerSeries κ, Z (Fin.succ ⟨0, hm⟩) k' = (PowerSeries.X : PowerSeries κ) ^ b * w) (hminor : PowerSeries.constantCoeff (selectedMinor Z rows).det = 1) : ∃ (lambda : Fin d → PowerSeries κ) (c : ℕ) (tau : Fin (m + 1) → PowerSeries κ), (∀ l, PowerSeries.derivative (R := κ) (q (rows l)) = Z.mulVec lambda (rows l)) ∧ (∀ lambda' : Fin d → PowerSeries κ, (∀ l, PowerSeries.derivative (R := κ) (q (rows l)) = Z.mulVec lambda' (rows l)) → lambda' = lambda) ∧ (∀ l, PowerSeries.derivative (R := κ) (q (rows l)) - Z.mulVec lambda (rows l) = 0) ∧ (PowerSeries.derivative (R := κ) (q 0) - Z.mulVec lambda 0).order = (a - 1 : ℕ) ∧ (∀ i, (c : ℕ∞) ≤ (PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i).order) ∧ (∃ i, (PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i).order = c) ∧ c ≤ a - 1 ∧ (∀ i, PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i = (PowerSeries.X : PowerSeries κ) ^ c * tau i) ∧ (∃ i, PowerSeries.constantCoeff (tau i) ≠ 0) ∧ (∀ l, tau (rows l) = 0) ∧ (∃ s : PowerSeries κ, tau (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ (b - a) * s) ∧ PowerSeries.constantCoeff (tau (Fin.succ ⟨0, hm⟩)) = 0 ∧ tau j = 0 := by classical have hminor0 : PowerSeries.constantCoeff (selectedMinor Z rows).det ≠ 0 := by rw [hminor]; exact one_ne_zero -- `Z̃(0)` is the identity, so `Z̃` is invertible over `κ[[t]]` and -- `λ = Z̃⁻¹ (q'_{i_1}, …, q'_{i_{d'}})` solves `Z̃ λ = (q'_{i_l})`. let v : Fin (m + 1) → PowerSeries κ := fun i ↦ PowerSeries.derivative (R := κ) (q i) let lambda : Fin d → PowerSeries κ := correctionCoefficients Z rows v have hsel : ∀ l, PowerSeries.derivative (R := κ) (q (rows l)) = Z.mulVec lambda (rows l) := fun l ↦ sub_eq_zero.mp (correctedVector_selectedRow_eq_zero Z rows v hminor0 l) -- The solution is unique. have huniq : ∀ lambda' : Fin d → PowerSeries κ, (∀ l, PowerSeries.derivative (R := κ) (q (rows l)) = Z.mulVec lambda' (rows l)) → lambda' = lambda := by intro lambda' h' have hunit : IsUnit (selectedMinor Z rows) := (Matrix.isUnit_iff_isUnit_det _).mpr (PowerSeries.isUnit_iff_constantCoeff.mpr (by rw [hminor]; exact isUnit_one)) apply Matrix.mulVec_injective_of_isUnit hunit funext l change Z.mulVec lambda' (rows l) = Z.mulVec lambda (rows l) rw [← h', hsel] -- `w = q' − Zλ` has `w_{i_l} = 0`. let w : Fin (m + 1) → PowerSeries κ := fun i ↦ PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i have hwrows : ∀ l, w (rows l) = 0 := fun l ↦ sub_eq_zero.mpr (hsel l) -- `ord w_0 = a − 1`: `q'_0` has order exactly `a − 1` (its coefficient `a u_0(0)` of -- `t^{a-1}` is nonzero in characteristic zero), while `Z_{0•} λ` has order `≥ a`. obtain ⟨r₀, hr₀, hr₀c⟩ := derivative_sub_mulVec_eq_X_pow_mul q Z lambda 0 a ha u₀ hq0 hZ0 have hr₀ne : PowerSeries.constantCoeff r₀ ≠ 0 := by rw [hr₀c] exact mul_ne_zero (Nat.cast_ne_zero.mpr ha.ne') hu₀ have hw0 : (PowerSeries.derivative (R := κ) (q 0) - Z.mulVec lambda 0).order = (a - 1 : ℕ) := by rw [hr₀] exact order_X_pow_mul_of_constantCoeff_ne_zero _ _ hr₀ne -- `c` is the minimal order among the entries of `w`; then `c ≤ a − 1`, and -- `τ = t^{-c} w` lies in `κ[[t]]^{n+1}` with `τ(0) ≠ 0`. have hcoeff : PowerSeries.coeff (a - 1) (w 0) ≠ 0 := by change PowerSeries.coeff (a - 1) (PowerSeries.derivative (R := κ) (q 0) - Z.mulVec lambda 0) ≠ 0 rw [hr₀] simpa [PowerSeries.coeff_zero_eq_constantCoeff_apply] using (show PowerSeries.coeff (0 + (a - 1)) ((PowerSeries.X : PowerSeries κ) ^ (a - 1) * r₀) ≠ 0 by rw [PowerSeries.coeff_X_pow_mul] simpa [PowerSeries.coeff_zero_eq_constantCoeff_apply] using hr₀ne) obtain ⟨c, hc, tau, hfactor, hprim⟩ := exists_primitive_commonPower_of_coeff_ne_zero w 0 (a - 1) hcoeff have hfactor' : ∀ i, PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i = (PowerSeries.X : PowerSeries κ) ^ c * tau i := hfactor have htau_of_w : ∀ i, w i = 0 → tau i = 0 := fun i hi ↦ (mul_eq_zero.mp ((hfactor i).symm.trans hi)).resolve_left (pow_ne_zero c PowerSeries.X_ne_zero) have htaurows : ∀ l, tau (rows l) = 0 := fun l ↦ htau_of_w _ (hwrows l) -- In row `1`, `q'_1` and `Z_{1•} λ` have order `≥ b − 1`, so `ord τ_1 ≥ b − 1 − c ≥ b − a`. obtain ⟨r₁, hr₁, -⟩ := derivative_sub_mulVec_eq_X_pow_mul q Z lambda _ b (ha.trans hab) u₁ hq1 hZ1 have htau1' : tau (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ (b - a) * ((PowerSeries.X : PowerSeries κ) ^ (a - 1 - c) * r₁) := by apply mul_left_cancel₀ (pow_ne_zero c (PowerSeries.X_ne_zero (R := κ))) rw [← hfactor', hr₁, ← mul_assoc, ← mul_assoc, ← pow_add, ← pow_add] congr 2 omega have htau1 : PowerSeries.constantCoeff (tau (Fin.succ ⟨0, hm⟩)) = 0 := by rw [htau1', map_mul, map_pow, PowerSeries.constantCoeff_X, zero_pow (Nat.sub_ne_zero_of_lt hab), zero_mul] -- `q'_j = 0` and `Z_{j•} = 0` (as `q_j = 1`), so `τ_j = 0`. have htauj : tau j = 0 := by apply htau_of_w change PowerSeries.derivative (R := κ) (q j) - Z.mulVec lambda j = 0 rw [hqj, Derivation.map_one_eq_zero, zero_sub, neg_eq_zero] simp [Matrix.mulVec, dotProduct, hZj] refine ⟨lambda, c, tau, hsel, huniq, hwrows, hw0, ?_, ?_, hc, hfactor', hprim, htaurows, ⟨_, htau1'⟩, htau1, htauj⟩ · intro i rw [hfactor' i, PowerSeries.order_mul, PowerSeries.order_X_pow] exact le_self_add · obtain ⟨i, hi⟩ := hprim exact ⟨i, by rw [hfactor' i]; exact order_X_pow_mul_of_constantCoeff_ne_zero c _ hi⟩ omit [CharZero κ] in
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Paragraph "`M(0)` has full column rank".** For `M = (q, Z_{•1}, …, Z_{•d'}, τ)`: `M(0)` has full column rank (row `j`, the identity block on the rows `i_l`, and `τ(0) ≠ 0`); so some `(d'+2) × (d'+2)` minor `M_ι(0)` is nonzero, the corresponding minor of `M(t)` is a unit of `κ[[t]]`, and the columns of `M(t)` are linearly independent over `κ((t))`. Every column of `M(0)` has vanishing `1`-coordinate. -/ theorem step_M_minor {m d : ℕ} (hm : 0 < m) (q : Fin (m + 1) → PowerSeries κ) (Z : Matrix (Fin (m + 1)) (Fin d) (PowerSeries κ)) (tau : Fin (m + 1) → PowerSeries κ) (rows : Fin d ↪ Fin (m + 1)) (j : Fin (m + 1)) (b : ℕ) (u₁ : PowerSeries κ) (hqj : q j = 1) (hZj : ∀ k', Z j k' = 0) (htauj : tau j = 0) (htaurows : ∀ l, tau (rows l) = 0) (hminor : PowerSeries.constantCoeff (selectedMinor Z rows).det = 1) (hprim : ∃ i, PowerSeries.constantCoeff (tau i) ≠ 0) (hb : 0 < b) (hq1 : q (Fin.succ ⟨0, hm⟩) = (PowerSeries.X : PowerSeries κ) ^ b * u₁) (hZ1 : ∀ k', ∃ w : PowerSeries κ, Z (Fin.succ ⟨0, hm⟩) k' = (PowerSeries.X : PowerSeries κ) ^ b * w) (htau1 : PowerSeries.constantCoeff (tau (Fin.succ ⟨0, hm⟩)) = 0) : Function.Injective (residueMatrix (formalTangentMatrix q Z tau)).mulVec ∧ (∃ rowsι : FormalTangentColumn (Fin d) ↪ Fin (m + 1), PowerSeries.constantCoeff (selectedMinor (formalTangentMatrix q Z tau) rowsι).det ≠ 0 ∧ IsUnit (selectedMinor (formalTangentMatrix q Z tau) rowsι).det) ∧ LinearIndependent (LaurentSeries κ) (fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col)) ∧ ∀ col, PowerSeries.constantCoeff (formalTangentMatrix q Z tau (Fin.succ ⟨0, hm⟩) col) = 0 := by classical have hinj := residue_formalTangentMatrix_mulVec_injective q Z tau rows j hqj hZj htauj htaurows (by rw [hminor]; exact one_ne_zero) hprim obtain ⟨rowsι, hρ⟩ := powerSeries_exists_selectedMinor_constantCoeff_ne_zero_of_residue_mulVec_injective _ hinj obtain ⟨hunit, hli⟩ := linearIndependent_laurent_of_selectedMinor _ rowsι hρ exact ⟨hinj, ⟨rowsι, hρ, hunit⟩, hli, constantCoeff_formalTangentMatrix_axis_eq_zero q Z tau _ b u₁ hb hq1 hZ1 htau1⟩ omit [CharZero k] [CharZero κ] in
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Lemmas 5.12(a), 5.12(b) and 5.13 at the Laurent generic point of the formal proof.** For the column `q(t)` of `witness_lemma_5_11`, with `y = q/q_0 ∈ κ((t))^n`: `dim Θ_y = dim_F Ω_{F/K}`, this equals `trdeg_K F = m`, and `m ≤ d' + 1` by the printed argument (`witness_lemma_5_12_a_bound`: second fundamental exact sequence for `V → κ`, Nakayama with `Ω_{V/K}` finitely generated by Lemma 5.10(c), and `F ⊗_V Ω_{V/K} = Ω_{F/K}`); `dim 𝒞 = dim Θ_y + 1`; `δq ∈ 𝒞` for every `K`-derivation `δ` of `κ((t))` and `q ∈ 𝒞`; and every row annihilating `𝒞` has conormal affine part. -/ theorem witness_lemma_5_12_5_13_paper {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hfin : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k])) : let C := w.column let W := C.W let F := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W let E := LaurentSeries κ letI : Algebra k E := (groundLaurentMap (k := k) (K := κ)).toAlgebra letI : SMul k E := (inferInstance : Algebra k E).toSMul letI : Module k E := (inferInstance : Algebra k E).toModule let q : Fin (m + 1) → E := laurentColumn fun a ↦ retainedToCompletedPowerSeries W (C.q a) (Module.finrank E (tangentSpaceAt q P.asIdeal) = Module.finrank F (Ω[F⁄k]) ∧ (Module.finrank F (Ω[F⁄k]) : Cardinal) = Algebra.trdeg k F ∧ Module.finrank F (Ω[F⁄k]) ≤ Module.finrank κ (Ω[κ⁄k]) + 1 ∧ Module.finrank E (tangentCone q P.asIdeal) = Module.finrank E (tangentSpaceAt q P.asIdeal) + 1) ∧ ((∀ δ : Derivation k E E, (fun a ↦ δ (q a)) ∈ tangentCone q P.asIdeal) ∧ q ∈ tangentCone q P.asIdeal) ∧ (∀ ell : Fin (m + 1) → E, (∀ v ∈ tangentCone q P.asIdeal, ∑ a, ell a * v a = 0) → coordinateCovector (fun i : Fin m ↦ ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint q) (P.asIdeal.map (MvPolynomial.map (algebraMap k E)))) := witness_lemma_5_12_5_13_of_bound hm P w (witness_lemma_5_12_a_bound hm P w hfin)
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Lemma 5.12(a), `m ≤ d' + 1`, at the place of the proof.** For the discrete valuation ring `V` of the witness, with `Ω_{V/K}` finitely generated (Lemma 5.10(c)), `dim_F Ω_{F/K} ≤ dim_κ Ω_{κ/K} + 1`; this is `lemma_5_12_a_bound` (second fundamental exact sequence for `V → κ`, Nakayama, `F ⊗_V Ω_{V/K} = Ω_{F/K}`), read with the witness's `K`-algebra structure on `κ`. -/ theorem witness_lemma_5_12_a_bound {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hfin : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k])) : let C := w.column let W := C.W let F := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W Module.finrank F (Ω[F⁄k]) ≤ Module.finrank κ (Ω[κ⁄k]) + 1 := by intro C W F letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra intro V letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra intro κ have hfrac : IsFractionRing (↥W.place.valuation) (ComponentFractionField P) := inferInstance letI : Algebra k V := C.groundCoeff.toAlgebra letI : Algebra V F := V.subtype.toAlgebra letI : IsScalarTower k V F := C.groundTower letI : IsFractionRing V F := hfrac letI : Module.Finite V (Ω[V⁄k]) := hfin -- The bound with `κ` a `K`-algebra through `V`, i.e. `c ↦ res (c)`. have hbound := lemma_5_12_a_bound (k := k) (V := V) (F := F) -- That structure is the witness's retained one. have hres : (retainedResidueGroundAlgebra P ⟨0, hm⟩ W : Algebra k κ) = (inferInstance : Algebra k κ) := by apply Algebra.algebra_ext intro c have h := RingHom.congr_fun w.retainedGroundMap c change (@algebraMap k κ _ _ (retainedResidueGroundAlgebra P ⟨0, hm⟩ W)) c = residue V (retainedComponentCoefficientMap P ⟨0, hm⟩ W c) at h rw [h, ← C.groundCoeff_eq_retained] rfl letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W change Module.finrank F (Ω[F⁄k]) ≤ (letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W; Module.finrank κ (Ω[κ⁄k])) + 1 rw [kaehler_finrank_eq_of_algebra_eq _ _ hres] exact hbound
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureNormalizationSource ↗open Stafford38.Geometry.CompletedDVRPowerSeriesSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperResidueGroundAlgebraTransportSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗universe u
/-- **Paragraph "the columns of `M` span `𝒞`".** The columns `q`, `Z_{•k}` and `q'` have the form `δq` for `K`-derivations `δ` of `κ((t))` (the `D_k` extended coefficientwise, and `d/dt`), so they lie in `𝒞` by Lemma 5.12(b); `τ = t^{-c}(q' − Zλ)` is a `κ((t))`-combination of them. With `dim 𝒞 = dim Θ_y + 1` (Lemma 5.12(a)) and `dim Θ_y ≤ d' + 1`, the `d' + 2` independent columns of `M` span `𝒞`. -/ theorem step_M_spans_cone {m d : ℕ} (I : Ideal (MvPolynomial (Fin m) k)) (q : Fin (m + 1) → PowerSeries κ) (hq0 : q 0 ≠ 0) (hbase : ∀ f ∈ I, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) (dehomogenizedPoint (laurentColumn q)) f = 0) (D : Fin d → Derivation k κ κ) (Z : Matrix (Fin (m + 1)) (Fin d) (PowerSeries κ)) (hZ : Z = coefficientwiseTangentMatrix q D) (lambda : Fin d → PowerSeries κ) (c : ℕ) (tau : Fin (m + 1) → PowerSeries κ) (hfactor : ∀ i, PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i = (PowerSeries.X : PowerSeries κ) ^ c * tau i) (hli : LinearIndependent (LaurentSeries κ) (fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col))) (hbound : Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I) ≤ d + 1) : (∀ col, (fun i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col)) ∈ tangentCone (laurentColumn q) I) ∧ Submodule.span (LaurentSeries κ) (Set.range fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col)) = tangentCone (laurentColumn q) I := by classical let L := LaurentSeries κ let f := algebraMap (PowerSeries κ) L let qL := laurentColumn q let 𝒞 := tangentCone qL I have hq0L : qL 0 ≠ 0 := laurentColumn_ne_zero_of_ne_zero q hq0 -- Lemma 5.12(b) at `y = q/q₀`. obtain ⟨hδ, hq⟩ := lemma_5_12_b qL hq0L I hbase have hder : ∀ δ : Derivation k (PowerSeries κ) (PowerSeries κ), laurentColumn (fun i ↦ δ (q i)) ∈ 𝒞 := by intro δ have h := hδ (extendPowerSeriesDerivation δ) have hcol : (fun a ↦ extendPowerSeriesDerivation δ (qL a)) = laurentColumn (fun i ↦ δ (q i)) := by funext a exact extendPowerSeriesDerivation_on_powerSeries δ (q a) rw [hcol] at h exact h have hZcol : ∀ l, (fun i ↦ f (Z i l)) ∈ 𝒞 := by intro l have h := hder (coefficientwiseDerivation (D l)) rw [hZ] exact h have hq' : (fun i ↦ f (PowerSeries.derivative (R := κ) (q i))) ∈ 𝒞 := hder (uniformizerDerivation (k := k) (K := κ)) -- `τ = t^{-c} (q' − Σ λ_l Z_{•l})`. let Xc : L := f ((PowerSeries.X : PowerSeries κ) ^ c) have hXc : Xc ≠ 0 := by simp [Xc, f] have htau : (fun i ↦ f (tau i)) ∈ 𝒞 := by have hcomb := 𝒞.smul_mem Xc⁻¹ (𝒞.sub_mem hq' (𝒞.sum_mem fun l (_ : l ∈ Finset.univ) ↦ 𝒞.smul_mem (f (lambda l)) (hZcol l))) convert hcomb using 1 funext i have hi := congrArg f (hfactor i) simp only [map_sub, map_mul, Matrix.mulVec, dotProduct, map_sum] at hi simp only [Pi.smul_apply, Pi.sub_apply, Finset.sum_apply, smul_eq_mul] rw [eq_inv_mul_iff_mul_eq₀ hXc] rw [← hi] congr 1 refine Finset.sum_congr rfl fun l _ ↦ ?_ ring have hcols : ∀ col, (fun i ↦ f (formalTangentMatrix q Z tau i col)) ∈ 𝒞 := by intro col rcases col with _ | (l | _) · exact hq · exact hZcol l · exact htau refine ⟨hcols, ?_⟩ have hle : Submodule.span L (Set.range fun col i ↦ f (formalTangentMatrix q Z tau i col)) ≤ 𝒞 := Submodule.span_le.mpr (by rintro _ ⟨col, rfl⟩; exact hcols col) apply Submodule.eq_of_le_of_finrank_le hle rw [finrank_span_eq_card hli] -- Lemma 5.12(a): `dim 𝒞 = dim Θ_y + 1 ≤ d' + 2`. have hcone : Module.finrank (LaurentSeries κ) (tangentCone (laurentColumn q) I) = Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I) + 1 := lemma_5_12_a_cone (laurentColumn q) hq0L I change Module.finrank (LaurentSeries κ) (tangentCone (laurentColumn q) I) ≤ _ simp only [Fintype.card_sum, Fintype.card_unit, Fintype.card_fin] omega
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Paragraph lifting `u`.** Let `M(0)` have full column rank and `u M(0) = 0`. There is a set `ι` of row indices with `M_ι(0)` invertible; then `det M_ι` has nonzero constant term, so `M_ι` is invertible over `κ[[t]]`, and `ℓ_υ = u_υ`, `ℓ_ι = −u_υ M_υ M_ι⁻¹` satisfies `ℓ M = 0` and `ℓ(0) = u`. -/ theorem step_lift_ell_minor {R ι : Type*} [Fintype R] [Fintype ι] [DecidableEq R] [DecidableEq ι] (M : Matrix R ι (PowerSeries κ)) (hinj : Function.Injective (residueMatrix M).mulVec) (u : R → κ) (hu : rowMul u (residueMatrix M) = 0) : ∃ rowsι : ι ↪ R, (selectedMinor (residueMatrix M) rowsι).det ≠ 0 ∧ IsUnit (selectedMinor M rowsι).det ∧ (∀ i, (¬ ∃ c, rowsι c = i) → liftRow M rowsι u i = PowerSeries.C (u i)) ∧ (∀ c, liftRow M rowsι u (rowsι c) = liftBlock M rowsι u c) ∧ rowMul (liftRow M rowsι u) M = 0 ∧ residueColumn (liftRow M rowsι u) = u := by classical obtain ⟨rowsι, h0⟩ := exists_selectedMinor_det_ne_zero_of_mulVec_injective _ hinj have h0' : PowerSeries.constantCoeff (selectedMinor M rowsι).det ≠ 0 := by rwa [constantCoeff_selectedMinor_det] have hunit : IsUnit (selectedMinor M rowsι).det := PowerSeries.isUnit_iff_constantCoeff.mpr (IsUnit.mk0 _ h0') set ℓ := liftRow M rowsι u -- `ℓ M = ℓ_ι M_ι + ℓ_υ M_υ = −u_υ M_υ + u_υ M_υ = 0`. have hblock : Matrix.vecMul (liftBlock M rowsι u) (selectedMinor M rowsι) = -rowMul (complementRow rowsι u) M := by rw [liftBlock, Matrix.neg_vecMul, Matrix.vecMul_vecMul, Matrix.nonsing_inv_mul _ hunit, Matrix.vecMul_one] have hell : rowMul ℓ M = 0 := by funext col have h := congrFun hblock col simp only [Matrix.vecMul, dotProduct, selectedMinor, Pi.neg_apply] at h have hsplit : rowMul ℓ M col = rowMul (complementRow rowsι u) M col + ∑ c, liftBlock M rowsι u c * M (rowsι c) col := by simp only [ℓ, liftRow, rowMul, add_mul, Finset.sum_add_distrib, Finset.sum_mul] congr 1 rw [Finset.sum_comm] refine Finset.sum_congr rfl fun c _ ↦ ?_ simp only [ite_mul, zero_mul] rw [Finset.sum_ite_eq] simp rw [hsplit, h, Pi.zero_apply, add_neg_cancel] refine ⟨rowsι, h0, hunit, liftRow_apply_of_not_mem M rowsι u, liftRow_apply_rows M rowsι u, hell, ?_⟩ -- `ℓ(0) − u` vanishes off `ι` and is killed by `M(0)`, hence vanishes on `ι`. let δ : R → κ := residueColumn ℓ - u have hoff : ∀ i, (¬ ∃ c, rowsι c = i) → δ i = 0 := by intro i hi simp [δ, residueColumn, ℓ, liftRow_apply_of_not_mem M rowsι u i hi] have hδ : Matrix.vecMul (fun c ↦ δ (rowsι c)) (selectedMinor (residueMatrix M) rowsι) = 0 := by funext col have hℓ := congrArg PowerSeries.constantCoeff (congrFun hell col) have hu' := congrFun hu col simp only [rowMul, map_sum, map_mul, Pi.zero_apply, map_zero] at hℓ hu' have hsum : ∑ i, δ i * residueMatrix M i col = 0 := by simp only [δ, Pi.sub_apply, sub_mul, Finset.sum_sub_distrib] rw [hu', sub_zero] exact hℓ rw [Pi.zero_apply, ← hsum] simp only [Matrix.vecMul, dotProduct, selectedMinor] refine Fintype.sum_of_injective rowsι rowsι.injective _ _ ?_ fun c ↦ rfl intro i hi rw [hoff i (by simpa using hi), zero_mul] have hA : IsUnit (selectedMinor (residueMatrix M) rowsι) := (Matrix.isUnit_iff_isUnit_det _).mpr (IsUnit.mk0 _ h0) have hzero := Matrix.vecMul_injective_of_isUnit hA (hδ.trans (Matrix.zero_vecMul _).symm) funext i by_cases hi : ∃ c, rowsι c = i · obtain ⟨c, rfl⟩ := hi exact sub_eq_zero.mp (congrFun hzero c) · exact sub_eq_zero.mp (hoff i hi) omit [CharZero k] [CharZero κ] in
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Paragraph "`ℓ` annihilates `𝒞`".** If the columns of `M` span `𝒞` and `ℓ M = 0`, then `ℓ` annihilates `𝒞`, so by Lemma 5.13 its affine part `ℓ̃ = (ℓ_1, …, ℓ_n)` lies in `span {df_y : f ∈ 𝔭}`; and `ℓ(0) = u` gives `ℓ̃(0) = (1, 0, …, 0)`. -/ theorem step_ell_conormal {m : ℕ} (hm : 0 < m) {ι : Type*} [Fintype ι] (I : Ideal (MvPolynomial (Fin m) k)) (q : Fin (m + 1) → PowerSeries κ) (hq0 : q 0 ≠ 0) (M : Matrix (Fin (m + 1)) ι (PowerSeries κ)) (hspan : Submodule.span (LaurentSeries κ) (Set.range fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (M i col)) = tangentCone (laurentColumn q) I) (ell : Fin (m + 1) → PowerSeries κ) (hell : rowMul ell M = 0) (hres : residueColumn ell = axisRow (k := κ) (Fin.succ ⟨0, hm⟩)) : coordinateCovector (fun i : Fin m ↦ laurentColumn ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint (laurentColumn q)) (I.map (MvPolynomial.map (algebraMap k (LaurentSeries κ)))) ∧ residueColumn (fun i : Fin m ↦ ell i.succ) = fun i ↦ if i = ⟨0, hm⟩ then 1 else 0 := by classical let L := LaurentSeries κ let f := algebraMap (PowerSeries κ) L let φ : (Fin (m + 1) → L) →ₗ[L] L := { toFun := fun v ↦ ∑ a, laurentColumn ell a * v a map_add' := fun v w ↦ by simp [mul_add, Finset.sum_add_distrib] map_smul' := fun c v ↦ by simp only [Pi.smul_apply, smul_eq_mul, RingHom.id_apply] rw [Finset.mul_sum] exact Finset.sum_congr rfl fun a _ ↦ by ring } have hle : tangentCone (laurentColumn q) I ≤ LinearMap.ker φ := by rw [← hspan, Submodule.span_le] rintro _ ⟨col, rfl⟩ have h := congrArg f (congrFun hell col) simp only [rowMul, map_sum, map_mul, Pi.zero_apply, map_zero] at h exact h have hann : ∀ v ∈ tangentCone (laurentColumn q) I, ∑ a, laurentColumn ell a * v a = 0 := fun v hv ↦ hle hv refine ⟨lemma_5_13 (laurentColumn q) (laurentColumn_ne_zero_of_ne_zero q hq0) I (laurentColumn ell) hann, ?_⟩ calc residueColumn (fun i : Fin m ↦ ell i.succ) = (fun i : Fin m ↦ residueColumn ell i.succ) := residueColumn_tail ell _ = (fun i : Fin m ↦ axisRow (k := κ) (Fin.succ ⟨0, hm⟩) i.succ) := by rw [hres] _ = (fun i : Fin m ↦ if i = ⟨0, hm⟩ then 1 else 0) := by funext i by_cases hi : i = ⟨0, hm⟩ · subst i simp [axisRow] · have hne : i.succ ≠ Fin.succ ⟨0, hm⟩ := fun h ↦ hi (Fin.succ_injective m h) rw [axisRow_apply_of_ne hne] simp [hi]
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
/-- **Final paragraph.** Let `P ∈ K[ξ]` vanish on `N(Y)`, let the equations of `Y` vanish at the generic point `y ∈ κ((t))^n`, and let `ℓ̃(t) ∈ κ[[t]]^n` lie in the `κ((t))`-span of `{dg_y : g ∈ 𝔭}`. Then `P(ℓ̃(0)) = 0`. -/ theorem step_polynomial_limit {k κ : Type u} [Field k] [IsAlgClosed k] [Field κ] [Algebra k κ] {m : ℕ} (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (Pol : MvPolynomial (Fin m) k) (hPol : Pol ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) (y : Fin m → LaurentSeries κ) (hbase : ∀ f ∈ I.asIdeal, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) y f = 0) (ell : Fin m → PowerSeries κ) (hell : coordinateCovector (fun i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (ell i)) ∈ equationCovectorSpan y (I.asIdeal.map (MvPolynomial.map (algebraMap k (LaurentSeries κ))))) : MvPolynomial.eval₂ (algebraMap k κ) (residueColumn ell) Pol = 0 := by apply eval_residue_eq_zero_of_mem_groundEquationConormalLocus I Pol hPol y ell refine ⟨?_, ?_⟩ · have hker : I.asIdeal.map (groundPolynomialMap (k := k) (K := κ) (Fin m)) ≤ RingHom.ker (MvPolynomial.eval y) := by rw [Ideal.map_le_iff_le_comap] intro f hf rw [Ideal.mem_comap, RingHom.mem_ker] change MvPolynomial.eval y (MvPolynomial.map (algebraMap k (LaurentSeries κ)) f) = 0 rw [MvPolynomial.eval_map] exact hbase f hf intro f hf exact RingHom.mem_ker.mp (hker hf) · change coordinateCovector (fun i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (ell i)) ∈ affineConormalSpace y (I.asIdeal.map (MvPolynomial.map (algebraMap k (LaurentSeries κ)))) rw [affineConormalSpace_eq_equationCovectorSpan] exact hell
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Proof of Theorem 5.7(ii) from (i).** Let every polynomial vanishing on `N(Y)` vanish at `ξ ≠ 0`, and let `G` be homogeneous of degree `d` and vanish on all classes `[ξ']` with `0 ≠ ξ' ∈ N(Y)`. By homogeneity `G` vanishes on every nonzero `ξ' ∈ N(Y)`, and at `0` if `d > 0`, hence on `N(Y)`; so `G(ξ) = 0`, and `G` vanishes at every representative `a ξ` of `[ξ]`. If `d = 0`, `G` is a constant vanishing at a class of `N(Y)`, so `G = 0`; that such a class exists follows from the hypothesis itself, applied to a coordinate `ξ_i ≠ 0`. -/ theorem step_dir_of_fibre {k : Type*} [Field k] [IsAlgClosed k] {n : ℕ} (I : Ideal (MvPolynomial (Fin n) k)) (ξ : Fin n → k) (hξ : ξ ≠ 0) (hvan : ξ ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I))) : Projectivization.mk k ξ hξ ∈ projectiveHomogeneousClosure (projectivizedDirectionSet (smoothConormalDirectionSet I)) := by intro G d hG hT -- `G` vanishes on every nonzero `ξ' ∈ N(Y)`. have hnonzero : ∀ v ∈ smoothConormalFibreProjection I, v ≠ 0 → MvPolynomial.eval v G = 0 := by intro v hv hv0 rcases hv with ⟨y, hy, hc⟩ exact hT ⟨v, ⟨y, hy, hv0, hc⟩, hv0, rfl⟩ v hv0 rfl have hξG : MvPolynomial.eval ξ G = 0 := by rcases Nat.eq_zero_or_pos d with hd | hd · -- `d = 0`: `G` is constant, and some nonzero `ξ' ∈ N(Y)` exists. subst hd have hconst : ∀ x : Fin n → k, MvPolynomial.eval x G = MvPolynomial.eval 0 G := by intro x have h := eval_smul_of_isHomogeneous G 0 hG 0 x rw [zero_smul, pow_zero, one_mul] at h exact h.symm obtain ⟨i, hi⟩ : ∃ i, ξ i ≠ 0 := by simpa [funext_iff] using hξ obtain ⟨v, hv, hv0⟩ : ∃ v ∈ smoothConormalFibreProjection I, v ≠ 0 := by by_contra hnone push Not at hnone apply hi have hXi : MvPolynomial.X i ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I) := by rw [MvPolynomial.mem_vanishingIdeal_iff] intro v hv rw [hnone v hv] simp simpa using (MvPolynomial.mem_zeroLocus_iff.mp hvan) _ hXi rw [hconst, ← hconst v] exact hnonzero v hv hv0 · -- `d > 0`: `G(0) = 0` too, so `G` vanishes on `N(Y)`. have hGvan : G ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I) := by rw [MvPolynomial.mem_vanishingIdeal_iff] intro v hv by_cases hv0 : v = 0 · subst hv0 have h := eval_smul_of_isHomogeneous G d hG 0 0 rw [smul_zero, zero_pow hd.ne', zero_mul] at h exact h · exact hnonzero v hv hv0 exact (MvPolynomial.mem_zeroLocus_iff.mp hvan) _ hGvan -- Every representative `a ξ` of `[ξ]`. intro v hv hz obtain ⟨a, ha⟩ := (Projectivization.mk_eq_mk_iff k ξ v hξ hv).mp hz have hscaled := eval_smul_of_isHomogeneous G d hG (a : k) v have ha' : (a : k) • v = ξ := by rw [← Units.smul_def] exact ha rw [ha', hξG] at hscaled exact (mul_eq_zero.mp hscaled.symm).resolve_left (pow_ne_zero d (Units.ne_zero a))
/-- **Lemma 5.9.** If `x̄_1` is transcendental over `K`, there are a valuation ring `K ⊆ W ⊊ F` with `x̄_1 ∈ 𝔪_W` and an index `j` with `x̄_i / x̄_j ∈ W` for all `0 ≤ i ≤ n`. -/ theorem lemma_5_9 {k F : Type*} [Field k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (htrans : Transcendental k (x i)) : ∃ (W : ValuationSubring F) (j : Fin (n + 1)), (∀ c : k, algebraMap k F c ∈ W) ∧ W ≠ ⊤ ∧ (∃ hx : x i ∈ W, (⟨x i, hx⟩ : W.toSubring) ∈ maximalIdeal W.toSubring) ∧ homogeneousCoordinates x j ≠ 0 ∧ ∀ a, homogeneousCoordinates x a / homogeneousCoordinates x j ∈ W := by classical -- `W` dominates `K[x̄_1]` localized at `x̄_1` (Stacks, Tag 00IA). obtain ⟨W, hkW, hxW, hxinvW⟩ := stage1_exists_valuationSubring_of_transcendental (x i) htrans have hx0 : x i ≠ 0 := fun h ↦ htrans (h ▸ isAlgebraic_zero) -- Definition 5.3(a): `𝔪_W = W ∖ W^×` have hxmax : (⟨x i, hxW⟩ : W.toSubring) ∈ maximalIdeal W.toSubring := by rw [(definition_5_3.2.1 W).2] rintro ⟨u, hu⟩ apply hxinvW have hmul : x i * ((u⁻¹ : W.toSubringˣ) : W.toSubring) = 1 := by have := congrArg Subtype.val (Units.mul_inv u) simpa [hu] using this have : (x i)⁻¹ = ((u⁻¹ : W.toSubringˣ) : W.toSubring) := (eq_inv_of_mul_eq_one_right hmul).symm rw [this] exact ((u⁻¹ : W.toSubringˣ) : W.toSubring).2 have hWtop : W ≠ ⊤ := by rintro rfl exact hxinvW (ValuationSubring.mem_top _) -- Among the nonzero `x̄_i` choose one that is maximal for `a ≼ b ↔ x̄_a / x̄_b ∈ W`. let Y := homogeneousCoordinates x let S : Set (Fin (n + 1)) := {a | Y a ≠ 0} have hS : S.Nonempty := ⟨0, by simp [S, Y]⟩ letI : LE (Fin (n + 1)) := ⟨fun a b ↦ Y b ≠ 0 ∧ Y a / Y b ∈ W⟩ have htr : IsTrans (Fin (n + 1)) LE.le := by constructor intro a b c hab hbc rcases hab with ⟨hb, hab⟩ rcases hbc with ⟨hc, hbc⟩ refine ⟨hc, ?_⟩ have := W.toSubring.mul_mem hab hbc rwa [div_mul_div_cancel₀ hb] at this letI : IsTrans (Fin (n + 1)) LE.le := htr obtain ⟨j, hjS, hjmax⟩ := S.toFinite.exists_maximalFor id S hS refine ⟨W, j, hkW, hWtop, ⟨hxW, hxmax⟩, hjS, fun a ↦ ?_⟩ by_cases ha : Y a = 0 · change Y a / Y j ∈ W rw [ha, zero_div] exact W.zero_mem -- Definition 5.3(a): `a/b ∈ W` or `b/a ∈ W` for nonzero `a, b` rcases (definition_5_3.1 W.toSubring).2 ⟨W, rfl⟩ (Y a / Y j) (div_ne_zero ha hjS) with h | h · exact h · rw [inv_div] at h exact (hjmax (show a ∈ S from ha) ⟨ha, h⟩).2
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- **Lemma 5.10.** For `q_i = x̄_i / x̄_j` from Lemma 5.9 there is a discrete valuation ring `K ⊆ V ⊊ F` (`Frac V = F` holds for every valuation subring) with (a) `q_i ∈ V` and `x̄_1 ∈ 𝔪_V`, (b) `κ` algebraic over `K(res q_0, …, res q_n)`, (c) `Ω_{V/K}` finitely generated; the conclusions are the fields of `Lemma510Place`, which also records `V = A_{𝔭_D}` from the proof. The clause `R ⊄ V` of (a) is `lemma_5_10_a_not_R_le`, stated for every valuation ring with `x̄_1 ∈ 𝔪_V`, which is how the proof of Theorem 5.7 uses it. -/ theorem lemma_5_10 {k F : Type u} [Field k] [CharZero k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (hgen : IntermediateField.adjoin k (Set.range x) = ⊤) (htrans : Transcendental k (x i)) (W : ValuationSubring F) (j : Fin (n + 1)) (hkW : ∀ c : k, algebraMap k F c ∈ W) (hxW : ∃ hx : x i ∈ W, (⟨x i, hx⟩ : W.toSubring) ∈ maximalIdeal W.toSubring) (hj : homogeneousCoordinates x j ≠ 0) (hqW : ∀ a, homogeneousCoordinates x a / homogeneousCoordinates x j ∈ W) : Nonempty (Lemma510Place k x i j) := by classical let Y := homogeneousCoordinates x obtain ⟨hxiW, hxmW⟩ := hxW have hy0 : x i ≠ 0 := fun h ↦ htrans (h ▸ isAlgebraic_zero) have hxiInvW : (x i)⁻¹ ∉ W := by intro h rw [(definition_5_3.2.1 W).2] at hxmW apply hxmW refine IsUnit.of_mul_eq_one (⟨(x i)⁻¹, h⟩ : W.toSubring) ?_ apply Subtype.ext simp [hy0] let s : F := (Y j)⁻¹ have hs : s ≠ 0 := inv_ne_zero hj have hjs : Y j * s = 1 := mul_inv_cancel₀ hj have hYsW : ∀ a, Y a * s ∈ W := fun a ↦ by simpa [s, div_eq_mul_inv] using hqW a -- `C_0 = K[q_0, …, q_n, x̄_1] ⊆ W`, and `A` is its integral closure in `F`. let C₀ : Subalgebra k F := Algebra.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) let SW : Subalgebra k F := { W.toSubring with algebraMap_mem' := hkW } have hC₀W : C₀ ≤ SW := by apply Algebra.adjoin_le rintro z (⟨a, rfl⟩ | hz) · exact hYsW a · rw [Set.mem_singleton_iff.mp hz] exact hxiW have htop : IntermediateField.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) = ⊤ := by apply top_unique rw [← hgen] apply IntermediateField.adjoin_le_iff.2 rintro _ ⟨a, rfl⟩ apply IntermediateField.mem_adjoin_iff_div.mpr refine ⟨Y (Fin.succ a) * s, ?_, s, ?_, ?_⟩ · exact Algebra.subset_adjoin (R := k) (A := F) (Set.mem_union_left _ (Set.mem_range_self _)) · have h0 : Y 0 * s ∈ C₀ := Algebra.subset_adjoin (R := k) (A := F) (Set.mem_union_left _ (Set.mem_range_self (0 : Fin (n + 1)))) simpa only [Y, homogeneousCoordinates, Fin.cases_zero, one_mul] using h0 · simp only [Y, Fin.cases_succ] exact (eq_div_iff hs).2 rfl letI : IsFractionRing C₀ F := IsFractionRing.of_field C₀ F fun z ↦ by have hz : z ∈ IntermediateField.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) := by rw [htop] exact Set.mem_univ z rw [IntermediateField.mem_adjoin_iff_div] at hz rcases hz with ⟨a, ha, b, hb, hab⟩ exact ⟨⟨a, ha⟩, ⟨b, hb⟩, hab⟩ let A₀ : Subalgebra C₀ F := integralClosure C₀ F let A : Subalgebra k F := A₀.restrictScalars k have hC₀A : C₀ ≤ A := fun z hz ↦ A₀.algebraMap_mem ⟨z, hz⟩ -- `W` is integrally closed, so `A ⊆ W` (Atiyah–Macdonald, Prop. 5.18). letI : IsIntegrallyClosedIn W.toSubring F := inferInstanceAs (IsIntegrallyClosedIn W F) have hA₀W : A₀.toSubring ≤ W.toSubring := by change (integralClosure C₀ F).toSubring ≤ W.toSubring apply (Subring.integralClosure_subring_le_iff (S := C₀) (T := W.toSubring)).2 exact hC₀W have hyA : x i ∈ A := A₀.algebraMap_mem ⟨x i, Algebra.subset_adjoin (Set.mem_union_right _ (Set.mem_singleton _))⟩ have hnormA : ∀ a, Y a * s ∈ A := fun a ↦ A₀.algebraMap_mem ⟨Y a * s, Algebra.subset_adjoin (Set.mem_union_left _ (Set.mem_range_self a))⟩ have hinvA : (x i)⁻¹ ∉ A := fun h ↦ hxiInvW (hA₀W h) -- `A` is a finite `C_0`-module (Hartshorne, Thm. I.3.9A). haveI hC₀ft : Algebra.FiniteType k C₀ := Algebra.FiniteType.adjoin_of_finite (Set.Finite.union (Set.finite_range _) (Set.finite_singleton _)) let D : Subalgebra C₀ (FractionRing C₀) := integralClosure C₀ (FractionRing C₀) haveI : Module.Finite C₀ D := finite_normalization_of_fg_domain k C₀ D let e : FractionRing C₀ ≃ₐ[C₀] F := IsLocalization.algEquiv C₀⁰ (FractionRing C₀) F haveI : Module.Finite C₀ A₀ := Module.Finite.equiv e.mapIntegralClosure.toLinearEquiv have hsub : A.toSubring = A₀.toSubring := rfl let er : A ≃+* A₀ := RingEquiv.subringCongr hsub letI : Algebra C₀ A := (er.symm.toRingHom.comp (algebraMap C₀ A₀)).toAlgebra let ea : A₀ ≃ₐ[C₀] A := AlgEquiv.ofRingEquiv (f := er.symm) (fun _ ↦ rfl) letI : IsScalarTower k C₀ A := IsScalarTower.of_algebraMap_eq fun _ ↦ by apply Subtype.ext rfl letI : IsScalarTower C₀ A F := IsScalarTower.of_algebraMap_eq fun _ ↦ rfl haveI : Module.Finite C₀ A := Module.Finite.equiv ea.toLinearEquiv haveI hAft : Algebra.FiniteType k A := Algebra.FiniteType.trans hC₀ft (inferInstance : Algebra.FiniteType C₀ A) have hAic : IsIntegrallyClosedIn A F := by apply Subring.isIntegrallyClosedIn_iff.mpr intro z hz change z ∈ integralClosure C₀ F exact isIntegral_trans (R := C₀) (A := A) z hz letI : IsFractionRing A F := IsFractionRing.of_field A F fun z ↦ by obtain ⟨a, b, hb, hab⟩ := IsFractionRing.div_surjective C₀ z refine ⟨⟨a, hC₀A a.2⟩, ⟨b, hC₀A b.2⟩, ?_⟩ change z = (a : F) / (b : F) exact hab.symm have haint : ∀ a : A, IsIntegral C₀ (a : F) := fun a ↦ a.2 letI : IsNoetherianRing A := Algebra.FiniteType.isNoetherianRing k A letI : IsIntegrallyClosed A := (isIntegrallyClosed_iff_isIntegrallyClosedIn F).mpr hAic -- `x̄_1` is a nonzero nonunit of `A`; a minimal prime over it has height one (Krull). let xA : A := ⟨x i, hyA⟩ have hxA0 : xA ≠ 0 := fun h ↦ hy0 (congrArg Subtype.val h) have hxAu : ¬ IsUnit xA := by intro hu obtain ⟨z, hz⟩ := isUnit_iff_exists_inv.mp hu have hzK : x i * (z : F) = 1 := congrArg Subtype.val hz apply hinvA rw [show (x i)⁻¹ = (z : F) by simpa using congrArg Inv.inv (eq_inv_of_mul_eq_one_left hzK)] exact z.2 -- `𝔭_D` is minimal among the primes containing `x̄_1`; by Krull's principal ideal -- theorem `ht 𝔭_D ≤ 1`, and `ht 𝔭_D = 1` because `(0)` is the only prime of height `0`. obtain ⟨p, hpmin⟩ := (Ideal.span {xA}).nonempty_minimalPrimes (by simpa [Ideal.span_singleton_eq_top] using hxAu) letI : p.IsPrime := hpmin.isPrime have hxp : xA ∈ p := hpmin.1.2 (Ideal.mem_span_singleton_self xA) have hpheight : p.height = 1 := by refine le_antisymm (Ideal.height_le_one_of_isPrincipal_of_mem_minimalPrimes _ _ hpmin) ?_ have hpne : p ≠ ⊥ := fun h ↦ hxA0 (by simpa [h] using hxp) rw [Order.one_le_iff_ne_zero] simpa [Ideal.height_eq_zero_iff_eq_bot] using hpne -- `V := A_{𝔭_D}` is a Noetherian local domain. let S := placeSubalgebra A p letI : IsLocalization.AtPrime S p := by unfold S placeSubalgebra infer_instance letI : IsLocalRing S := IsLocalization.AtPrime.isLocalRing S p letI : IsNoetherianRing S := IsLocalization.isNoetherianRing p.primeCompl S inferInstance -- It is normal, as a localization of an integrally closed domain (A–M, Prop. 5.12), letI : IsIntegrallyClosed S := isIntegrallyClosed_of_isLocalization S p.primeCompl p.primeCompl_le_nonZeroDivisors -- and its dimension is `ht 𝔭_D = 1`. have hdim : ringKrullDim S = 1 := by rw [IsLocalization.AtPrime.ringKrullDim_eq_height p S, hpheight] rfl -- A Noetherian local normal domain of dimension one is a DVR (Matsumura, Thm. 11.2). have hSdvr : IsDiscreteValuationRing S := by have hnf : ¬ IsField S := fun hf ↦ by letI := hf.toField rw [ringKrullDim_eq_zero_of_isField hf] at hdim exact zero_ne_one hdim have h3 : IsIntegrallyClosed S ∧ ∃! P : Ideal S, P ≠ ⊥ ∧ P.IsPrime := by refine ⟨inferInstance, IsLocalRing.maximalIdeal _, ⟨IsLocalRing.isField_iff_maximalIdeal_eq.not.mp hnf, inferInstance⟩, ?_⟩ have hkd : Ring.KrullDimLE 1 S := by rw [Ring.krullDimLE_iff, hdim] norm_num rintro P ⟨hPb, hPp⟩ exact IsLocalRing.eq_maximalIdeal (hPp.isMaximal_of_ne_bot hPb) exact ((IsDiscreteValuationRing.TFAE S hnf).out 4 1).mp h3 -- In particular `V` is a valuation ring of `F`. have hplace : ∀ z : F, z ∈ S.toSubring ∨ z⁻¹ ∈ S.toSubring := by intro z rcases ValuationRing.isInteger_or_isInteger S z with ⟨s, hs⟩ | ⟨s, hs⟩ · left; rw [← hs]; exact s.2 · right; rw [← hs]; exact s.2 let V : ValuationSubring F := placeValuationSubring A p hplace have hAV : ∀ a : A, (a : F) ∈ V.toSubring := fun a ↦ ofField_mem_algebraMap_v A p hplace a have hpV : ∀ a : A, (⟨a, hAV a⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ↔ a ∈ p := fun a ↦ ofField_mem_maximalIdeal_iff_v A p hplace a have hsurj : ∀ z : ResidueField V.toSubring, ∃ (a b : A), b ∉ p ∧ z = residue V.toSubring ⟨a, hAV a⟩ / residue V.toSubring ⟨b, hAV b⟩ := fun z ↦ ofField_residue_surjective_v A p hplace z have hkV : ∀ c : k, algebraMap k F c ∈ V.toSubring := fun c ↦ by rw [show algebraMap k F c = algebraMap A F (algebraMap k A c) from (IsScalarTower.algebraMap_apply k A F c)] exact hAV (algebraMap k A c) have hqV : ∀ a, Y a / Y j ∈ V.toSubring := fun a ↦ by simpa [s, div_eq_mul_inv] using hAV ⟨Y a * s, hnormA a⟩ -- `Frac V = F`, since `A ⊆ V ⊆ F` and `Frac A = F`. have _hfracV : IsFractionRing V.toSubring F := IsFractionRing.of_field V.toSubring F fun z ↦ by obtain ⟨a, b, hb, hab⟩ := IsFractionRing.div_surjective A z exact ⟨⟨a, hAV a⟩, ⟨b, hAV b⟩, hab.symm⟩ letI : Algebra k V.toSubring := ((algebraMap k F).codRestrict V.toSubring hkV).toAlgebra letI : IsScalarTower k V.toSubring F := IsScalarTower.of_algebraMap_eq fun _ ↦ rfl have hxV : x i ∈ V.toSubring := hAV xA have hxm : (⟨x i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := (hpV xA).2 hxp have hVdvr : IsDiscreteValuationRing V.toSubring := hSdvr have hqA : ∀ a, Y a / Y j ∈ A := fun a ↦ by simpa [s, div_eq_mul_inv] using hnormA a refine ⟨Lemma510Place.mk A p hplace hVdvr hqA hyA hxp ?_ ?_⟩ · -- (b) `A/𝔭_D` is integral over the image of `C_0`, whose residues are `res q_i`. let c' : (Fin (n + 1) ⊕ Unit) → F := fun a ↦ match a with | Sum.inl b => Y b * s | Sum.inr _ => x i have hc'A : ∀ a, c' a ∈ A := fun a ↦ by cases a with | inl b => exact hnormA b | inr _ => exact hyA have hset : Set.range c' = Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i} := by ext z constructor · rintro ⟨a, rfl⟩ cases a with | inl b => exact Or.inl ⟨b, rfl⟩ | inr u => exact Or.inr (Set.mem_singleton _) · rintro (⟨b, rfl⟩ | hz) · exact ⟨Sum.inl b, rfl⟩ · exact ⟨Sum.inr (), (Set.mem_singleton_iff.mp hz).symm⟩ have hBig := Stafford38.Geometry.DivisorialVisibleFrameStage4.stage4_residueField_isAlgebraic_of_isIntegral A p V hAV hpV hsurj c' hc'A (fun a ↦ by rw [hset] exact haint a) have hyres : residue V.toSubring (⟨x i, hxV⟩ : V.toSubring) = 0 := (residue_eq_zero_iff _).2 hxm have hEq : IntermediateField.adjoin k (Set.range fun a ↦ residue V.toSubring ⟨c' a, hAV ⟨c' a, hc'A a⟩⟩) = IntermediateField.adjoin k (Set.range fun a ↦ residue V.toSubring ⟨Y a / Y j, hqV a⟩) := by have hcoord : ∀ b, (⟨c' (Sum.inl b), hAV ⟨c' (Sum.inl b), hc'A (Sum.inl b)⟩⟩ : V.toSubring) = ⟨Y b / Y j, hqV b⟩ := fun b ↦ Subtype.ext (by simp [c', s, div_eq_mul_inv]) apply le_antisymm · apply IntermediateField.adjoin_le_iff.mpr rintro _ ⟨a, rfl⟩ cases a with | inl b => dsimp only rw [hcoord b] exact IntermediateField.subset_adjoin k _ ⟨b, rfl⟩ | inr _ => change residue V.toSubring (⟨x i, hxV⟩ : V.toSubring) ∈ _ rw [hyres] exact zero_mem _ · apply IntermediateField.adjoin_le_iff.mpr rintro _ ⟨b, rfl⟩ dsimp only rw [← hcoord b] exact IntermediateField.subset_adjoin k _ ⟨Sum.inl b, rfl⟩ rw [← hEq] convert hBig using 3 · -- (c) `Ω_{A/K}` is finite and localizes to `Ω_{V/K}` (Hartshorne, Prop. II.8.2A). exact stage7_kaehler_finite_ofField A p hkV
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- **Lemma 5.10(a), `R ⊄ V`.** For any valuation ring `V` of `F` with `x̄_1 ∈ 𝔪_V`: `x̄_1 ∈ R^×` since `1 = ḡ x̄_1` with `ḡ ∈ R`, while `x̄_1 ∈ 𝔪_V`, so `R ⊄ V`. The paper derives `1 = ḡ x̄_1` from `Y ∩ H = ∅` by the Nullstellensatz. -/ theorem lemma_5_10_a_not_R_le {k F : Type*} [Field k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (V : ValuationSubring F) (hxV : x i ∈ V.toSubring) (hxm : (⟨x i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring) (hunit : ∃ g ∈ Algebra.adjoin k (Set.range x), x i * g = 1) : ¬ (Algebra.adjoin k (Set.range x)).toSubring ≤ V.toSubring := by obtain ⟨g, hgR, hxg⟩ := hunit have hy0 : x i ≠ 0 := left_ne_zero_of_mul_eq_one hxg have hinvR : (x i)⁻¹ ∈ Algebra.adjoin k (Set.range x) := by rw [← eq_inv_of_mul_eq_one_right hxg] exact hgR intro hRV rw [mem_maximalIdeal, mem_nonunits_iff] at hxm apply hxm refine IsUnit.of_mul_eq_one (⟨(x i)⁻¹, hRV hinvR⟩ : V.toSubring) ?_ apply Subtype.ext simp [hy0]
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- **Lemma 5.11.** -/ theorem lemma_5_11 {k κ : Type u} [Field k] [CharZero k] [Field κ] [Algebra k κ] {m : ℕ} (hm : 0 < m) (r : Fin (m + 1) → κ) (j : Fin (m + 1)) (hr0 : r 0 = 0) (hr1 : r (Fin.succ ⟨0, hm⟩) = 0) (hrj : r j = 1) (halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range r) : IntermediateField k κ) κ) : ∃ (d' : ℕ) (rows : Fin d' → Fin (m + 1)), Function.Injective rows ∧ (∀ l, rows l ≠ 0 ∧ rows l ≠ Fin.succ ⟨0, hm⟩ ∧ rows l ≠ j) ∧ LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) ∧ Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = ⊤ ∧ d' = Module.finrank κ (Ω[κ⁄k]) := by classical let axis : Fin (m + 1) := Fin.succ ⟨0, hm⟩ let good : Set (Fin (m + 1)) := {a | a ≠ 0 ∧ a ≠ axis ∧ a ≠ j} let S : Set κ := r '' good -- The excluded residues are `0` and `1`, so `L` is generated by `S`. have hL : IntermediateField.adjoin k (Set.range r) = IntermediateField.adjoin k S := by apply le_antisymm · rw [IntermediateField.adjoin_le_iff] rintro _ ⟨a, rfl⟩ by_cases ha : a ∈ good · exact IntermediateField.subset_adjoin k S ⟨a, ha, rfl⟩ · simp only [good, Set.mem_ofPred_eq, not_and_or, not_not] at ha rcases ha with rfl | rfl | rfl · rw [hr0]; exact zero_mem _ · rw [hr1]; exact zero_mem _ · rw [hrj]; exact one_mem _ · apply IntermediateField.adjoin.mono rintro _ ⟨a, -, rfl⟩ exact ⟨a, rfl⟩ -- `L` is generated by the finite set `S`, which therefore contains a transcendence -- basis `B` of `L/K` (Definition 5.1, closing sentence; Lang, Thm. VIII.1.1). let SL : Set (IntermediateField.adjoin k S) := Subtype.val ⁻¹' S have hSL : IntermediateField.adjoin k SL = ⊤ := by apply IntermediateField.lift_injective (IntermediateField.adjoin k S) rw [IntermediateField.lift_adjoin, IntermediateField.lift_top] congr 1 change Subtype.val '' (Subtype.val ⁻¹' S) = S ext z refine ⟨fun ⟨w, hw, hwz⟩ => hwz ▸ hw, fun hz => ?_⟩ exact ⟨⟨z, IntermediateField.subset_adjoin k S hz⟩, hz, rfl⟩ have hSLfin : SL.Finite := ((Set.toFinite good).image r).preimage Subtype.val_injective.injOn obtain ⟨BL, hBLS, hBL⟩ := (definition_5_1_closing (K := k) (L := IntermediateField.adjoin k S)).2 SL hSLfin hSL -- `κ/L` is algebraic (Lemma 5.10(b)), so `B` is a transcendence basis of `κ/K`. have : Algebra.IsAlgebraic (IntermediateField.adjoin k S) κ := hL ▸ halg let t : Set κ := Set.range (algebraMap (IntermediateField.adjoin k S) κ ∘ ((↑) : BL → _)) have ht : IsTranscendenceBasis k ((↑) : t → κ) := hBL.algebraMap_comp.to_subtype_range have htS : t ⊆ S := by rintro _ ⟨b, rfl⟩ exact hBLS b.2 have htfin : t.Finite := (Set.toFinite good |>.image r).subset htS let : Fintype t := htfin.fintype -- Matsumura, Thm. 26.5: the transcendence basis is a differential basis. have hrank : Module.finrank κ (Ω[κ⁄k]) = Fintype.card t := Stafford38.Geometry.kaehlerFinrankOfTranscendenceBasis k κ t (↑) ht have hspanT : Submodule.span κ (Set.range fun x : t ↦ KaehlerDifferential.D k κ (x : κ)) = ⊤ := Stafford38.Geometry.KaehlerDVRVisibility.kaehler_span_of_isAlgebraic_adjoin (fun x : t ↦ (x : κ)) (by simpa using (definition_5_1_g_h ht).1) -- Index the basis elements by residues `q_i(0)` with `i ∉ {0, 1, j}`. have hpre : ∀ x : t, ∃ a ∈ good, r a = x := fun x ↦ htS x.2 choose idx hidx hridx using hpre let e := Fintype.equivFin t let rows : Fin (Fintype.card t) → Fin (m + 1) := fun l ↦ idx (e.symm l) have hrows : ∀ l, r (rows l) = (e.symm l : κ) := fun l ↦ hridx _ have hfam : (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = (fun x : t ↦ KaehlerDifferential.D k κ (x : κ)) ∘ e.symm := by funext l simp [hrows] have hspan : Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = ⊤ := by rw [hfam, Set.range_comp, e.symm.surjective.range_eq, Set.image_univ, hspanT] have hli : LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) := linearIndependent_of_top_le_span_of_card_eq_finrank hspan.ge (by rw [Fintype.card_fin, hrank]) refine ⟨Fintype.card t, rows, ?_, fun l ↦ hidx _, hli, hspan, hrank.symm⟩ intro l l' h apply e.symm.injective apply Subtype.ext rw [← hrows, ← hrows, h]
/-- **Lemma 5.12(a), `dim_F Ω_{F/K} ≤ d' + 1`.** For a discrete valuation ring `V ⊇ K` with fraction field `F`, residue field `κ` and `Ω_{V/K}` finitely generated: the second fundamental exact sequence for `V → κ` gives `dim_κ (κ ⊗_V Ω_{V/K}) ≤ d' + 1`, Nakayama's lemma gives that many generators of `Ω_{V/K}`, and `F ⊗_V Ω_{V/K} = Ω_{F/K}` bounds `dim_F Ω_{F/K}`. -/ theorem lemma_5_12_a_bound {F : Type u} [Field F] [Algebra V F] [IsFractionRing V F] [Algebra k F] [IsScalarTower k V F] [Module.Finite V (Ω[V⁄k])] : Module.finrank F (Ω[F⁄k]) ≤ Module.finrank (ResidueField V) (Ω[ResidueField V⁄k]) + 1 := by obtain ⟨f, hf⟩ := exists_kaehler_generators (k := k) (V := V) have h := finrank_fraction_kaehler_le_of_span (F := F) f hf rw [Fintype.card_fin] at h exact h.trans finrank_residue_tensor_kaehler_le
/-- **Lemma 5.12(a), cone part.** `dim 𝒞 = dim Θ_y + 1`. -/ theorem lemma_5_12_a_cone (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) : Module.finrank L (tangentCone q I) = Module.finrank L (tangentSpaceAt q I) + 1 := by classical let f : tangentCone q I →ₗ[L] tangentSpaceAt q I := (rho q).restrict (p := tangentCone q I) (q := tangentSpaceAt q I) (fun v hv ↦ hv) have hf : Function.Surjective f := by rintro ⟨w, hw⟩ obtain ⟨v, rfl⟩ := rho_surjective q hq0 w exact ⟨⟨v, hw⟩, rfl⟩ have hq : q ∈ tangentCone q I := by change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _ have hker : Module.finrank L (LinearMap.ker f) = 1 := by have hq' : (⟨q, hq⟩ : tangentCone q I) ≠ 0 := by intro h exact hq0 (congrFun (congrArg Subtype.val h) 0) rw [finrank_eq_one_iff_of_nonzero' (⟨⟨q, hq⟩, ?_⟩ : LinearMap.ker f) ?_] · rintro ⟨⟨v, hv⟩, hvk⟩ have hv0 : v ∈ LinearMap.ker (rho q) := by rw [LinearMap.mem_ker] exact congrArg Subtype.val (LinearMap.mem_ker.mp hvk) rw [ker_rho q hq0, Submodule.mem_span_singleton] at hv0 obtain ⟨c, rfl⟩ := hv0 exact ⟨c, rfl⟩ · rw [LinearMap.mem_ker] apply Subtype.ext change rho q q = 0 rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q · intro h exact hq' (congrArg Subtype.val h) have := LinearMap.finrank_range_add_finrank_ker f rw [LinearMap.range_eq_top.mpr hf, finrank_top, hker] at this omega
/-- **Lemma 5.12(b).** If the equations vanish at `y`, then `δq ∈ 𝒞` for every `K`-derivation `δ` of `L`, and `q ∈ 𝒞`. -/ theorem lemma_5_12_b (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) (hbase : ∀ f ∈ I, MvPolynomial.eval₂ (algebraMap k L) (dehomogenizedPoint q) f = 0) : (∀ δ : Derivation k L L, (fun a ↦ δ (q a)) ∈ tangentCone q I) ∧ q ∈ tangentCone q I := by constructor · intro δ -- `δ(f(y)) = df_y · δy`, so `δy ∈ Θ_y`. have htan := Stafford38.Geometry.PowerSeriesArcTangency.derivationVector_mem_zariskiTangentSpace_of_eval₂_eq_zero δ I (dehomogenizedPoint q) hbase -- Quotient rule: `ρ(δq) = δy`. have hquot : rho q (fun a ↦ δ (q a)) = fun i ↦ δ (dehomogenizedPoint q i) := by funext i rw [rho_apply, dehomogenizedPoint, Derivation.leibniz_div] simp only [smul_eq_mul] field_simp change rho q (fun a ↦ δ (q a)) ∈ tangentSpaceAt q I rw [hquot] exact htan · change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _
/-- **Lemma 5.13.** A row vector annihilating `𝒞` has its affine part `(ℓ_1, …, ℓ_n)` in `span {df_y : f ∈ 𝔭}`. -/ theorem lemma_5_13 (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) (ell : Fin (n + 1) → L) (hell : ∀ v ∈ tangentCone q I, ∑ a, ell a * v a = 0) : coordinateCovector (fun i : Fin n ↦ ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint q) (I.map (MvPolynomial.map (algebraMap k L))) := by rw [← affineConormalSpace_eq_equationCovectorSpan, affineConormalSpace, Submodule.mem_dualAnnihilator] intro w hw -- `ρ(𝒞) = Θ_y`. have hw' : w ∈ (tangentCone q I).map (rho q) := by rw [rho_map_tangentCone q hq0] exact hw obtain ⟨v, hv, rfl⟩ := hw' have hq : q ∈ tangentCone q I := by change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _ -- `ℓ(q) = 0` and `ℓ(v) = 0` give `Σ ℓ_i ρ(v)_i = 0`. have hellq := hell q hq have hellv := hell v hv rw [Fin.sum_univ_succ] at hellq hellv rw [coordinateCovector_apply] simp only [rho_apply] have hsum : ∑ i : Fin n, ell i.succ * ((v i.succ * q 0 - q i.succ * v 0) / q 0 ^ 2) = (q 0 * ∑ i : Fin n, ell i.succ * v i.succ - v 0 * ∑ i : Fin n, ell i.succ * q i.succ) / q 0 ^ 2 := by rw [Finset.mul_sum, Finset.mul_sum, ← Finset.sum_sub_distrib, Finset.sum_div] refine Finset.sum_congr rfl fun i _ ↦ ?_ ring rw [hsum] have h1 : ∑ i : Fin n, ell i.succ * v i.succ = -(ell 0 * v 0) := by linear_combination hellv have h2 : ∑ i : Fin n, ell i.succ * q i.succ = -(ell 0 * q 0) := by linear_combination hellq rw [h1, h2] ring_nf
/-- The original-prime coordinate axis lies in the smooth conormal fibre closure by splitting into the direct algebraic case and the retained-witness route. -/ theorem coordinate_axis_mem_smooth_fibre_closure {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := by rcases Stafford38.Geometry.OriginalPrimeCoordinateAvoidanceWitness.coordinate_axis_or_visible_frame_of_avoidance hm I havoid with h | hw · exact h · obtain ⟨w⟩ := hw exact axis_mem_smoothConormalFibreProjection_closure_of_groundPointOutput hm I w (Stafford38.Geometry.ActualSameWitnessGroundPointCompletion.exists_actual_same_witness_groundpoint_chart hm I w) (Stafford38.Geometry.exists_nonzero_smooth_away_quotient I.asIdeal)
theorem coordinate_axis_mem_smooth_fibre_closure {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (I : PrimeSpectrum (MvPolynomial (Fin m) k)) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k I.asIdeal, y ⟨0, hm⟩ ≠ 0) : (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection I.asIdeal)) := by exact Stafford38.Geometry.SameWitness.coordinate_axis_mem_smooth_fibre_closure hm I havoid
Variables and notation
Source ↗open Stafford38.Geometry.GeneralAsymptoticLaurentAxisSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗universe u
Comparison notes
theorem_5_7_i_paper is the printed proof of (i), one declaration per paragraph, in the printed order. The algebraic case is step_coordinate_algebraic, the first paragraph: g(x₁) = λ∏(x₁ − c_i), a factor x₁ − c ∈ 𝔭 with c ≠ 0, and dx₁ = d(x₁ − c) conormal at a smooth point of Y (exists_smoothAffinePoint), so (1, 0, …, 0) ∈ N(Y). In the transcendental case the discrete valuation ring V comes from Lemmas 5.9 and 5.10 (generalDivisorialVisibleFrameWithKaehlerFinite through LaneC.divisorialVisibleFrameExistenceWithChart, which calls lemma_5_9 and lemma_5_10; the column is q_a = x̄_a/x̄_j for the chart j of Lemma 5.9), keeping Lemma 5.10(c), Ω_{V/K} finitely generated. Then step_completion (V̂ ≅ κ[[t]], F ↪ κ((t))), step_orders (0 < a < b, j ∉ {0, 1}; a > 0 by Lemma 5.10(a) in step_a_pos), step_generic_point (equation (generic)), Lemma 5.11 at V (witness_lemma_5_11), step_dual_derivations (D_k, Z(t)), step_lambda_w_tau (λ, w = q′ − Zλ = t^c τ), step_M_minor (M(0) has full column rank; a unit minor M_ι), Lemma 5.12(a) at V with m ≤ d′ + 1 by the second fundamental sequence and Nakayama (witness_lemma_5_12_5_13_paper, witness_lemma_5_12_a_bound), step_M_spans_cone (Lemmas 5.12(a), (b)), step_lift_ell_minor (ℓ_υ = u_υ, ℓ_ι = −u_υM_υM_ι⁻¹, ℓM = 0, ℓ(0) = u), step_ell_conormal (Lemma 5.13) and step_polynomial_limit (the final paragraph). One point is done differently: the place is followed by a field E ⊆ V over which κ is finite, which Lean uses to obtain the coefficient field of V̂ (Matsumura Thm. 28.3). The last paragraph, (ii), is step_dir_of_fibre. The earlier route's GeneralAsymptoticConormal.coordinate_axis_mem_smooth_fibre_closure proves (i) by a second, chart-based route through SameWitness.coordinate_axis_mem_smooth_fibre_closure.
Review notes
Lemma 5.9 — A valuation ring at infinity in which x̄₁ vanishes
/-- **Lemma 5.9.** If `x̄_1` is transcendental over `K`, there are a valuation ring `K ⊆ W ⊊ F` with `x̄_1 ∈ 𝔪_W` and an index `j` with `x̄_i / x̄_j ∈ W` for all `0 ≤ i ≤ n`. -/ theorem lemma_5_9 {k F : Type*} [Field k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (htrans : Transcendental k (x i)) : ∃ (W : ValuationSubring F) (j : Fin (n + 1)), (∀ c : k, algebraMap k F c ∈ W) ∧ W ≠ ⊤ ∧ (∃ hx : x i ∈ W, (⟨x i, hx⟩ : W.toSubring) ∈ maximalIdeal W.toSubring) ∧ homogeneousCoordinates x j ≠ 0 ∧ ∀ a, homogeneousCoordinates x a / homogeneousCoordinates x j ∈ W := by classical -- `W` dominates `K[x̄_1]` localized at `x̄_1` (Stacks, Tag 00IA). obtain ⟨W, hkW, hxW, hxinvW⟩ := stage1_exists_valuationSubring_of_transcendental (x i) htrans have hx0 : x i ≠ 0 := fun h ↦ htrans (h ▸ isAlgebraic_zero) -- Definition 5.3(a): `𝔪_W = W ∖ W^×` have hxmax : (⟨x i, hxW⟩ : W.toSubring) ∈ maximalIdeal W.toSubring := by rw [(definition_5_3.2.1 W).2] rintro ⟨u, hu⟩ apply hxinvW have hmul : x i * ((u⁻¹ : W.toSubringˣ) : W.toSubring) = 1 := by have := congrArg Subtype.val (Units.mul_inv u) simpa [hu] using this have : (x i)⁻¹ = ((u⁻¹ : W.toSubringˣ) : W.toSubring) := (eq_inv_of_mul_eq_one_right hmul).symm rw [this] exact ((u⁻¹ : W.toSubringˣ) : W.toSubring).2 have hWtop : W ≠ ⊤ := by rintro rfl exact hxinvW (ValuationSubring.mem_top _) -- Among the nonzero `x̄_i` choose one that is maximal for `a ≼ b ↔ x̄_a / x̄_b ∈ W`. let Y := homogeneousCoordinates x let S : Set (Fin (n + 1)) := {a | Y a ≠ 0} have hS : S.Nonempty := ⟨0, by simp [S, Y]⟩ letI : LE (Fin (n + 1)) := ⟨fun a b ↦ Y b ≠ 0 ∧ Y a / Y b ∈ W⟩ have htr : IsTrans (Fin (n + 1)) LE.le := by constructor intro a b c hab hbc rcases hab with ⟨hb, hab⟩ rcases hbc with ⟨hc, hbc⟩ refine ⟨hc, ?_⟩ have := W.toSubring.mul_mem hab hbc rwa [div_mul_div_cancel₀ hb] at this letI : IsTrans (Fin (n + 1)) LE.le := htr obtain ⟨j, hjS, hjmax⟩ := S.toFinite.exists_maximalFor id S hS refine ⟨W, j, hkW, hWtop, ⟨hxW, hxmax⟩, hjS, fun a ↦ ?_⟩ by_cases ha : Y a = 0 · change Y a / Y j ∈ W rw [ha, zero_div] exact W.zero_mem -- Definition 5.3(a): `a/b ∈ W` or `b/a ∈ W` for nonzero `a, b` rcases (definition_5_3.1 W.toSubring).2 ⟨W, rfl⟩ (Y a / Y j) (div_ne_zero ha hjS) with h | h · exact h · rw [inv_div] at h exact (hjmax (show a ∈ S from ha) ⟨ha, h⟩).2
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- **Definition 5.3.** Let `F` be a field. (a) A subring `W ⊆ F` is a valuation ring of `F` iff `a ∈ W` or `a⁻¹ ∈ W` for every `a ∈ F^×`; it is a local ring with maximal ideal `𝔪_W = W ∖ W^×`. (b) A valuation ring `V` is a discrete valuation ring iff it is Noetherian and `V ≠ F`; then `𝔪_V` is generated by some `ϖ ∈ V`, and every `f ∈ F^×` is `f = ϖ^k u` with `u ∈ V^×` and a unique `k = ord_V(f) ∈ ℤ`. (c) For a field `κ`, `κ((t))` (`LaurentSeries κ`) is the fraction field of `κ[[t]]` (`PowerSeries κ`). -/ theorem definition_5_3 {F : Type u} [Field F] : (∀ S : Subring F, (∀ a : F, a ≠ 0 → a ∈ S ∨ a⁻¹ ∈ S) ↔ ∃ W : ValuationSubring F, W.toSubring = S) ∧ (∀ W : ValuationSubring F, IsLocalRing W.toSubring ∧ ∀ w : W.toSubring, w ∈ maximalIdeal W.toSubring ↔ ¬ IsUnit w) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring ↔ IsNoetherianRing V.toSubring ∧ V ≠ ⊤) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring → ∃ ϖ : V.toSubring, maximalIdeal V.toSubring = Ideal.span {ϖ} ∧ ∀ f : F, f ≠ 0 → ∃! k : ℤ, ∃ u : V.toSubringˣ, f = (ϖ : F) ^ k * ((u : V.toSubring) : F)) ∧ ∀ (κ : Type u) [Field κ], IsFractionRing (PowerSeries κ) (LaurentSeries κ) := by refine ⟨fun S => ⟨fun h => ?_, ?_⟩, fun W => ⟨inferInstance, fun w => ?_⟩, fun V => ⟨fun hV => ?_, fun ⟨hN, hV⟩ => ?_⟩, fun V hV => ?_, fun κ _ => inferInstance⟩ · refine ⟨⟨S, fun a => ?_⟩, rfl⟩ by_cases ha : a = 0 · exact Or.inl (ha ▸ S.zero_mem) · exact h a ha · rintro ⟨W, rfl⟩ a _ exact W.mem_or_inv_mem a · rw [mem_maximalIdeal, mem_nonunits_iff] · refine ⟨inferInstance, fun hVtop => IsDiscreteValuationRing.not_isField V.toSubring ?_⟩ exact (step_5_3_isField_iff V).2 hVtop · have hfield : ¬ IsField V.toSubring := fun h => hV ((step_5_3_isField_iff V).1 h) have := step_5_3_valuationRing V exact ((IsDiscreteValuationRing.TFAE V.toSubring hfield).out 1 2).2 this · obtain ⟨ϖ, hϖ⟩ := IsDiscreteValuationRing.exists_irreducible V.toSubring refine ⟨ϖ, hϖ.maximalIdeal_eq, fun f hf => ?_⟩ obtain ⟨k, u, hku⟩ := step_5_3_ord_exists V hϖ hf exact ⟨k, ⟨u, hku⟩, fun l ⟨u', hl⟩ => step_5_3_ord_unique V hϖ (hl.symm.trans hku)⟩
More Lean proofs and definitions
/-- The place at infinity of the paper's proof of Theorem 5.7: the valuation ring and chart of Lemma 5.9, the discrete valuation ring `V = A_{𝔭_D}` of Lemma 5.10, and a coefficient field `E` over which `κ` is finite (the input to the coefficient field of the completion, Matsumura Thm. 28.3). -/ theorem divisorialVisibleFrameExistenceWithChart (k K : Type u) [Field k] [CharZero k] [Field K] [Algebra k K] (r : ℕ) (y : Fin r → K) (i : Fin r) (hgen : IntermediateField.adjoin k (Set.range y) = ⊤) (hx : Transcendental k (y i)) : ∃ (E : IntermediateField k K) (V : ValuationSubring K) (hEV : ∀ z : E, (z : K) ∈ V.toSubring) (hVdvr : IsDiscreteValuationRing V.toSubring) (hxV : y i ∈ V.toSubring), letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra letI : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra Transcendental E (y i) ∧ (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ∧ Module.Finite E (ResidueField V.toSubring) ∧ Module.Finite V.toSubring (Ω[V.toSubring⁄k]) ∧ (∀ (s : K) (q : Fin (r + 1) → V.toSubring), (∀ a, (q a : K) = s * Fin.cases 1 y a) → (∃ j, q j = 1) → ¬ IsUnit (q 0) → Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun j : Fin r ↦ residue V.toSubring (q (Fin.succ j))) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) ∧ -- The chart `j` of Lemma 5.9, with `q_a = x̄_a / x̄_j ∈ V` (Lemma 5.10(a)). ∃ j, (Fin.cases 1 y j : K) ≠ 0 ∧ ∀ a, (Fin.cases 1 y a : K) / Fin.cases 1 y j ∈ V.toSubring := by -- Lemma 5.9. obtain ⟨W, j, hkW, -, hxW, hj, hqW⟩ := Stafford38.Paper.Section5.lemma_5_9 y i hx -- Lemma 5.10. obtain ⟨L⟩ := Stafford38.Paper.Section5.lemma_5_10 y i hgen hx W j hkW hxW hj hqW let A := L.A let p := L.p let V : ValuationSubring K := L.V have hVdvr : IsDiscreteValuationRing V.toSubring := L.isDVR letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing have hAV : ∀ a : A, (a : K) ∈ V.toSubring := fun a => ofField_mem_algebraMap_v A p L.place a have hpV : ∀ a : A, (⟨a, hAV a⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ↔ a ∈ p := fun a => ofField_mem_maximalIdeal_iff_v A p L.place a have hsurj : ∀ z : ResidueField V.toSubring, ∃ (a b : A), b ∉ p ∧ z = residue V.toSubring ⟨a, hAV a⟩ / residue V.toSubring ⟨b, hAV b⟩ := fun z => ofField_residue_surjective_v A p L.place z have hy0 : y i ≠ 0 := fun h => hx (h ▸ isAlgebraic_zero) have hxV : y i ∈ V.toSubring := hAV ⟨y i, L.x_mem⟩ have hxm : (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := (hpV ⟨y i, L.x_mem⟩).2 L.x_mem_p obtain ⟨E, hEV, hEfin⟩ := stage5_of_verified A p V hAV hpV hsurj letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra have htrans : Transcendental E (y i) := stage6_transcendental_of_mem_maximalIdeal E V hEV (y i) hy0 hxV hxm have hk : ∀ c : k, algebraMap k K c ∈ V.toSubring := fun c => hAV (algebraMap k A c) let cod : Algebra k V.toSubring := ((algebraMap k K).codRestrict _ hk).toAlgebra let ground : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra have hcod_ground : cod = ground := by apply Algebra.algebra_ext intro c apply Subtype.ext rfl letI : Algebra k V.toSubring := ground letI : IsScalarTower k V.toSubring K := IsScalarTower.of_algebraMap_eq fun c => by change algebraMap k K c = algebraMap k K c rfl have hkaehler : Module.Finite V.toSubring (Ω[V.toSubring⁄k]) := by have h := L.kaehler_finite change (letI : Algebra k V.toSubring := cod; Module.Finite V.toSubring (Ω[V.toSubring⁄k])) at h rw [hcod_ground] at h exact h -- The normalized column `q_a = x̄_a / x̄_j` of Lemma 5.10. let s : K := (Fin.cases 1 y j : K)⁻¹ have hjs : (Fin.cases 1 y j : K) * s = 1 := mul_inv_cancel₀ hj let c : Fin (r + 1) → V.toSubring := fun a => ⟨(Fin.cases 1 y a : K) / Fin.cases 1 y j, hAV ⟨_, L.q_mem a⟩⟩ have hc : ∀ a, (c a : K) = (Fin.cases 1 y a : K) * s := fun a => div_eq_mul_inv _ _ have hcj : c j = 1 := Subtype.ext (div_self hj) have halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring) := by have h := L.residue_algebraic change (letI : Algebra k V.toSubring := cod; Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) at h rw [hcod_ground] at h exact h refine ⟨E, V, hEV, hVdvr, hxV, htrans, hxm, hEfin, hkaehler, ?_, j, hj, fun a ↦ (c a).2⟩ intro s' q hq hq1 hq0 by_cases hj0 : j = 0 · obtain ⟨j', hj'⟩ := hq1 exfalso have hsone : s = 1 := by simpa [hj0] using hjs have hprod : ∀ a, q a = q 0 * c a := by intro a apply Subtype.ext change (q a : K) = (q 0 : K) * (c a : K) rw [hq a, hq 0, hc a] simp [hsone] have hq0res : residue V.toSubring (q 0) = 0 := by by_contra hnz exact hq0 ((residue_ne_zero_iff_isUnit _).1 hnz) have hqjres : residue V.toSubring (q j') = 0 := by rw [hprod j', map_mul, hq0res, zero_mul] have hqjunit : IsUnit (q j') := by rw [hj']; exact isUnit_one exact ((residue_ne_zero_iff_isUnit _).2 hqjunit) hqjres · exact DivisorialVisibleFrameStage4.stage4'_isAlgebraic_of_normalized_column V y j hj0 s c hc hcj halg s' q hq hq1 hq0
Comparison notes
lemma_5_9: if x̄₁ is transcendental over K, there are a valuation ring W of F with K ⊆ W ≠ F and x̄₁ ∈ 𝔪_W, and an index j with x̄_j ≠ 0 and x̄_i/x̄_j ∈ W for 0 ≤ i ≤ n, where x̄₀ = 1. The proof is the printed one: a valuation ring dominating K[x̄₁] localized at x̄₁ (Stacks Tag 00IA), then j maximal for the preorder x̄_a/x̄_b ∈ W; Definition 5.3(a) enters through definition_5_3 (𝔪_W = W ∖ W^×, and a ∈ W or a⁻¹ ∈ W). The proof of Theorem 5.7 calls it through LaneC.divisorialVisibleFrameExistenceWithChart and takes the column q_a = x̄_a/x̄_j with its chart j.
Review notes
Lemma 5.10 — A divisorial place with a vanishing first coordinate
- for all , and ;
- is algebraic over ;
- is a finitely generated -module.
Put . This is a Noetherian local domain. It is normal, since a localization of an integrally closed domain is integrally closed [Atiyah 1969, Prop 5.12], and its dimension equals . A Noetherian local normal domain of dimension one is a discrete valuation ring [Matsumura 1986, Thm. 11.2], so is a discrete valuation ring.
It remains to check the listed properties.
- : we have , and the domain embeds into its localization .
- : since and .
- for all : since .
- : since and . , so by Theorem 2.11, i.e. for some , . Modulo this gives , so . If , then , contradicting .
(b) Since and , we have , and the residue map restricts to the quotient map . As is integral over and , the ring is integral over the image of in , which is . So every element of , and hence of its fraction field , is algebraic over .
(c) is a finitely generated -algebra, so is a finite -module, and since is a localization of [hartshorne, Prop II.8.2A]. ∎
/-- **Lemma 5.10.** For `q_i = x̄_i / x̄_j` from Lemma 5.9 there is a discrete valuation ring `K ⊆ V ⊊ F` (`Frac V = F` holds for every valuation subring) with (a) `q_i ∈ V` and `x̄_1 ∈ 𝔪_V`, (b) `κ` algebraic over `K(res q_0, …, res q_n)`, (c) `Ω_{V/K}` finitely generated; the conclusions are the fields of `Lemma510Place`, which also records `V = A_{𝔭_D}` from the proof. The clause `R ⊄ V` of (a) is `lemma_5_10_a_not_R_le`, stated for every valuation ring with `x̄_1 ∈ 𝔪_V`, which is how the proof of Theorem 5.7 uses it. -/ theorem lemma_5_10 {k F : Type u} [Field k] [CharZero k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (hgen : IntermediateField.adjoin k (Set.range x) = ⊤) (htrans : Transcendental k (x i)) (W : ValuationSubring F) (j : Fin (n + 1)) (hkW : ∀ c : k, algebraMap k F c ∈ W) (hxW : ∃ hx : x i ∈ W, (⟨x i, hx⟩ : W.toSubring) ∈ maximalIdeal W.toSubring) (hj : homogeneousCoordinates x j ≠ 0) (hqW : ∀ a, homogeneousCoordinates x a / homogeneousCoordinates x j ∈ W) : Nonempty (Lemma510Place k x i j) := by classical let Y := homogeneousCoordinates x obtain ⟨hxiW, hxmW⟩ := hxW have hy0 : x i ≠ 0 := fun h ↦ htrans (h ▸ isAlgebraic_zero) have hxiInvW : (x i)⁻¹ ∉ W := by intro h rw [(definition_5_3.2.1 W).2] at hxmW apply hxmW refine IsUnit.of_mul_eq_one (⟨(x i)⁻¹, h⟩ : W.toSubring) ?_ apply Subtype.ext simp [hy0] let s : F := (Y j)⁻¹ have hs : s ≠ 0 := inv_ne_zero hj have hjs : Y j * s = 1 := mul_inv_cancel₀ hj have hYsW : ∀ a, Y a * s ∈ W := fun a ↦ by simpa [s, div_eq_mul_inv] using hqW a -- `C_0 = K[q_0, …, q_n, x̄_1] ⊆ W`, and `A` is its integral closure in `F`. let C₀ : Subalgebra k F := Algebra.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) let SW : Subalgebra k F := { W.toSubring with algebraMap_mem' := hkW } have hC₀W : C₀ ≤ SW := by apply Algebra.adjoin_le rintro z (⟨a, rfl⟩ | hz) · exact hYsW a · rw [Set.mem_singleton_iff.mp hz] exact hxiW have htop : IntermediateField.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) = ⊤ := by apply top_unique rw [← hgen] apply IntermediateField.adjoin_le_iff.2 rintro _ ⟨a, rfl⟩ apply IntermediateField.mem_adjoin_iff_div.mpr refine ⟨Y (Fin.succ a) * s, ?_, s, ?_, ?_⟩ · exact Algebra.subset_adjoin (R := k) (A := F) (Set.mem_union_left _ (Set.mem_range_self _)) · have h0 : Y 0 * s ∈ C₀ := Algebra.subset_adjoin (R := k) (A := F) (Set.mem_union_left _ (Set.mem_range_self (0 : Fin (n + 1)))) simpa only [Y, homogeneousCoordinates, Fin.cases_zero, one_mul] using h0 · simp only [Y, Fin.cases_succ] exact (eq_div_iff hs).2 rfl letI : IsFractionRing C₀ F := IsFractionRing.of_field C₀ F fun z ↦ by have hz : z ∈ IntermediateField.adjoin k (Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i}) := by rw [htop] exact Set.mem_univ z rw [IntermediateField.mem_adjoin_iff_div] at hz rcases hz with ⟨a, ha, b, hb, hab⟩ exact ⟨⟨a, ha⟩, ⟨b, hb⟩, hab⟩ let A₀ : Subalgebra C₀ F := integralClosure C₀ F let A : Subalgebra k F := A₀.restrictScalars k have hC₀A : C₀ ≤ A := fun z hz ↦ A₀.algebraMap_mem ⟨z, hz⟩ -- `W` is integrally closed, so `A ⊆ W` (Atiyah–Macdonald, Prop. 5.18). letI : IsIntegrallyClosedIn W.toSubring F := inferInstanceAs (IsIntegrallyClosedIn W F) have hA₀W : A₀.toSubring ≤ W.toSubring := by change (integralClosure C₀ F).toSubring ≤ W.toSubring apply (Subring.integralClosure_subring_le_iff (S := C₀) (T := W.toSubring)).2 exact hC₀W have hyA : x i ∈ A := A₀.algebraMap_mem ⟨x i, Algebra.subset_adjoin (Set.mem_union_right _ (Set.mem_singleton _))⟩ have hnormA : ∀ a, Y a * s ∈ A := fun a ↦ A₀.algebraMap_mem ⟨Y a * s, Algebra.subset_adjoin (Set.mem_union_left _ (Set.mem_range_self a))⟩ have hinvA : (x i)⁻¹ ∉ A := fun h ↦ hxiInvW (hA₀W h) -- `A` is a finite `C_0`-module (Hartshorne, Thm. I.3.9A). haveI hC₀ft : Algebra.FiniteType k C₀ := Algebra.FiniteType.adjoin_of_finite (Set.Finite.union (Set.finite_range _) (Set.finite_singleton _)) let D : Subalgebra C₀ (FractionRing C₀) := integralClosure C₀ (FractionRing C₀) haveI : Module.Finite C₀ D := finite_normalization_of_fg_domain k C₀ D let e : FractionRing C₀ ≃ₐ[C₀] F := IsLocalization.algEquiv C₀⁰ (FractionRing C₀) F haveI : Module.Finite C₀ A₀ := Module.Finite.equiv e.mapIntegralClosure.toLinearEquiv have hsub : A.toSubring = A₀.toSubring := rfl let er : A ≃+* A₀ := RingEquiv.subringCongr hsub letI : Algebra C₀ A := (er.symm.toRingHom.comp (algebraMap C₀ A₀)).toAlgebra let ea : A₀ ≃ₐ[C₀] A := AlgEquiv.ofRingEquiv (f := er.symm) (fun _ ↦ rfl) letI : IsScalarTower k C₀ A := IsScalarTower.of_algebraMap_eq fun _ ↦ by apply Subtype.ext rfl letI : IsScalarTower C₀ A F := IsScalarTower.of_algebraMap_eq fun _ ↦ rfl haveI : Module.Finite C₀ A := Module.Finite.equiv ea.toLinearEquiv haveI hAft : Algebra.FiniteType k A := Algebra.FiniteType.trans hC₀ft (inferInstance : Algebra.FiniteType C₀ A) have hAic : IsIntegrallyClosedIn A F := by apply Subring.isIntegrallyClosedIn_iff.mpr intro z hz change z ∈ integralClosure C₀ F exact isIntegral_trans (R := C₀) (A := A) z hz letI : IsFractionRing A F := IsFractionRing.of_field A F fun z ↦ by obtain ⟨a, b, hb, hab⟩ := IsFractionRing.div_surjective C₀ z refine ⟨⟨a, hC₀A a.2⟩, ⟨b, hC₀A b.2⟩, ?_⟩ change z = (a : F) / (b : F) exact hab.symm have haint : ∀ a : A, IsIntegral C₀ (a : F) := fun a ↦ a.2 letI : IsNoetherianRing A := Algebra.FiniteType.isNoetherianRing k A letI : IsIntegrallyClosed A := (isIntegrallyClosed_iff_isIntegrallyClosedIn F).mpr hAic -- `x̄_1` is a nonzero nonunit of `A`; a minimal prime over it has height one (Krull). let xA : A := ⟨x i, hyA⟩ have hxA0 : xA ≠ 0 := fun h ↦ hy0 (congrArg Subtype.val h) have hxAu : ¬ IsUnit xA := by intro hu obtain ⟨z, hz⟩ := isUnit_iff_exists_inv.mp hu have hzK : x i * (z : F) = 1 := congrArg Subtype.val hz apply hinvA rw [show (x i)⁻¹ = (z : F) by simpa using congrArg Inv.inv (eq_inv_of_mul_eq_one_left hzK)] exact z.2 -- `𝔭_D` is minimal among the primes containing `x̄_1`; by Krull's principal ideal -- theorem `ht 𝔭_D ≤ 1`, and `ht 𝔭_D = 1` because `(0)` is the only prime of height `0`. obtain ⟨p, hpmin⟩ := (Ideal.span {xA}).nonempty_minimalPrimes (by simpa [Ideal.span_singleton_eq_top] using hxAu) letI : p.IsPrime := hpmin.isPrime have hxp : xA ∈ p := hpmin.1.2 (Ideal.mem_span_singleton_self xA) have hpheight : p.height = 1 := by refine le_antisymm (Ideal.height_le_one_of_isPrincipal_of_mem_minimalPrimes _ _ hpmin) ?_ have hpne : p ≠ ⊥ := fun h ↦ hxA0 (by simpa [h] using hxp) rw [Order.one_le_iff_ne_zero] simpa [Ideal.height_eq_zero_iff_eq_bot] using hpne -- `V := A_{𝔭_D}` is a Noetherian local domain. let S := placeSubalgebra A p letI : IsLocalization.AtPrime S p := by unfold S placeSubalgebra infer_instance letI : IsLocalRing S := IsLocalization.AtPrime.isLocalRing S p letI : IsNoetherianRing S := IsLocalization.isNoetherianRing p.primeCompl S inferInstance -- It is normal, as a localization of an integrally closed domain (A–M, Prop. 5.12), letI : IsIntegrallyClosed S := isIntegrallyClosed_of_isLocalization S p.primeCompl p.primeCompl_le_nonZeroDivisors -- and its dimension is `ht 𝔭_D = 1`. have hdim : ringKrullDim S = 1 := by rw [IsLocalization.AtPrime.ringKrullDim_eq_height p S, hpheight] rfl -- A Noetherian local normal domain of dimension one is a DVR (Matsumura, Thm. 11.2). have hSdvr : IsDiscreteValuationRing S := by have hnf : ¬ IsField S := fun hf ↦ by letI := hf.toField rw [ringKrullDim_eq_zero_of_isField hf] at hdim exact zero_ne_one hdim have h3 : IsIntegrallyClosed S ∧ ∃! P : Ideal S, P ≠ ⊥ ∧ P.IsPrime := by refine ⟨inferInstance, IsLocalRing.maximalIdeal _, ⟨IsLocalRing.isField_iff_maximalIdeal_eq.not.mp hnf, inferInstance⟩, ?_⟩ have hkd : Ring.KrullDimLE 1 S := by rw [Ring.krullDimLE_iff, hdim] norm_num rintro P ⟨hPb, hPp⟩ exact IsLocalRing.eq_maximalIdeal (hPp.isMaximal_of_ne_bot hPb) exact ((IsDiscreteValuationRing.TFAE S hnf).out 4 1).mp h3 -- In particular `V` is a valuation ring of `F`. have hplace : ∀ z : F, z ∈ S.toSubring ∨ z⁻¹ ∈ S.toSubring := by intro z rcases ValuationRing.isInteger_or_isInteger S z with ⟨s, hs⟩ | ⟨s, hs⟩ · left; rw [← hs]; exact s.2 · right; rw [← hs]; exact s.2 let V : ValuationSubring F := placeValuationSubring A p hplace have hAV : ∀ a : A, (a : F) ∈ V.toSubring := fun a ↦ ofField_mem_algebraMap_v A p hplace a have hpV : ∀ a : A, (⟨a, hAV a⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ↔ a ∈ p := fun a ↦ ofField_mem_maximalIdeal_iff_v A p hplace a have hsurj : ∀ z : ResidueField V.toSubring, ∃ (a b : A), b ∉ p ∧ z = residue V.toSubring ⟨a, hAV a⟩ / residue V.toSubring ⟨b, hAV b⟩ := fun z ↦ ofField_residue_surjective_v A p hplace z have hkV : ∀ c : k, algebraMap k F c ∈ V.toSubring := fun c ↦ by rw [show algebraMap k F c = algebraMap A F (algebraMap k A c) from (IsScalarTower.algebraMap_apply k A F c)] exact hAV (algebraMap k A c) have hqV : ∀ a, Y a / Y j ∈ V.toSubring := fun a ↦ by simpa [s, div_eq_mul_inv] using hAV ⟨Y a * s, hnormA a⟩ -- `Frac V = F`, since `A ⊆ V ⊆ F` and `Frac A = F`. have _hfracV : IsFractionRing V.toSubring F := IsFractionRing.of_field V.toSubring F fun z ↦ by obtain ⟨a, b, hb, hab⟩ := IsFractionRing.div_surjective A z exact ⟨⟨a, hAV a⟩, ⟨b, hAV b⟩, hab.symm⟩ letI : Algebra k V.toSubring := ((algebraMap k F).codRestrict V.toSubring hkV).toAlgebra letI : IsScalarTower k V.toSubring F := IsScalarTower.of_algebraMap_eq fun _ ↦ rfl have hxV : x i ∈ V.toSubring := hAV xA have hxm : (⟨x i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := (hpV xA).2 hxp have hVdvr : IsDiscreteValuationRing V.toSubring := hSdvr have hqA : ∀ a, Y a / Y j ∈ A := fun a ↦ by simpa [s, div_eq_mul_inv] using hnormA a refine ⟨Lemma510Place.mk A p hplace hVdvr hqA hyA hxp ?_ ?_⟩ · -- (b) `A/𝔭_D` is integral over the image of `C_0`, whose residues are `res q_i`. let c' : (Fin (n + 1) ⊕ Unit) → F := fun a ↦ match a with | Sum.inl b => Y b * s | Sum.inr _ => x i have hc'A : ∀ a, c' a ∈ A := fun a ↦ by cases a with | inl b => exact hnormA b | inr _ => exact hyA have hset : Set.range c' = Set.range (fun a : Fin (n + 1) ↦ Y a * s) ∪ {x i} := by ext z constructor · rintro ⟨a, rfl⟩ cases a with | inl b => exact Or.inl ⟨b, rfl⟩ | inr u => exact Or.inr (Set.mem_singleton _) · rintro (⟨b, rfl⟩ | hz) · exact ⟨Sum.inl b, rfl⟩ · exact ⟨Sum.inr (), (Set.mem_singleton_iff.mp hz).symm⟩ have hBig := Stafford38.Geometry.DivisorialVisibleFrameStage4.stage4_residueField_isAlgebraic_of_isIntegral A p V hAV hpV hsurj c' hc'A (fun a ↦ by rw [hset] exact haint a) have hyres : residue V.toSubring (⟨x i, hxV⟩ : V.toSubring) = 0 := (residue_eq_zero_iff _).2 hxm have hEq : IntermediateField.adjoin k (Set.range fun a ↦ residue V.toSubring ⟨c' a, hAV ⟨c' a, hc'A a⟩⟩) = IntermediateField.adjoin k (Set.range fun a ↦ residue V.toSubring ⟨Y a / Y j, hqV a⟩) := by have hcoord : ∀ b, (⟨c' (Sum.inl b), hAV ⟨c' (Sum.inl b), hc'A (Sum.inl b)⟩⟩ : V.toSubring) = ⟨Y b / Y j, hqV b⟩ := fun b ↦ Subtype.ext (by simp [c', s, div_eq_mul_inv]) apply le_antisymm · apply IntermediateField.adjoin_le_iff.mpr rintro _ ⟨a, rfl⟩ cases a with | inl b => dsimp only rw [hcoord b] exact IntermediateField.subset_adjoin k _ ⟨b, rfl⟩ | inr _ => change residue V.toSubring (⟨x i, hxV⟩ : V.toSubring) ∈ _ rw [hyres] exact zero_mem _ · apply IntermediateField.adjoin_le_iff.mpr rintro _ ⟨b, rfl⟩ dsimp only rw [← hcoord b] exact IntermediateField.subset_adjoin k _ ⟨Sum.inl b, rfl⟩ rw [← hEq] convert hBig using 3 · -- (c) `Ω_{A/K}` is finite and localizes to `Ω_{V/K}` (Hartshorne, Prop. II.8.2A). exact stage7_kaehler_finite_ofField A p hkV
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- **Lemma 5.10(a), `R ⊄ V`.** For any valuation ring `V` of `F` with `x̄_1 ∈ 𝔪_V`: `x̄_1 ∈ R^×` since `1 = ḡ x̄_1` with `ḡ ∈ R`, while `x̄_1 ∈ 𝔪_V`, so `R ⊄ V`. The paper derives `1 = ḡ x̄_1` from `Y ∩ H = ∅` by the Nullstellensatz. -/ theorem lemma_5_10_a_not_R_le {k F : Type*} [Field k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (V : ValuationSubring F) (hxV : x i ∈ V.toSubring) (hxm : (⟨x i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring) (hunit : ∃ g ∈ Algebra.adjoin k (Set.range x), x i * g = 1) : ¬ (Algebra.adjoin k (Set.range x)).toSubring ≤ V.toSubring := by obtain ⟨g, hgR, hxg⟩ := hunit have hy0 : x i ≠ 0 := left_ne_zero_of_mul_eq_one hxg have hinvR : (x i)⁻¹ ∈ Algebra.adjoin k (Set.range x) := by rw [← eq_inv_of_mul_eq_one_right hxg] exact hgR intro hRV rw [mem_maximalIdeal, mem_nonunits_iff] at hxm apply hxm refine IsUnit.of_mul_eq_one (⟨(x i)⁻¹, hRV hinvR⟩ : V.toSubring) ?_ apply Subtype.ext simp [hy0]
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
More Lean proofs and definitions
/-- The discrete valuation ring of Lemma 5.10 together with the construction `V = A_{𝔭_D}` of its proof: `A` is the integral closure in `F` of `C_0 = K[q_0, …, q_n, x̄_1]` and `𝔭_D` is a height-one prime of `A` containing `x̄_1`. The fields `isDVR`, `q_mem`, `x_mem_p`, `residue_algebraic` and `kaehler_finite` are the conclusions (a)–(c) of the lemma; the clause `R ⊄ V` of (a) is `lemma_5_10_a_not_R_le`. -/ structure Lemma510Place (k : Type u) {F : Type u} [Field k] [Field F] [Algebra k F] {n : ℕ} (x : Fin n → F) (i : Fin n) (j : Fin (n + 1)) where A : Subalgebra k F [finiteType : Algebra.FiniteType k A] [fractionRing : IsFractionRing A F] p : Ideal A [isPrime : p.IsPrime] place : ∀ z : F, z ∈ (placeSubalgebra A p).toSubring ∨ z⁻¹ ∈ (placeSubalgebra A p).toSubring isDVR : IsDiscreteValuationRing (placeValuationSubring A p place).toSubring q_mem : ∀ a, homogeneousCoordinates x a / homogeneousCoordinates x j ∈ A x_mem : x i ∈ A x_mem_p : (⟨x i, x_mem⟩ : A) ∈ p residue_algebraic : letI : Algebra k (placeValuationSubring A p place).toSubring := ((algebraMap k F).codRestrict (placeValuationSubring A p place).toSubring (fun c ↦ ofField_mem_algebraMap_v A p place (algebraMap k A c))).toAlgebra Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a ↦ residue (placeValuationSubring A p place).toSubring ⟨homogeneousCoordinates x a / homogeneousCoordinates x j, ofField_mem_algebraMap_v A p place ⟨_, q_mem a⟩⟩) : IntermediateField k (ResidueField (placeValuationSubring A p place).toSubring)) (ResidueField (placeValuationSubring A p place).toSubring) kaehler_finite : letI : Algebra k (placeValuationSubring A p place).toSubring := ((algebraMap k F).codRestrict (placeValuationSubring A p place).toSubring (fun c ↦ ofField_mem_algebraMap_v A p place (algebraMap k A c))).toAlgebra Module.Finite (placeValuationSubring A p place).toSubring (Ω[(placeValuationSubring A p place).toSubring⁄k])
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.LaneCSource ↗open Stafford38.Geometry.NormalizationHeightOneSource ↗open scoped nonZeroDivisorsSource ↗universe uSource ↗
local instance valuationSubringToSubringIsLocal {F : Type*} [Field F] (V : ValuationSubring F) : IsLocalRing V.toSubring :=
/-- The place at infinity of the paper's proof of Theorem 5.7: the valuation ring and chart of Lemma 5.9, the discrete valuation ring `V = A_{𝔭_D}` of Lemma 5.10, and a coefficient field `E` over which `κ` is finite (the input to the coefficient field of the completion, Matsumura Thm. 28.3). -/ theorem divisorialVisibleFrameExistenceWithChart (k K : Type u) [Field k] [CharZero k] [Field K] [Algebra k K] (r : ℕ) (y : Fin r → K) (i : Fin r) (hgen : IntermediateField.adjoin k (Set.range y) = ⊤) (hx : Transcendental k (y i)) : ∃ (E : IntermediateField k K) (V : ValuationSubring K) (hEV : ∀ z : E, (z : K) ∈ V.toSubring) (hVdvr : IsDiscreteValuationRing V.toSubring) (hxV : y i ∈ V.toSubring), letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra letI : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra Transcendental E (y i) ∧ (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ∧ Module.Finite E (ResidueField V.toSubring) ∧ Module.Finite V.toSubring (Ω[V.toSubring⁄k]) ∧ (∀ (s : K) (q : Fin (r + 1) → V.toSubring), (∀ a, (q a : K) = s * Fin.cases 1 y a) → (∃ j, q j = 1) → ¬ IsUnit (q 0) → Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun j : Fin r ↦ residue V.toSubring (q (Fin.succ j))) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) ∧ -- The chart `j` of Lemma 5.9, with `q_a = x̄_a / x̄_j ∈ V` (Lemma 5.10(a)). ∃ j, (Fin.cases 1 y j : K) ≠ 0 ∧ ∀ a, (Fin.cases 1 y a : K) / Fin.cases 1 y j ∈ V.toSubring := by -- Lemma 5.9. obtain ⟨W, j, hkW, -, hxW, hj, hqW⟩ := Stafford38.Paper.Section5.lemma_5_9 y i hx -- Lemma 5.10. obtain ⟨L⟩ := Stafford38.Paper.Section5.lemma_5_10 y i hgen hx W j hkW hxW hj hqW let A := L.A let p := L.p let V : ValuationSubring K := L.V have hVdvr : IsDiscreteValuationRing V.toSubring := L.isDVR letI : IsLocalRing V.toSubring := hVdvr.toIsLocalRing have hAV : ∀ a : A, (a : K) ∈ V.toSubring := fun a => ofField_mem_algebraMap_v A p L.place a have hpV : ∀ a : A, (⟨a, hAV a⟩ : V.toSubring) ∈ maximalIdeal V.toSubring ↔ a ∈ p := fun a => ofField_mem_maximalIdeal_iff_v A p L.place a have hsurj : ∀ z : ResidueField V.toSubring, ∃ (a b : A), b ∉ p ∧ z = residue V.toSubring ⟨a, hAV a⟩ / residue V.toSubring ⟨b, hAV b⟩ := fun z => ofField_residue_surjective_v A p L.place z have hy0 : y i ≠ 0 := fun h => hx (h ▸ isAlgebraic_zero) have hxV : y i ∈ V.toSubring := hAV ⟨y i, L.x_mem⟩ have hxm : (⟨y i, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := (hpV ⟨y i, L.x_mem⟩).2 L.x_mem_p obtain ⟨E, hEV, hEfin⟩ := stage5_of_verified A p V hAV hpV hsurj letI : Algebra E V.toSubring := (coeffHom E V hEV).toAlgebra have htrans : Transcendental E (y i) := stage6_transcendental_of_mem_maximalIdeal E V hEV (y i) hy0 hxV hxm have hk : ∀ c : k, algebraMap k K c ∈ V.toSubring := fun c => hAV (algebraMap k A c) let cod : Algebra k V.toSubring := ((algebraMap k K).codRestrict _ hk).toAlgebra let ground : Algebra k V.toSubring := (groundHom E V hEV).toAlgebra have hcod_ground : cod = ground := by apply Algebra.algebra_ext intro c apply Subtype.ext rfl letI : Algebra k V.toSubring := ground letI : IsScalarTower k V.toSubring K := IsScalarTower.of_algebraMap_eq fun c => by change algebraMap k K c = algebraMap k K c rfl have hkaehler : Module.Finite V.toSubring (Ω[V.toSubring⁄k]) := by have h := L.kaehler_finite change (letI : Algebra k V.toSubring := cod; Module.Finite V.toSubring (Ω[V.toSubring⁄k])) at h rw [hcod_ground] at h exact h -- The normalized column `q_a = x̄_a / x̄_j` of Lemma 5.10. let s : K := (Fin.cases 1 y j : K)⁻¹ have hjs : (Fin.cases 1 y j : K) * s = 1 := mul_inv_cancel₀ hj let c : Fin (r + 1) → V.toSubring := fun a => ⟨(Fin.cases 1 y a : K) / Fin.cases 1 y j, hAV ⟨_, L.q_mem a⟩⟩ have hc : ∀ a, (c a : K) = (Fin.cases 1 y a : K) * s := fun a => div_eq_mul_inv _ _ have hcj : c j = 1 := Subtype.ext (div_self hj) have halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring) := by have h := L.residue_algebraic change (letI : Algebra k V.toSubring := cod; Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a => residue V.toSubring (c a)) : IntermediateField k (ResidueField V.toSubring)) (ResidueField V.toSubring)) at h rw [hcod_ground] at h exact h refine ⟨E, V, hEV, hVdvr, hxV, htrans, hxm, hEfin, hkaehler, ?_, j, hj, fun a ↦ (c a).2⟩ intro s' q hq hq1 hq0 by_cases hj0 : j = 0 · obtain ⟨j', hj'⟩ := hq1 exfalso have hsone : s = 1 := by simpa [hj0] using hjs have hprod : ∀ a, q a = q 0 * c a := by intro a apply Subtype.ext change (q a : K) = (q 0 : K) * (c a : K) rw [hq a, hq 0, hc a] simp [hsone] have hq0res : residue V.toSubring (q 0) = 0 := by by_contra hnz exact hq0 ((residue_ne_zero_iff_isUnit _).1 hnz) have hqjres : residue V.toSubring (q j') = 0 := by rw [hprod j', map_mul, hq0res, zero_mul] have hqjunit : IsUnit (q j') := by rw [hj']; exact isUnit_one exact ((residue_ne_zero_iff_isUnit _).2 hqjunit) hqjres · exact DivisorialVisibleFrameStage4.stage4'_isAlgebraic_of_normalized_column V y j hj0 s c hc hcj halg s' q hq hq1 hq0
/-- **Paragraph "`a = ord q_0`", `a > 0`.** If `a = 0`, then `q_0` is a unit of `V`, hence all `x̄_i = q_i/q_0` lie in `V`, so `R ⊆ V` since also `K ⊆ V`, contradicting Lemma 5.10(a) (`lemma_5_10_a_not_R_le`). So `q_0 ∉ V^×`. -/ theorem step_a_pos {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra ¬ IsUnit (w.column.q 0) := by let C := w.column let W := C.W let K := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) K := W.ambientAlgebra let V := W.place.valuation letI : IsDiscreteValuationRing V.toSubring := W.place.isDiscrete let x : Fin m → K := fun j ↦ componentCoordinate P j -- `1 = ḡ x̄_1` with `ḡ ∈ R = K[x̄_1, …, x̄_n]`. obtain ⟨g, hg⟩ := hunit let phi : MvPolynomial (Fin m) k →+* K := (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) K).comp (Ideal.Quotient.mk P.asIdeal) have hphiC : phi.comp MvPolynomial.C = algebraMap k K := by ext c exact IsScalarTower.algebraMap_apply k (MvPolynomial (Fin m) k ⧸ P.asIdeal) K c have hpoly : phi g = MvPolynomial.eval₂ (algebraMap k K) x g := by rw [MvPolynomial.map_mvPolynomial_eq_eval₂ phi g] change MvPolynomial.eval₂Hom (phi.comp MvPolynomial.C) (fun j ↦ phi (MvPolynomial.X j)) g = MvPolynomial.eval₂Hom (algebraMap k K) x g exact MvPolynomial.eval₂Hom_congr hphiC rfl rfl have hinverse : x ⟨0, hm⟩ * MvPolynomial.eval₂ (algebraMap k K) x g = 1 := by have hzero : phi (MvPolynomial.X ⟨0, hm⟩ * g - 1) = 0 := by have hmk : Ideal.Quotient.mk P.asIdeal (MvPolynomial.X ⟨0, hm⟩ * g - 1) = 0 := Ideal.Quotient.eq_zero_iff_mem.mpr hg simpa [phi] using congrArg (algebraMap (MvPolynomial (Fin m) k ⧸ P.asIdeal) K) hmk rw [map_sub, map_mul, map_one, sub_eq_zero, hpoly] at hzero exact hzero have hgR : MvPolynomial.eval₂ (algebraMap k K) x g ∈ Algebra.adjoin k (Set.range x) := by rw [← MvPolynomial.aeval_def, Algebra.adjoin_range_eq_range_aeval] exact ⟨g, rfl⟩ -- `x̄_1 = ϖ ∈ 𝔪_V`. have hxV : x ⟨0, hm⟩ ∈ V.toSubring := by change componentCoordinate P ⟨0, hm⟩ ∈ V.toSubring rw [← W.parameter_eq_coordinate] exact W.place.parameter.2 have hxm : (⟨x ⟨0, hm⟩, hxV⟩ : V.toSubring) ∈ maximalIdeal V.toSubring := by have : (⟨x ⟨0, hm⟩, hxV⟩ : V.toSubring) = W.place.parameter := Subtype.ext W.parameter_eq_coordinate.symm rw [this, mem_maximalIdeal, mem_nonunits_iff] exact W.place.parameter_nonunit have hR := lemma_5_10_a_not_R_le x ⟨0, hm⟩ V hxV hxm ⟨_, hgR, hinverse⟩ rintro ⟨u, hu⟩ apply hR -- `q_0 ∈ V^×` puts every `x̄_i = q_i/q_0` in `V`. have hq0 : (C.q 0 : K) = C.scale := by simpa [componentProjectivePoint] using C.q_commonScale 0 have hxjV : ∀ j, x j ∈ V.toSubring := by intro j have hqj : (C.q (Fin.succ j) : K) = C.scale * x j := by simpa [componentProjectivePoint] using C.q_commonScale (Fin.succ j) have hscale : ((u : V.toSubring) : K) = C.scale := by rw [hu]; exact hq0 have hu1 : ((u : V.toSubring) : K) * ((↑(u⁻¹) : V.toSubring) : K) = 1 := by rw [← Subring.coe_mul, Units.mul_inv] rfl have : x j = (C.q (Fin.succ j) : K) * ((↑(u⁻¹) : V.toSubring) : K) := by rw [hqj, ← hscale, mul_comm ((u : V.toSubring) : K), mul_assoc, hu1, mul_one] rw [this] exact V.toSubring.mul_mem (C.q _).2 (↑(u⁻¹) : V.toSubring).2 intro z hz induction hz using Algebra.adjoin_induction with | mem z hz => obtain ⟨j, rfl⟩ := hz exact hxjV j | algebraMap c => rw [← C.groundCoeff_commutes] exact (C.groundCoeff c).2 | add a b _ _ ha hb => exact V.toSubring.add_mem ha hb | mul a b _ _ ha hb => exact V.toSubring.mul_mem ha hb
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Definition 5.3.** Let `F` be a field. (a) A subring `W ⊆ F` is a valuation ring of `F` iff `a ∈ W` or `a⁻¹ ∈ W` for every `a ∈ F^×`; it is a local ring with maximal ideal `𝔪_W = W ∖ W^×`. (b) A valuation ring `V` is a discrete valuation ring iff it is Noetherian and `V ≠ F`; then `𝔪_V` is generated by some `ϖ ∈ V`, and every `f ∈ F^×` is `f = ϖ^k u` with `u ∈ V^×` and a unique `k = ord_V(f) ∈ ℤ`. (c) For a field `κ`, `κ((t))` (`LaurentSeries κ`) is the fraction field of `κ[[t]]` (`PowerSeries κ`). -/ theorem definition_5_3 {F : Type u} [Field F] : (∀ S : Subring F, (∀ a : F, a ≠ 0 → a ∈ S ∨ a⁻¹ ∈ S) ↔ ∃ W : ValuationSubring F, W.toSubring = S) ∧ (∀ W : ValuationSubring F, IsLocalRing W.toSubring ∧ ∀ w : W.toSubring, w ∈ maximalIdeal W.toSubring ↔ ¬ IsUnit w) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring ↔ IsNoetherianRing V.toSubring ∧ V ≠ ⊤) ∧ (∀ V : ValuationSubring F, IsDiscreteValuationRing V.toSubring → ∃ ϖ : V.toSubring, maximalIdeal V.toSubring = Ideal.span {ϖ} ∧ ∀ f : F, f ≠ 0 → ∃! k : ℤ, ∃ u : V.toSubringˣ, f = (ϖ : F) ^ k * ((u : V.toSubring) : F)) ∧ ∀ (κ : Type u) [Field κ], IsFractionRing (PowerSeries κ) (LaurentSeries κ) := by refine ⟨fun S => ⟨fun h => ?_, ?_⟩, fun W => ⟨inferInstance, fun w => ?_⟩, fun V => ⟨fun hV => ?_, fun ⟨hN, hV⟩ => ?_⟩, fun V hV => ?_, fun κ _ => inferInstance⟩ · refine ⟨⟨S, fun a => ?_⟩, rfl⟩ by_cases ha : a = 0 · exact Or.inl (ha ▸ S.zero_mem) · exact h a ha · rintro ⟨W, rfl⟩ a _ exact W.mem_or_inv_mem a · rw [mem_maximalIdeal, mem_nonunits_iff] · refine ⟨inferInstance, fun hVtop => IsDiscreteValuationRing.not_isField V.toSubring ?_⟩ exact (step_5_3_isField_iff V).2 hVtop · have hfield : ¬ IsField V.toSubring := fun h => hV ((step_5_3_isField_iff V).1 h) have := step_5_3_valuationRing V exact ((IsDiscreteValuationRing.TFAE V.toSubring hfield).out 1 2).2 this · obtain ⟨ϖ, hϖ⟩ := IsDiscreteValuationRing.exists_irreducible V.toSubring refine ⟨ϖ, hϖ.maximalIdeal_eq, fun f hf => ?_⟩ obtain ⟨k, u, hku⟩ := step_5_3_ord_exists V hϖ hf exact ⟨k, ⟨u, hku⟩, fun l ⟨u', hl⟩ => step_5_3_ord_unique V hϖ (hl.symm.trans hku)⟩
Comparison notes
lemma_5_10 takes W and j from Lemma 5.9 and gives a discrete valuation ring V of F containing K with (a) q_i ∈ V and x̄₁ ∈ 𝔪_V, (b) κ algebraic over K(res q₀, …, res q_n) and (c) Ω_{V/K} finitely generated, as the fields of Lemma510Place, which also records V = A_{𝔭_D}. Frac V = F holds for every valuation subring of F; the proof also derives it from A ⊆ V as printed. The clause R ⊄ V of (a) is lemma_5_10_a_not_R_le, for every valuation ring with x̄₁ ∈ 𝔪_V; its hypothesis 1 = ḡx̄₁ with ḡ ∈ R is what the paper derives from Y ∩ H = ∅ by the Nullstellensatz. The proof is the printed one: the normalization A of C₀ = K[q₀, …, q_n, x̄₁], a prime 𝔭_D minimal over x̄₁, of height one by Krull’s principal ideal theorem, and V = A_{𝔭_D}, a Noetherian local normal domain of dimension one and hence a discrete valuation ring [Matsumura, Thm. 11.2]; (b) from the integrality of A/𝔭_D over the image of C₀ and (c) by localizing Ω_{A/K}. Definition 5.3 (definition_5_3) gives 𝔪_W = W ∖ W^×. The proof of Theorem 5.7 calls lemma_5_10 through LaneC.divisorialVisibleFrameExistenceWithChart and lemma_5_10_a_not_R_le in step_a_pos.
Review notes
Lemma 5.11 — Residue field and differential basis
Since and , the field is generated over by the finitely many with , so some subset of them is a transcendence basis of [Lang 2012, Thm. VIII.1.1]. As and are algebraic (Lemma 5.10(b)), so is ; hence is also a transcendence basis of . Since has characteristic zero, a transcendence basis of is a differential basis, i.e. its differentials form a -basis of [Matsumura 1986, Thm. 26.5]. ∎
/-- **Lemma 5.11.** -/ theorem lemma_5_11 {k κ : Type u} [Field k] [CharZero k] [Field κ] [Algebra k κ] {m : ℕ} (hm : 0 < m) (r : Fin (m + 1) → κ) (j : Fin (m + 1)) (hr0 : r 0 = 0) (hr1 : r (Fin.succ ⟨0, hm⟩) = 0) (hrj : r j = 1) (halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range r) : IntermediateField k κ) κ) : ∃ (d' : ℕ) (rows : Fin d' → Fin (m + 1)), Function.Injective rows ∧ (∀ l, rows l ≠ 0 ∧ rows l ≠ Fin.succ ⟨0, hm⟩ ∧ rows l ≠ j) ∧ LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) ∧ Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = ⊤ ∧ d' = Module.finrank κ (Ω[κ⁄k]) := by classical let axis : Fin (m + 1) := Fin.succ ⟨0, hm⟩ let good : Set (Fin (m + 1)) := {a | a ≠ 0 ∧ a ≠ axis ∧ a ≠ j} let S : Set κ := r '' good -- The excluded residues are `0` and `1`, so `L` is generated by `S`. have hL : IntermediateField.adjoin k (Set.range r) = IntermediateField.adjoin k S := by apply le_antisymm · rw [IntermediateField.adjoin_le_iff] rintro _ ⟨a, rfl⟩ by_cases ha : a ∈ good · exact IntermediateField.subset_adjoin k S ⟨a, ha, rfl⟩ · simp only [good, Set.mem_ofPred_eq, not_and_or, not_not] at ha rcases ha with rfl | rfl | rfl · rw [hr0]; exact zero_mem _ · rw [hr1]; exact zero_mem _ · rw [hrj]; exact one_mem _ · apply IntermediateField.adjoin.mono rintro _ ⟨a, -, rfl⟩ exact ⟨a, rfl⟩ -- `L` is generated by the finite set `S`, which therefore contains a transcendence -- basis `B` of `L/K` (Definition 5.1, closing sentence; Lang, Thm. VIII.1.1). let SL : Set (IntermediateField.adjoin k S) := Subtype.val ⁻¹' S have hSL : IntermediateField.adjoin k SL = ⊤ := by apply IntermediateField.lift_injective (IntermediateField.adjoin k S) rw [IntermediateField.lift_adjoin, IntermediateField.lift_top] congr 1 change Subtype.val '' (Subtype.val ⁻¹' S) = S ext z refine ⟨fun ⟨w, hw, hwz⟩ => hwz ▸ hw, fun hz => ?_⟩ exact ⟨⟨z, IntermediateField.subset_adjoin k S hz⟩, hz, rfl⟩ have hSLfin : SL.Finite := ((Set.toFinite good).image r).preimage Subtype.val_injective.injOn obtain ⟨BL, hBLS, hBL⟩ := (definition_5_1_closing (K := k) (L := IntermediateField.adjoin k S)).2 SL hSLfin hSL -- `κ/L` is algebraic (Lemma 5.10(b)), so `B` is a transcendence basis of `κ/K`. have : Algebra.IsAlgebraic (IntermediateField.adjoin k S) κ := hL ▸ halg let t : Set κ := Set.range (algebraMap (IntermediateField.adjoin k S) κ ∘ ((↑) : BL → _)) have ht : IsTranscendenceBasis k ((↑) : t → κ) := hBL.algebraMap_comp.to_subtype_range have htS : t ⊆ S := by rintro _ ⟨b, rfl⟩ exact hBLS b.2 have htfin : t.Finite := (Set.toFinite good |>.image r).subset htS let : Fintype t := htfin.fintype -- Matsumura, Thm. 26.5: the transcendence basis is a differential basis. have hrank : Module.finrank κ (Ω[κ⁄k]) = Fintype.card t := Stafford38.Geometry.kaehlerFinrankOfTranscendenceBasis k κ t (↑) ht have hspanT : Submodule.span κ (Set.range fun x : t ↦ KaehlerDifferential.D k κ (x : κ)) = ⊤ := Stafford38.Geometry.KaehlerDVRVisibility.kaehler_span_of_isAlgebraic_adjoin (fun x : t ↦ (x : κ)) (by simpa using (definition_5_1_g_h ht).1) -- Index the basis elements by residues `q_i(0)` with `i ∉ {0, 1, j}`. have hpre : ∀ x : t, ∃ a ∈ good, r a = x := fun x ↦ htS x.2 choose idx hidx hridx using hpre let e := Fintype.equivFin t let rows : Fin (Fintype.card t) → Fin (m + 1) := fun l ↦ idx (e.symm l) have hrows : ∀ l, r (rows l) = (e.symm l : κ) := fun l ↦ hridx _ have hfam : (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = (fun x : t ↦ KaehlerDifferential.D k κ (x : κ)) ∘ e.symm := by funext l simp [hrows] have hspan : Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (r (rows l))) = ⊤ := by rw [hfam, Set.range_comp, e.symm.surjective.range_eq, Set.image_univ, hspanT] have hli : LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (r (rows l))) := linearIndependent_of_top_le_span_of_card_eq_finrank hspan.ge (by rw [Fintype.card_fin, hrank]) refine ⟨Fintype.card t, rows, ?_, fun l ↦ hidx _, hli, hspan, hrank.symm⟩ intro l l' h apply e.symm.injective apply Subtype.ext rw [← hrows, ← hrows, h]
/-- **Definition 5.1(g)–(h).** For a transcendence basis `B` of `L/K`, `L` is algebraic over `K(B)`, and `|B| = trdeg_K L`. -/ theorem definition_5_1_g_h {B : Set L} (hB : IsTranscendenceBasis K ((↑) : B → L)) : Algebra.IsAlgebraic (IntermediateField.adjoin K (Set.range ((↑) : B → L))) L ∧ Cardinal.mk B = Algebra.trdeg K L := ⟨hB.isAlgebraic_field, hB.cardinalMk_eq_trdeg⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
/-- **Definition 5.1, closing sentence.** The elements of `L` that are algebraic over `K` form a subfield, and every finite set generating `L` over `K` contains a transcendence basis ([Lang, Thm. VIII.1.1]). -/ theorem definition_5_1_closing : (∃ E : IntermediateField K L, ∀ z : L, z ∈ E ↔ IsAlgebraic K z) ∧ ∀ S : Set L, S.Finite → IntermediateField.adjoin K S = ⊤ → ∃ B ⊆ S, IsTranscendenceBasis K ((↑) : B → L) := by refine ⟨⟨algebraicClosure K L, fun z => mem_algebraicClosure_iff⟩, fun S _ hS => ?_⟩ have : Algebra.IsAlgebraic (Algebra.adjoin K S) L := by rw [← IntermediateField.isAlgebraic_adjoin_iff_top, hS] exact ⟨fun z => ⟨Polynomial.X - Polynomial.C ⟨z, trivial⟩, Polynomial.X_sub_C_ne_zero _, by simp [Polynomial.aeval_def]⟩⟩ exact exists_isTranscendenceBasis_subset S
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗variable {k : Type*} [Field k] {n : ℕ}Source ↗
variable {K L : Type*} [Field K] [Field L] [Algebra K L]
More Lean proofs and definitions
/-- **Lemma 5.11 for the valuation ring of the formal proof.** The residues `res q_i` of its normalized column satisfy `res q_0 = res q_1 = 0`, `res q_j = 1`, and some `res q_i` with `i ∉ {0, 1, j}` have differentials forming a basis of `Ω_{κ/K}`. -/ theorem witness_lemma_5_11 {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hunit : ∃ g : MvPolynomial (Fin m) k, MvPolynomial.X ⟨0, hm⟩ * g - 1 ∈ P.asIdeal) : let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W residue V (C.q 0) = 0 ∧ residue V (C.q (Fin.succ ⟨0, hm⟩)) = 0 ∧ residue V (C.q C.chart) = 1 ∧ ∃ (d' : ℕ) (rows : Fin d' → Fin (m + 1)), Function.Injective rows ∧ (∀ l, rows l ≠ 0 ∧ rows l ≠ Fin.succ ⟨0, hm⟩ ∧ rows l ≠ C.chart) ∧ LinearIndependent κ (fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l)))) ∧ Submodule.span κ (Set.range fun l ↦ KaehlerDifferential.D k κ (residue V (C.q (rows l)))) = ⊤ ∧ d' = Module.finrank κ (Ω[κ⁄k]) := by classical let C := w.column let W := C.W letI : Algebra (CoordinateZeroLocalRing W.coefficientField) (ComponentFractionField P) := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W letI : IsLocalRing V := W.place.isDiscrete.toIsLocalRing obtain ⟨a, b, u₀, u₁, ha, hab, hq0, -, hq1⟩ := order_gap_of_actual_witness hm P w hunit have hres := constantCoeff_retainedToCompleted_eq_residue (k := k) W have hr0 : residue V (C.q 0) = 0 := by rw [← hres, hq0] simp [ha.ne'] have hr1 : residue V (C.q (Fin.succ ⟨0, hm⟩)) = 0 := by rw [← hres, hq1] simp [(ha.trans hab).ne'] have hrj : residue V (C.q C.chart) = 1 := by rw [C.chart_one, map_one] have halg : Algebra.IsAlgebraic (IntermediateField.adjoin k (Set.range fun a ↦ residue V (C.q a)) : IntermediateField k κ) κ := by have h := w.halg have hle : IntermediateField.adjoin k (Set.range fun j : Fin m ↦ residue V (C.q (Fin.succ j))) ≤ IntermediateField.adjoin k (Set.range fun a ↦ residue V (C.q a)) := IntermediateField.adjoin.mono _ _ _ (by rintro _ ⟨j, rfl⟩; exact ⟨_, rfl⟩) exact ⟨fun z ↦ IsAlgebraic.tower_top_of_subalgebra_le (IntermediateField.toSubalgebra_le_toSubalgebra.mpr hle) (h.isAlgebraic z)⟩ exact ⟨hr0, hr1, hrj, lemma_5_11 hm (fun a ↦ residue V (C.q a)) C.chart hr0 hr1 hrj halg⟩
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
Comparison notes
lemma_5_11 takes residues r₀, …, r_n ∈ κ with r₀ = r₁ = 0, r_j = 1 and κ algebraic over K(r₀, …, r_n), and gives distinct indices outside {0, 1, j} whose differentials form a basis of Ω_{κ/K}; their number is dim_κ Ω_{κ/K}. The proof is the printed one: the residues with indices outside {0, 1, j} generate L = K(r₀, …, r_n), so they contain a transcendence basis B of L/K (closing sentence of Definition 5.1, definition_5_1_closing; Lang Thm. VIII.1.1); B is a transcendence basis of κ/K because κ/L is algebraic, and in characteristic zero it is a differential basis (Definition 5.1(g), Matsumura Thm. 26.5). witness_lemma_5_11 applies it to the ring V of the proof of Theorem 5.7, which uses the basis in step_dual_derivations.
Review notes
Lemma 5.12 — Tangent space at the Laurent generic point
- ; hence .
- For every -derivation of the column lies in ; moreover .
The second fundamental exact sequence for is
(b) Let and . By (5.1), , and vanishes on , so
∎
/-- **Lemma 5.12(a), `dim Θ_y = dim Ω_{F/k}`.** By the exact sequence and its localization (`fieldKaehlerMap_exact`), `span_F {df(x̄) : f ∈ 𝔭}` has dimension `n − dim_F Ω_{F/K}`. Rank does not change under the field extension `F ⊆ E` (in the paper `E = κ((t))`), so `span_E {df_y : f ∈ 𝔭}` has the same dimension, and `Θ_y = ⋂_f ker df_y` has dimension `n − (n − dim_F Ω_{F/K})`. -/ theorem generic_tangent_finrank_eq_component_kaehler_finrank {k F E : Type u} [Field k] [Field F] [Field E] [Algebra k F] [Algebra k E] [Algebra F E] [IsScalarTower k F E] {m : ℕ} {I : Ideal (MvPolynomial (Fin m) k)} [I.IsPrime] [Algebra (MvPolynomial (Fin m) k ⧸ I) F] [IsScalarTower k (MvPolynomial (Fin m) k ⧸ I) F] [IsFractionRing (MvPolynomial (Fin m) k ⧸ I) F] : Module.finrank E (zariskiTangentSpace (fun i ↦ algebraMap F E (GenericPointKaehlerConormal.genericPoint I F i)) (I.map (MvPolynomial.map (algebraMap k E)))) = Module.finrank F (Ω[F⁄k]) := by classical let yF := GenericPointKaehlerConormal.genericPoint I F let yE : Fin m → E := fun i ↦ algebraMap F E (yF i) let IE : Ideal (MvPolynomial (Fin m) E) := I.map (MvPolynomial.map (algebraMap k E)) obtain ⟨hrange, hker⟩ := fieldKaehlerMap_exact (F := F) I -- `dim_F span_F {df(x̄)} = n − dim_F Ω_{F/K}`. let T : Submodule F (Fin m → F) := Submodule.span F (Set.range fun f : I ↦ genericGradient (F := F) I f.1) have hTdim : Module.finrank F T + Module.finrank F (Ω[F⁄k]) = m := by have h := LinearMap.finrank_range_add_finrank_ker (fieldKaehlerMap (F := F) I) rw [hrange, hker, finrank_top, Module.finrank_fin_fun] at h rw [add_comm] exact h -- Rank does not change under `F ⊆ E`. let gE : I → Fin m → E := fun f i ↦ algebraMap F E (genericGradient (F := F) I f.1 i) let TE : Submodule E (Fin m → E) := Submodule.span E (Set.range gE) have hTE : Module.finrank E TE = Module.finrank F T := by let b := Module.finBasis F T let z : Fin (Module.finrank F T) → Fin m → F := fun j ↦ (b j : Fin m → F) have hliF : LinearIndependent F z := b.linearIndependent.map' T.subtype (Submodule.ker_subtype T) have hli : LinearIndependent E (fun j i ↦ algebraMap F E (z j i)) := (linearIndependent_algebraMap_comp_iff (R := F) (S := E) (ι' := Fin m)).2 hliF have hspanz : Submodule.span F (Set.range z) = T := by have h := congrArg (Submodule.map T.subtype) b.span_eq rw [Submodule.map_span, ← Set.range_comp, Submodule.map_top, Submodule.range_subtype] at h exact h have hTEeq : TE = Submodule.span E (Set.range fun j i ↦ algebraMap F E (z j i)) := by apply le_antisymm · rw [Submodule.span_le] rintro _ ⟨f, rfl⟩ have hf : genericGradient (F := F) I f.1 ∈ Submodule.span F (Set.range z) := by rw [hspanz] exact Submodule.subset_span ⟨f, rfl⟩ exact algebraMap_comp_mem_span (B := E) z hf · rw [Submodule.span_le] rintro _ ⟨j, rfl⟩ exact algebraMap_comp_mem_span (B := E) (fun f : I ↦ genericGradient (F := F) I f.1) (b j).2 rw [hTEeq, finrank_span_eq_card hli, Fintype.card_fin] -- `df_y` for `f ∈ 𝔭` is the image of `df(x̄)`, and `f(y) = 0`. have hevalF : ∀ p : MvPolynomial (Fin m) k, MvPolynomial.eval₂ (algebraMap k F) yF p = algebraMap (MvPolynomial (Fin m) k ⧸ I) F (Ideal.Quotient.mk I p) := by have hhom : MvPolynomial.aeval yF = (IsScalarTower.toAlgHom k (MvPolynomial (Fin m) k ⧸ I) F).comp (Ideal.Quotient.mkₐ k I) := by apply MvPolynomial.algHom_ext intro i simp [yF, GenericPointKaehlerConormal.genericPoint] intro p exact congrArg (fun φ ↦ φ p) hhom have hevalE : ∀ p : MvPolynomial (Fin m) k, MvPolynomial.eval₂ (algebraMap k E) yE p = algebraMap F E (MvPolynomial.eval₂ (algebraMap k F) yF p) := by intro p rw [MvPolynomial.eval₂_comp_left, ← IsScalarTower.algebraMap_eq k F E] rfl have hdf : ∀ f : MvPolynomial (Fin m) k, ∀ i, Stafford38.Geometry.CoisotropicTranslation.differentialAt yE (MvPolynomial.map (algebraMap k E) f) i = algebraMap F E (genericGradient (F := F) I f i) := by intro f i simp only [Stafford38.Geometry.CoisotropicTranslation.differentialAt, MvPolynomial.pderiv_map, MvPolynomial.eval_map] rw [hevalE, hevalF] rfl have hzero : ∀ g ∈ I, MvPolynomial.eval₂ (algebraMap k E) yE g = 0 := by intro g hg rw [hevalE, hevalF, (Ideal.Quotient.eq_zero_iff_mem).2 hg, map_zero, map_zero] -- `Θ_y = ⋂_{f ∈ 𝔭} ker df_y`. let C := TE.map (coordinateCovectorLinear (k := E) (n := m)) have hC : Module.finrank E C = Module.finrank E TE := (Submodule.equivMapOfInjective _ coordinateCovector_injective TE).finrank_eq.symm have hΘ : zariskiTangentSpace yE IE = C.dualCoannihilator := by apply le_antisymm · intro v hv rw [Submodule.mem_dualCoannihilator] rintro _ ⟨t, ht, rfl⟩ induction ht using Submodule.span_induction with | mem t ht => obtain ⟨f, rfl⟩ := ht have hfE : MvPolynomial.map (algebraMap k E) f.1 ∈ IE := Ideal.mem_map_of_mem _ f.2 have h := (Submodule.mem_dualCoannihilator _).1 hv _ (Submodule.subset_span ⟨⟨_, hfE⟩, rfl⟩) rw [differentialCovector_apply] at h change ∑ i, gE f i * v i = 0 simpa [gE, hdf] using h | zero => simp | add t t' _ _ ht ht' => simp [ht, ht'] | smul a t _ ht => simp [ht] · intro v hv rw [Submodule.mem_dualCoannihilator] at hv apply mem_zariskiTangentSpace_map_of_forall_mem I yE v hzero intro g hg have h := hv _ ⟨gE ⟨g, hg⟩, Submodule.subset_span ⟨⟨g, hg⟩, rfl⟩, rfl⟩ change ∑ i, gE ⟨g, hg⟩ i * v i = 0 at h have h' : ∀ i, MvPolynomial.eval₂ (algebraMap k E) yE (MvPolynomial.pderiv i g) = gE ⟨g, hg⟩ i := by intro i have := hdf g i simp only [Stafford38.Geometry.CoisotropicTranslation.differentialAt, MvPolynomial.pderiv_map, MvPolynomial.eval_map] at this exact this simp_rw [h'] exact h have hdual := Subspace.finrank_add_finrank_dualCoannihilator_eq C rw [← hΘ, hC, hTE, Module.finrank_fin_fun] at hdual change Module.finrank E (zariskiTangentSpace yE IE) = _ omega
/-- **`dim Ω_{F/k} = trdeg_k F`.** For a field `F` generated over `k` by finitely many elements, in characteristic zero, the Kähler rank is the transcendence degree (Matsumura, Thm. 26.5), which is finite. -/ theorem component_kaehler_finrank_eq_trdeg {k F : Type u} [Field k] [CharZero k] [Field F] [Algebra k F] {m : ℕ} (y : Fin m → F) (hgen : IntermediateField.adjoin k (Set.range y) = ⊤) : (Module.finrank F (Ω[F⁄k]) : Cardinal) = Algebra.trdeg k F ∧ Module.finrank F (Ω[F⁄k]) = (Algebra.trdeg k F).toNat := by classical have : Algebra.IsAlgebraic (Algebra.adjoin k (Set.range y)) F := by rw [← IntermediateField.isAlgebraic_adjoin_iff_top, hgen, Algebra.isAlgebraic_iff_isIntegral] exact Algebra.isIntegral_of_surjective IntermediateField.topEquiv.surjective obtain ⟨t, hts, ht⟩ := exists_isTranscendenceBasis_subset (R := k) (Set.range y) have htfin : t.Finite := (Set.finite_range y).subset hts let : Fintype t := htfin.fintype have hrank : Module.finrank F (Ω[F⁄k]) = Fintype.card t := kaehlerFinrankOfTranscendenceBasis k F t (↑) ht have hcard : (Fintype.card t : Cardinal) = Algebra.trdeg k F := by rw [← Cardinal.mk_fintype] exact (definition_5_1_g_h ht).2 refine ⟨by rw [hrank, hcard], ?_⟩ rw [← hcard, Cardinal.toNat_natCast, hrank]
More Lean proofs and definitions
/-- **Second fundamental exact sequence for `K[x] → R = K[x]/𝔭`.** In `R ⊗ Ω_{K[x]/K} = R^n` (basis `1 ⊗ dx_i`) the image of `f ∈ 𝔭` is `df(x̄)`, and the exact sequence `𝔭/𝔭² → R^n → Ω_{R/K} → 0` (Matsumura, Thm. 25.2) gives `Ω_{R/K} = R^n / span_R {df(x̄) : f ∈ 𝔭}`: the map `ξ ↦ Σ ξ_i dx̄_i` is onto, with kernel `span_R {df(x̄)}`. -/ theorem quotientKaehlerMap_exact : LinearMap.range (quotientKaehlerMap I) = ⊤ ∧ LinearMap.ker (quotientKaehlerMap I) = Submodule.span R (Set.range fun f : I ↦ quotientGradient I f.1) := by classical have h1 : ∀ ω : Ω[R⁄k], (1 : R) • ω = ω := fun ω ↦ one_smul R ω have hδ : ∀ ξ : Fin m → R, quotientKaehlerMap I ξ = ∑ i, ξ i • D k R (Ideal.Quotient.mk I (MvPolynomial.X i)) := fun ξ ↦ Fintype.linearCombination_apply _ _ _ let tb := Algebra.TensorProduct.basis R (mvPolynomialBasis k (Fin m)) have htb : ∀ i, mapBaseChange k A R (tb i) = D k R (Ideal.Quotient.mk I (MvPolynomial.X i)) := by intro i simp only [tb, Algebra.TensorProduct.basis_apply, mapBaseChange_tmul, mvPolynomialBasis_apply, map_D, h1] rfl -- The coordinates of `1 ⊗ df` in the basis `1 ⊗ dx_i` are `df(x̄)`. have hcoord : ∀ f : A, (1 : R) ⊗ₜ[A] D k A f = ∑ i, quotientGradient I f i • tb i := by intro f apply tb.repr.injective ext i rw [Algebra.TensorProduct.basis_repr_tmul, map_sum] simp [tb, quotientGradient, Finsupp.single_apply] -- Hence the image of `t ∈ R ⊗ Ω_{K[x]/K}` in `Ω_{R/K}` is `δ` of its coordinates. have hmap : ∀ ξ : Fin m → R, mapBaseChange k A R (∑ i, ξ i • tb i) = quotientKaehlerMap I ξ := by intro ξ rw [hδ, map_sum] simp_rw [map_smul, htb] have hsurj : Function.Surjective (algebraMap A R) := Ideal.Quotient.mk_surjective refine ⟨?_, le_antisymm ?_ ?_⟩ · -- `Ω_{R/K}` is spanned by the `dx̄_i`. rw [eq_top_iff] intro ω _ obtain ⟨t, rfl⟩ := mapBaseChange_surjective k A R hsurj ω refine ⟨tb.equivFun t, ?_⟩ rw [← hmap, tb.sum_equivFun] · -- `ξ ∈ ker δ`: `t = Σ ξ_i ⊗ dx_i` lies in the kernel of `R ⊗ Ω_{K[x]/K} → Ω_{R/K}`, -- hence in the image of `𝔭/𝔭²`, so `t = 1 ⊗ df` for some `f ∈ 𝔭` and `ξ = df(x̄)`. intro ξ h have hker : (∑ i, ξ i • tb i) ∈ LinearMap.ker (mapBaseChange k A R) := by rw [LinearMap.mem_ker, hmap] exact h obtain ⟨c, hc⟩ := (exact_kerCotangentToTensor_mapBaseChange k A R hsurj _).mp hker obtain ⟨⟨f, hf⟩, rfl⟩ := Ideal.toCotangent_surjective _ c rw [kerCotangentToTensor_toCotangent] at hc have hfI : f ∈ I := by simpa [RingHom.mem_ker, Ideal.Quotient.eq_zero_iff_mem] using hf have hξ : ξ = quotientGradient I f := by have := congrArg tb.equivFun ((hcoord f).symm.trans hc) simp only [map_sum, map_smul] at this funext i have hi := congrFun this i simpa [Finsupp.single_apply] using hi.symm rw [hξ] exact Submodule.subset_span ⟨⟨f, hfI⟩, rfl⟩ · -- `δ(df(x̄)) = d(f(x̄)) = 0` for `f ∈ 𝔭`. rw [Submodule.span_le] rintro _ ⟨⟨f, hf⟩, rfl⟩ rw [SetLike.mem_coe, LinearMap.mem_ker, ← hmap, ← hcoord, mapBaseChange_tmul, h1, map_D] change D k R (Ideal.Quotient.mk I f) = 0 rw [(Ideal.Quotient.eq_zero_iff_mem).2 hf, map_zero]
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k : Type*} [Field k] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k))Source ↗
local notation "A" => MvPolynomial (Fin m) kSource ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- **Localization of the exact sequence.** `Ω_{S⁻¹R/K} = S⁻¹ Ω_{R/K}` for `S = R ∖ {0}` (Hartshorne, Prop. II.8.2A), and localization is exact, so `Ω_{F/K} = F^n / span_F {df(x̄) : f ∈ 𝔭}`. -/ theorem fieldKaehlerMap_exact : LinearMap.range (fieldKaehlerMap (F := F) I) = ⊤ ∧ LinearMap.ker (fieldKaehlerMap (F := F) I) = Submodule.span F (Set.range fun f : I ↦ genericGradient (F := F) I f.1) := by classical obtain ⟨hrangeR, hkerR⟩ := quotientKaehlerMap_exact I let φ := KaehlerDifferential.map k k R F have hδF : ∀ ξ : Fin m → F, fieldKaehlerMap I ξ = ∑ i, ξ i • D k F (GenericPointKaehlerConormal.genericPoint I F i) := fun ξ ↦ Fintype.linearCombination_apply _ _ _ have hδR : ∀ ξ : Fin m → R, quotientKaehlerMap I ξ = ∑ i, ξ i • D k R (Ideal.Quotient.mk I (MvPolynomial.X i)) := fun ξ ↦ Fintype.linearCombination_apply _ _ _ -- `Ω_{R/K} → Ω_{F/K}` sends `Σ ξ_i dx̄_i` to `Σ ξ_i dx̄_i`. have hφ : ∀ ξ : Fin m → R, φ (quotientKaehlerMap I ξ) = fieldKaehlerMap I (fun i ↦ algebraMap R F (ξ i)) := by intro ξ rw [hδR, hδF, map_sum] refine Finset.sum_congr rfl fun i _ ↦ ?_ rw [map_smul, map_D] exact (algebraMap_smul F (ξ i) (D k F (GenericPointKaehlerConormal.genericPoint I F i))).symm refine ⟨?_, le_antisymm ?_ ?_⟩ · -- `Ω_{F/K}` is spanned by the images of `Ω_{R/K} = δ_R(R^n)`. rw [eq_top_iff, ← span_range_map_derivation_of_isLocalization k R F (nonZeroDivisors R), Submodule.span_le] rintro _ ⟨r, rfl⟩ obtain ⟨ξ, hξ⟩ : D k R r ∈ LinearMap.range (quotientKaehlerMap I) := by rw [hrangeR]; exact Submodule.mem_top refine ⟨fun i ↦ algebraMap R F (ξ i), ?_⟩ rw [← hφ, hξ] rfl · -- Clear denominators, `ξ = b⁻¹ ξ'`; then `Σ ξ'_i dx̄_i` dies in `F ⊗ Ω_{R/K}`, so -- `c Σ ξ'_i dx̄_i = 0` in `Ω_{R/K}` for some `c ≠ 0`, and `c ξ' ∈ span_R {df(x̄)}`. intro ξ hξ rw [LinearMap.mem_ker] at hξ obtain ⟨b, hb⟩ := IsLocalization.exist_integer_multiples_of_finite (nonZeroDivisors R) ξ choose ξ' hξ' using hb have hzero : φ (quotientKaehlerMap I ξ') = 0 := by rw [hφ] have : (fun i ↦ algebraMap R F (ξ' i)) = (algebraMap R F b) • ξ := by funext i rw [hξ' i, Pi.smul_apply, Algebra.smul_def, smul_eq_mul] rw [this, map_smul, hξ, smul_zero] obtain ⟨c, hc⟩ := (IsLocalizedModule.eq_zero_iff (nonZeroDivisors R) φ).mp hzero have hmem : (c : R) • ξ' ∈ Submodule.span R (Set.range fun f : I ↦ quotientGradient I f.1) := by rw [← hkerR, LinearMap.mem_ker, map_smul] exact hc have hF := algebraMap_comp_mem_span (B := F) _ hmem have hb0 : algebraMap R F b ≠ 0 := IsFractionRing.to_map_ne_zero_of_mem_nonZeroDivisors b.2 have hc0 : algebraMap R F c ≠ 0 := IsFractionRing.to_map_ne_zero_of_mem_nonZeroDivisors c.2 have hξeq : ξ = (algebraMap R F c * algebraMap R F b)⁻¹ • (fun i ↦ algebraMap R F (((c : R) • ξ') i)) := by funext i simp only [Pi.smul_apply, smul_eq_mul, map_mul, hξ' i] rw [Algebra.smul_def] field_simp rw [hξeq] exact Submodule.smul_mem _ _ hF · -- `δ_F(df(x̄))` is the image of `δ_R(df(x̄)) = 0`. rw [Submodule.span_le] rintro _ ⟨⟨f, hf⟩, rfl⟩ have hR : quotientKaehlerMap I (quotientGradient I f) = 0 := by rw [← LinearMap.mem_ker, hkerR] exact Submodule.subset_span ⟨⟨f, hf⟩, rfl⟩ rw [SetLike.mem_coe, LinearMap.mem_ker] have := hφ (quotientGradient I f) rw [hR, map_zero] at this exact this.symm
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k F : Type*} [Field k] [Field F] [Algebra k F] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k)) [I.IsPrime] [Algebra (MvPolynomial (Fin m) k ⧸ I) F] [IsScalarTower k (MvPolynomial (Fin m) k ⧸ I) F] [IsFractionRing (MvPolynomial (Fin m) k ⧸ I) F]Source ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- **Lemma 5.12(a), `dim_F Ω_{F/K} ≤ d' + 1`.** For a discrete valuation ring `V ⊇ K` with fraction field `F`, residue field `κ` and `Ω_{V/K}` finitely generated: the second fundamental exact sequence for `V → κ` gives `dim_κ (κ ⊗_V Ω_{V/K}) ≤ d' + 1`, Nakayama's lemma gives that many generators of `Ω_{V/K}`, and `F ⊗_V Ω_{V/K} = Ω_{F/K}` bounds `dim_F Ω_{F/K}`. -/ theorem lemma_5_12_a_bound {F : Type u} [Field F] [Algebra V F] [IsFractionRing V F] [Algebra k F] [IsScalarTower k V F] [Module.Finite V (Ω[V⁄k])] : Module.finrank F (Ω[F⁄k]) ≤ Module.finrank (ResidueField V) (Ω[ResidueField V⁄k]) + 1 := by obtain ⟨f, hf⟩ := exists_kaehler_generators (k := k) (V := V) have h := finrank_fraction_kaehler_le_of_span (F := F) f hf rw [Fintype.card_fin] at h exact h.trans finrank_residue_tensor_kaehler_le
/-- **Lemma 5.12(a), cone part.** `dim 𝒞 = dim Θ_y + 1`. -/ theorem lemma_5_12_a_cone (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) : Module.finrank L (tangentCone q I) = Module.finrank L (tangentSpaceAt q I) + 1 := by classical let f : tangentCone q I →ₗ[L] tangentSpaceAt q I := (rho q).restrict (p := tangentCone q I) (q := tangentSpaceAt q I) (fun v hv ↦ hv) have hf : Function.Surjective f := by rintro ⟨w, hw⟩ obtain ⟨v, rfl⟩ := rho_surjective q hq0 w exact ⟨⟨v, hw⟩, rfl⟩ have hq : q ∈ tangentCone q I := by change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _ have hker : Module.finrank L (LinearMap.ker f) = 1 := by have hq' : (⟨q, hq⟩ : tangentCone q I) ≠ 0 := by intro h exact hq0 (congrFun (congrArg Subtype.val h) 0) rw [finrank_eq_one_iff_of_nonzero' (⟨⟨q, hq⟩, ?_⟩ : LinearMap.ker f) ?_] · rintro ⟨⟨v, hv⟩, hvk⟩ have hv0 : v ∈ LinearMap.ker (rho q) := by rw [LinearMap.mem_ker] exact congrArg Subtype.val (LinearMap.mem_ker.mp hvk) rw [ker_rho q hq0, Submodule.mem_span_singleton] at hv0 obtain ⟨c, rfl⟩ := hv0 exact ⟨c, rfl⟩ · rw [LinearMap.mem_ker] apply Subtype.ext change rho q q = 0 rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q · intro h exact hq' (congrArg Subtype.val h) have := LinearMap.finrank_range_add_finrank_ker f rw [LinearMap.range_eq_top.mpr hf, finrank_top, hker] at this omega
/-- **Lemma 5.12(b).** If the equations vanish at `y`, then `δq ∈ 𝒞` for every `K`-derivation `δ` of `L`, and `q ∈ 𝒞`. -/ theorem lemma_5_12_b (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) (hbase : ∀ f ∈ I, MvPolynomial.eval₂ (algebraMap k L) (dehomogenizedPoint q) f = 0) : (∀ δ : Derivation k L L, (fun a ↦ δ (q a)) ∈ tangentCone q I) ∧ q ∈ tangentCone q I := by constructor · intro δ -- `δ(f(y)) = df_y · δy`, so `δy ∈ Θ_y`. have htan := Stafford38.Geometry.PowerSeriesArcTangency.derivationVector_mem_zariskiTangentSpace_of_eval₂_eq_zero δ I (dehomogenizedPoint q) hbase -- Quotient rule: `ρ(δq) = δy`. have hquot : rho q (fun a ↦ δ (q a)) = fun i ↦ δ (dehomogenizedPoint q i) := by funext i rw [rho_apply, dehomogenizedPoint, Derivation.leibniz_div] simp only [smul_eq_mul] field_simp change rho q (fun a ↦ δ (q a)) ∈ tangentSpaceAt q I rw [hquot] exact htan · change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _
/-- The column `df(x̄) = (∂_i f(x̄))_i ∈ R^n` of an equation `f`. -/ noncomputable def quotientGradient (f : A) : Fin m → R := fun i ↦ Ideal.Quotient.mk I (MvPolynomial.pderiv i f)
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k : Type*} [Field k] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k))Source ↗
local notation "A" => MvPolynomial (Fin m) kSource ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- `ξ ↦ Σ ξ_i dx̄_i`, from `R^n` to `Ω_{R/K}`. -/ noncomputable def quotientKaehlerMap : (Fin m → R) →ₗ[R] Ω[R⁄k] := Fintype.linearCombination R fun i ↦ D k R (Ideal.Quotient.mk I (MvPolynomial.X i))
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k : Type*} [Field k] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k))Source ↗
local notation "A" => MvPolynomial (Fin m) kSource ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- `df(x̄) ∈ F^n` for an equation `f`. -/ noncomputable def genericGradient (f : MvPolynomial (Fin m) k) : Fin m → F := fun i ↦ algebraMap R F (quotientGradient I f i)
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k F : Type*} [Field k] [Field F] [Algebra k F] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k)) [I.IsPrime] [Algebra (MvPolynomial (Fin m) k ⧸ I) F] [IsScalarTower k (MvPolynomial (Fin m) k ⧸ I) F] [IsFractionRing (MvPolynomial (Fin m) k ⧸ I) F]Source ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- `ξ ↦ Σ ξ_i dx̄_i`, from `F^n` to `Ω_{F/K}`. -/ noncomputable def fieldKaehlerMap : (Fin m → F) →ₗ[F] Ω[F⁄k] := Fintype.linearCombination F fun i ↦ D k F (GenericPointKaehlerConormal.genericPoint I F i)
Variables and notation
Source ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open KaehlerDifferentialSource ↗universe uSource ↗
variable {k F : Type*} [Field k] [Field F] [Algebra k F] {m : ℕ} (I : Ideal (MvPolynomial (Fin m) k)) [I.IsPrime] [Algebra (MvPolynomial (Fin m) k ⧸ I) F] [IsScalarTower k (MvPolynomial (Fin m) k ⧸ I) F] [IsFractionRing (MvPolynomial (Fin m) k ⧸ I) F]Source ↗
local notation "R" => MvPolynomial (Fin m) k ⧸ I
/-- A span is carried into the span of the image under a scalar extension `A → B`. -/ theorem algebraMap_comp_mem_span {A B ι : Type*} [CommRing A] [CommRing B] [Algebra A B] {n : ℕ} (S : ι → Fin n → A) {v : Fin n → A} (hv : v ∈ Submodule.span A (Set.range S)) : (fun i ↦ algebraMap A B (v i)) ∈ Submodule.span B (Set.range fun j i ↦ algebraMap A B (S j i)) := by induction hv using Submodule.span_induction with | mem w hw => obtain ⟨j, rfl⟩ := hw exact Submodule.subset_span ⟨j, rfl⟩ | zero => convert Submodule.zero_mem (Submodule.span B (Set.range fun j i ↦ algebraMap A B (S j i))) simp | add w w' _ _ hw hw' => convert Submodule.add_mem _ hw hw' using 1 funext i simp | smul a w _ hw => convert Submodule.smul_mem _ (algebraMap A B a) hw using 1 funext i simp
/-- **Lemma 5.12(a), `m ≤ d' + 1`, at the place of the proof.** For the discrete valuation ring `V` of the witness, with `Ω_{V/K}` finitely generated (Lemma 5.10(c)), `dim_F Ω_{F/K} ≤ dim_κ Ω_{κ/K} + 1`; this is `lemma_5_12_a_bound` (second fundamental exact sequence for `V → κ`, Nakayama, `F ⊗_V Ω_{V/K} = Ω_{F/K}`), read with the witness's `K`-algebra structure on `κ`. -/ theorem witness_lemma_5_12_a_bound {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hfin : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k])) : let C := w.column let W := C.W let F := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W Module.finrank F (Ω[F⁄k]) ≤ Module.finrank κ (Ω[κ⁄k]) + 1 := by intro C W F letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra intro V letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra intro κ have hfrac : IsFractionRing (↥W.place.valuation) (ComponentFractionField P) := inferInstance letI : Algebra k V := C.groundCoeff.toAlgebra letI : Algebra V F := V.subtype.toAlgebra letI : IsScalarTower k V F := C.groundTower letI : IsFractionRing V F := hfrac letI : Module.Finite V (Ω[V⁄k]) := hfin -- The bound with `κ` a `K`-algebra through `V`, i.e. `c ↦ res (c)`. have hbound := lemma_5_12_a_bound (k := k) (V := V) (F := F) -- That structure is the witness's retained one. have hres : (retainedResidueGroundAlgebra P ⟨0, hm⟩ W : Algebra k κ) = (inferInstance : Algebra k κ) := by apply Algebra.algebra_ext intro c have h := RingHom.congr_fun w.retainedGroundMap c change (@algebraMap k κ _ _ (retainedResidueGroundAlgebra P ⟨0, hm⟩ W)) c = residue V (retainedComponentCoefficientMap P ⟨0, hm⟩ W c) at h rw [h, ← C.groundCoeff_eq_retained] rfl letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W change Module.finrank F (Ω[F⁄k]) ≤ (letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W; Module.finrank κ (Ω[κ⁄k])) + 1 rw [kaehler_finrank_eq_of_algebra_eq _ _ hres] exact hbound
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureNormalizationSource ↗open Stafford38.Geometry.CompletedDVRPowerSeriesSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperResidueGroundAlgebraTransportSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗universe u
/-- **Lemmas 5.12(a), 5.12(b) and 5.13 at the Laurent generic point of the formal proof.** For the column `q(t)` of `witness_lemma_5_11`, with `y = q/q_0 ∈ κ((t))^n`: `dim Θ_y = dim_F Ω_{F/K}`, this equals `trdeg_K F = m`, and `m ≤ d' + 1` by the printed argument (`witness_lemma_5_12_a_bound`: second fundamental exact sequence for `V → κ`, Nakayama with `Ω_{V/K}` finitely generated by Lemma 5.10(c), and `F ⊗_V Ω_{V/K} = Ω_{F/K}`); `dim 𝒞 = dim Θ_y + 1`; `δq ∈ 𝒞` for every `K`-derivation `δ` of `κ((t))` and `q ∈ 𝒞`; and every row annihilating `𝒞` has conormal affine part. -/ theorem witness_lemma_5_12_5_13_paper {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hfin : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k])) : let C := w.column let W := C.W let F := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W let E := LaurentSeries κ letI : Algebra k E := (groundLaurentMap (k := k) (K := κ)).toAlgebra letI : SMul k E := (inferInstance : Algebra k E).toSMul letI : Module k E := (inferInstance : Algebra k E).toModule let q : Fin (m + 1) → E := laurentColumn fun a ↦ retainedToCompletedPowerSeries W (C.q a) (Module.finrank E (tangentSpaceAt q P.asIdeal) = Module.finrank F (Ω[F⁄k]) ∧ (Module.finrank F (Ω[F⁄k]) : Cardinal) = Algebra.trdeg k F ∧ Module.finrank F (Ω[F⁄k]) ≤ Module.finrank κ (Ω[κ⁄k]) + 1 ∧ Module.finrank E (tangentCone q P.asIdeal) = Module.finrank E (tangentSpaceAt q P.asIdeal) + 1) ∧ ((∀ δ : Derivation k E E, (fun a ↦ δ (q a)) ∈ tangentCone q P.asIdeal) ∧ q ∈ tangentCone q P.asIdeal) ∧ (∀ ell : Fin (m + 1) → E, (∀ v ∈ tangentCone q P.asIdeal, ∑ a, ell a * v a = 0) → coordinateCovector (fun i : Fin m ↦ ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint q) (P.asIdeal.map (MvPolynomial.map (algebraMap k E)))) := witness_lemma_5_12_5_13_of_bound hm P w (witness_lemma_5_12_a_bound hm P w hfin)
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Paragraph "the columns of `M` span `𝒞`".** The columns `q`, `Z_{•k}` and `q'` have the form `δq` for `K`-derivations `δ` of `κ((t))` (the `D_k` extended coefficientwise, and `d/dt`), so they lie in `𝒞` by Lemma 5.12(b); `τ = t^{-c}(q' − Zλ)` is a `κ((t))`-combination of them. With `dim 𝒞 = dim Θ_y + 1` (Lemma 5.12(a)) and `dim Θ_y ≤ d' + 1`, the `d' + 2` independent columns of `M` span `𝒞`. -/ theorem step_M_spans_cone {m d : ℕ} (I : Ideal (MvPolynomial (Fin m) k)) (q : Fin (m + 1) → PowerSeries κ) (hq0 : q 0 ≠ 0) (hbase : ∀ f ∈ I, MvPolynomial.eval₂ (algebraMap k (LaurentSeries κ)) (dehomogenizedPoint (laurentColumn q)) f = 0) (D : Fin d → Derivation k κ κ) (Z : Matrix (Fin (m + 1)) (Fin d) (PowerSeries κ)) (hZ : Z = coefficientwiseTangentMatrix q D) (lambda : Fin d → PowerSeries κ) (c : ℕ) (tau : Fin (m + 1) → PowerSeries κ) (hfactor : ∀ i, PowerSeries.derivative (R := κ) (q i) - Z.mulVec lambda i = (PowerSeries.X : PowerSeries κ) ^ c * tau i) (hli : LinearIndependent (LaurentSeries κ) (fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col))) (hbound : Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I) ≤ d + 1) : (∀ col, (fun i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col)) ∈ tangentCone (laurentColumn q) I) ∧ Submodule.span (LaurentSeries κ) (Set.range fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (formalTangentMatrix q Z tau i col)) = tangentCone (laurentColumn q) I := by classical let L := LaurentSeries κ let f := algebraMap (PowerSeries κ) L let qL := laurentColumn q let 𝒞 := tangentCone qL I have hq0L : qL 0 ≠ 0 := laurentColumn_ne_zero_of_ne_zero q hq0 -- Lemma 5.12(b) at `y = q/q₀`. obtain ⟨hδ, hq⟩ := lemma_5_12_b qL hq0L I hbase have hder : ∀ δ : Derivation k (PowerSeries κ) (PowerSeries κ), laurentColumn (fun i ↦ δ (q i)) ∈ 𝒞 := by intro δ have h := hδ (extendPowerSeriesDerivation δ) have hcol : (fun a ↦ extendPowerSeriesDerivation δ (qL a)) = laurentColumn (fun i ↦ δ (q i)) := by funext a exact extendPowerSeriesDerivation_on_powerSeries δ (q a) rw [hcol] at h exact h have hZcol : ∀ l, (fun i ↦ f (Z i l)) ∈ 𝒞 := by intro l have h := hder (coefficientwiseDerivation (D l)) rw [hZ] exact h have hq' : (fun i ↦ f (PowerSeries.derivative (R := κ) (q i))) ∈ 𝒞 := hder (uniformizerDerivation (k := k) (K := κ)) -- `τ = t^{-c} (q' − Σ λ_l Z_{•l})`. let Xc : L := f ((PowerSeries.X : PowerSeries κ) ^ c) have hXc : Xc ≠ 0 := by simp [Xc, f] have htau : (fun i ↦ f (tau i)) ∈ 𝒞 := by have hcomb := 𝒞.smul_mem Xc⁻¹ (𝒞.sub_mem hq' (𝒞.sum_mem fun l (_ : l ∈ Finset.univ) ↦ 𝒞.smul_mem (f (lambda l)) (hZcol l))) convert hcomb using 1 funext i have hi := congrArg f (hfactor i) simp only [map_sub, map_mul, Matrix.mulVec, dotProduct, map_sum] at hi simp only [Pi.smul_apply, Pi.sub_apply, Finset.sum_apply, smul_eq_mul] rw [eq_inv_mul_iff_mul_eq₀ hXc] rw [← hi] congr 1 refine Finset.sum_congr rfl fun l _ ↦ ?_ ring have hcols : ∀ col, (fun i ↦ f (formalTangentMatrix q Z tau i col)) ∈ 𝒞 := by intro col rcases col with _ | (l | _) · exact hq · exact hZcol l · exact htau refine ⟨hcols, ?_⟩ have hle : Submodule.span L (Set.range fun col i ↦ f (formalTangentMatrix q Z tau i col)) ≤ 𝒞 := Submodule.span_le.mpr (by rintro _ ⟨col, rfl⟩; exact hcols col) apply Submodule.eq_of_le_of_finrank_le hle rw [finrank_span_eq_card hli] -- Lemma 5.12(a): `dim 𝒞 = dim Θ_y + 1 ≤ d' + 2`. have hcone : Module.finrank (LaurentSeries κ) (tangentCone (laurentColumn q) I) = Module.finrank (LaurentSeries κ) (tangentSpaceAt (laurentColumn q) I) + 1 := lemma_5_12_a_cone (laurentColumn q) hq0L I change Module.finrank (LaurentSeries κ) (tangentCone (laurentColumn q) I) ≤ _ simp only [Fintype.card_sum, Fintype.card_unit, Fintype.card_fin] omega
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
Comparison notes
Lean states the lemma for a column q with q₀ ≠ 0 over a field L ⊇ K, with Θ_y and 𝒞 = ρ⁻¹(Θ_y) as in the text, and applies it at the Laurent point of the proof of Theorem 5.7 (witness_lemma_5_12_5_13_paper). (a) follows the printed proof: quotientKaehlerMap_exact is the second fundamental exact sequence for K[x] → R = K[x]/𝔭, Ω_{R/K} = R^n/span_R{df(x̄) : f ∈ 𝔭}; fieldKaehlerMap_exact is its localization, Ω_{F/K} = F^n/span_F{df(x̄) : f ∈ 𝔭}; generic_tangent_finrank_eq_component_kaehler_finrank gives dim Θ_y = dim_F Ω_{F/K} over every field E ⊇ F (in the paper E = κ((t))), since rank does not change under F ⊆ E and Θ_y is the common kernel of the df_y; component_kaehler_finrank_eq_trdeg gives dim_F Ω_{F/K} = trdeg_K F = m; lemma_5_12_a_bound gives m ≤ d′ + 1 for a discrete valuation ring with Ω_{V/K} finitely generated, by the second fundamental sequence for V → κ, Nakayama and F ⊗_V Ω_{V/K} = Ω_{F/K} (witness_lemma_5_12_a_bound at V); and lemma_5_12_a_cone gives dim 𝒞 = dim Θ_y + 1. (b) is lemma_5_12_b. The proof of Theorem 5.7 uses (a) and (b) in step_M_spans_cone.
Review notes
Lemma 5.13 — Annihilators of the tangent cone are conormal
/-- **Lemma 5.13.** A row vector annihilating `𝒞` has its affine part `(ℓ_1, …, ℓ_n)` in `span {df_y : f ∈ 𝔭}`. -/ theorem lemma_5_13 (q : Fin (n + 1) → L) (hq0 : q 0 ≠ 0) (I : Ideal (MvPolynomial (Fin n) k)) (ell : Fin (n + 1) → L) (hell : ∀ v ∈ tangentCone q I, ∑ a, ell a * v a = 0) : coordinateCovector (fun i : Fin n ↦ ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint q) (I.map (MvPolynomial.map (algebraMap k L))) := by rw [← affineConormalSpace_eq_equationCovectorSpan, affineConormalSpace, Submodule.mem_dualAnnihilator] intro w hw -- `ρ(𝒞) = Θ_y`. have hw' : w ∈ (tangentCone q I).map (rho q) := by rw [rho_map_tangentCone q hq0] exact hw obtain ⟨v, hv, rfl⟩ := hw' have hq : q ∈ tangentCone q I := by change rho q q ∈ tangentSpaceAt q I have : rho q q = 0 := by rw [← LinearMap.mem_ker, ker_rho q hq0] exact Submodule.mem_span_singleton_self q rw [this] exact zero_mem _ -- `ℓ(q) = 0` and `ℓ(v) = 0` give `Σ ℓ_i ρ(v)_i = 0`. have hellq := hell q hq have hellv := hell v hv rw [Fin.sum_univ_succ] at hellq hellv rw [coordinateCovector_apply] simp only [rho_apply] have hsum : ∑ i : Fin n, ell i.succ * ((v i.succ * q 0 - q i.succ * v 0) / q 0 ^ 2) = (q 0 * ∑ i : Fin n, ell i.succ * v i.succ - v 0 * ∑ i : Fin n, ell i.succ * q i.succ) / q 0 ^ 2 := by rw [Finset.mul_sum, Finset.mul_sum, ← Finset.sum_sub_distrib, Finset.sum_div] refine Finset.sum_congr rfl fun i _ ↦ ?_ ring rw [hsum] have h1 : ∑ i : Fin n, ell i.succ * v i.succ = -(ell 0 * v 0) := by linear_combination hellv have h2 : ∑ i : Fin n, ell i.succ * q i.succ = -(ell 0 * q 0) := by linear_combination hellq rw [h1, h2] ring_nf
More Lean proofs and definitions
/-- **Lemmas 5.12(a), 5.12(b) and 5.13 at the Laurent generic point of the formal proof.** For the column `q(t)` of `witness_lemma_5_11`, with `y = q/q_0 ∈ κ((t))^n`: `dim Θ_y = dim_F Ω_{F/K}`, this equals `trdeg_K F = m`, and `m ≤ d' + 1` by the printed argument (`witness_lemma_5_12_a_bound`: second fundamental exact sequence for `V → κ`, Nakayama with `Ω_{V/K}` finitely generated by Lemma 5.10(c), and `F ⊗_V Ω_{V/K} = Ω_{F/K}`); `dim 𝒞 = dim Θ_y + 1`; `δq ∈ 𝒞` for every `K`-derivation `δ` of `κ((t))` and `q ∈ 𝒞`; and every row annihilating `𝒞` has conormal affine part. -/ theorem witness_lemma_5_12_5_13_paper {k : Type u} [Field k] [CharZero k] {m : ℕ} (hm : 0 < m) (P : PrimeSpectrum (MvPolynomial (Fin m) k)) (w : GeneralDivisorialVisibleFrameWitness hm P) (hfin : letI : Algebra (CoordinateZeroLocalRing w.column.W.coefficientField) (ComponentFractionField P) := w.column.W.ambientAlgebra let V := w.column.W.place.valuation.toSubring letI : Algebra k V := w.column.groundCoeff.toAlgebra Module.Finite V (Ω[V⁄k])) : let C := w.column let W := C.W let F := ComponentFractionField P letI : Algebra (CoordinateZeroLocalRing W.coefficientField) F := W.ambientAlgebra let V := W.place.valuation.toSubring letI : IsDiscreteValuationRing V := W.place.isDiscrete letI : Algebra W.coefficientField V := (relativeCoefficientMap W.coefficientField W.place).toAlgebra let κ := ResidueField V letI : Algebra k κ := retainedResidueGroundAlgebra P ⟨0, hm⟩ W let E := LaurentSeries κ letI : Algebra k E := (groundLaurentMap (k := k) (K := κ)).toAlgebra letI : SMul k E := (inferInstance : Algebra k E).toSMul letI : Module k E := (inferInstance : Algebra k E).toModule let q : Fin (m + 1) → E := laurentColumn fun a ↦ retainedToCompletedPowerSeries W (C.q a) (Module.finrank E (tangentSpaceAt q P.asIdeal) = Module.finrank F (Ω[F⁄k]) ∧ (Module.finrank F (Ω[F⁄k]) : Cardinal) = Algebra.trdeg k F ∧ Module.finrank F (Ω[F⁄k]) ≤ Module.finrank κ (Ω[κ⁄k]) + 1 ∧ Module.finrank E (tangentCone q P.asIdeal) = Module.finrank E (tangentSpaceAt q P.asIdeal) + 1) ∧ ((∀ δ : Derivation k E E, (fun a ↦ δ (q a)) ∈ tangentCone q P.asIdeal) ∧ q ∈ tangentCone q P.asIdeal) ∧ (∀ ell : Fin (m + 1) → E, (∀ v ∈ tangentCone q P.asIdeal, ∑ a, ell a * v a = 0) → coordinateCovector (fun i : Fin m ↦ ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint q) (P.asIdeal.map (MvPolynomial.map (algebraMap k E)))) := witness_lemma_5_12_5_13_of_bound hm P w (witness_lemma_5_12_a_bound hm P w hfin)
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe u
/-- **Paragraph "`ℓ` annihilates `𝒞`".** If the columns of `M` span `𝒞` and `ℓ M = 0`, then `ℓ` annihilates `𝒞`, so by Lemma 5.13 its affine part `ℓ̃ = (ℓ_1, …, ℓ_n)` lies in `span {df_y : f ∈ 𝔭}`; and `ℓ(0) = u` gives `ℓ̃(0) = (1, 0, …, 0)`. -/ theorem step_ell_conormal {m : ℕ} (hm : 0 < m) {ι : Type*} [Fintype ι] (I : Ideal (MvPolynomial (Fin m) k)) (q : Fin (m + 1) → PowerSeries κ) (hq0 : q 0 ≠ 0) (M : Matrix (Fin (m + 1)) ι (PowerSeries κ)) (hspan : Submodule.span (LaurentSeries κ) (Set.range fun col i ↦ algebraMap (PowerSeries κ) (LaurentSeries κ) (M i col)) = tangentCone (laurentColumn q) I) (ell : Fin (m + 1) → PowerSeries κ) (hell : rowMul ell M = 0) (hres : residueColumn ell = axisRow (k := κ) (Fin.succ ⟨0, hm⟩)) : coordinateCovector (fun i : Fin m ↦ laurentColumn ell i.succ) ∈ equationCovectorSpan (dehomogenizedPoint (laurentColumn q)) (I.map (MvPolynomial.map (algebraMap k (LaurentSeries κ)))) ∧ residueColumn (fun i : Fin m ↦ ell i.succ) = fun i ↦ if i = ⟨0, hm⟩ then 1 else 0 := by classical let L := LaurentSeries κ let f := algebraMap (PowerSeries κ) L let φ : (Fin (m + 1) → L) →ₗ[L] L := { toFun := fun v ↦ ∑ a, laurentColumn ell a * v a map_add' := fun v w ↦ by simp [mul_add, Finset.sum_add_distrib] map_smul' := fun c v ↦ by simp only [Pi.smul_apply, smul_eq_mul, RingHom.id_apply] rw [Finset.mul_sum] exact Finset.sum_congr rfl fun a _ ↦ by ring } have hle : tangentCone (laurentColumn q) I ≤ LinearMap.ker φ := by rw [← hspan, Submodule.span_le] rintro _ ⟨col, rfl⟩ have h := congrArg f (congrFun hell col) simp only [rowMul, map_sum, map_mul, Pi.zero_apply, map_zero] at h exact h have hann : ∀ v ∈ tangentCone (laurentColumn q) I, ∑ a, laurentColumn ell a * v a = 0 := fun v hv ↦ hle hv refine ⟨lemma_5_13 (laurentColumn q) (laurentColumn_ne_zero_of_ne_zero q hq0) I (laurentColumn ell) hann, ?_⟩ calc residueColumn (fun i : Fin m ↦ ell i.succ) = (fun i : Fin m ↦ residueColumn ell i.succ) := residueColumn_tail ell _ = (fun i : Fin m ↦ axisRow (k := κ) (Fin.succ ⟨0, hm⟩) i.succ) := by rw [hres] _ = (fun i : Fin m ↦ if i = ⟨0, hm⟩ then 1 else 0) := by funext i by_cases hi : i = ⟨0, hm⟩ · subst i simp [axisRow] · have hne : i.succ ≠ Fin.succ ⟨0, hm⟩ := fun h ↦ hi (Fin.succ_injective m h) rw [axisRow_apply_of_ne hne] simp [hi]
Variables and notation
Source ↗open IsLocalRingSource ↗open Stafford38.Geometry.AffineComponentCoordinateSplitSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.ActualWitnessConormalAssemblySource ↗open Stafford38.Geometry.AsymptoticDivisorExistenceSource ↗open Stafford38.Geometry.ComponentFunctionFieldBoundarySource ↗open Stafford38.Geometry.ComponentProjectiveClosureSource ↗open Stafford38.Geometry.ConormalScalarExtensionVanishingSource ↗open Stafford38.Geometry.ContinuousPowerSeriesTangentFrameSource ↗open Stafford38.Geometry.ExactDivisorialVisibleFrameExistenceSource ↗open Stafford38.Geometry.FormalDivisorLaurentConormalSource ↗open Stafford38.Geometry.GeneralDivisorialVisibleFrameSource ↗open Stafford38.Geometry.KaehlerVisibleDerivationFrameSource ↗open Stafford38.Geometry.LaurentConormalDirectionSource ↗open Stafford38.Geometry.LaurentConormalResidueExtensionSource ↗open Stafford38.Geometry.PaperActualDivisorTangentSource ↗open Stafford38.Geometry.PaperLaurentChartDerivationSource ↗open Stafford38.Geometry.PaperResidueDerivationFrameSource ↗open Stafford38.Geometry.PaperRetainedChartAssemblySource ↗open Stafford38.Geometry.PaperSameWitnessComponentFieldRankSource ↗open Stafford38.Geometry.ProjectiveConormalDehomogenizationSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.RelativeCoefficientDVRSource ↗open Stafford38.Geometry.RelativeRetainedBoundaryPlaceSource ↗open Stafford38.Geometry.RetainedGroundMapIdentificationSource ↗open Stafford38.Geometry.RetainedPlaceConormalTransportSource ↗open Stafford38.Geometry.RetainedProjectiveCompletionSource ↗open Stafford38.Geometry.ScalarExtensionPointsSource ↗open Stafford38.Geometry.SmoothConormalFibreVanishingSource ↗open Stafford38.GeometryFormalDivisorAxisLiftSource ↗open Stafford38.GeometryFormalDivisorTangentSource ↗open Stafford38.GeometryPowerSeriesTangentLimitSource ↗open Stafford38.GeometryResidueMinorSelectionSource ↗open Stafford38.GeometryRetractionSpecializationSource ↗open Stafford38.GeometrySplitTangentMatrixSource ↗universe uSource ↗
variable {k κ : Type u} [Field k] [Field κ] [Algebra k κ] [CharZero k] [CharZero κ]
Comparison notes
lemma_5_13: for q with q₀ ≠ 0, a row vector ℓ annihilating 𝒞 has affine part (ℓ₁, …, ℓ_n) in span{df_y : f ∈ 𝔭}. The proof is the printed one: ℓ(q) = 0 and ℓ(v) = 0 give Σ ℓ_iρ(v)_i = 0, and ρ(𝒞) = Θ_y. witness_lemma_5_12_5_13_paper applies it at the Laurent point of the proof of Theorem 5.7, which uses it in step_ell_conormal.
Review notes
5.3. Proof of Theorem 5.8
Proof of 5.8 — Proof of the involutive exclusion theorem
Proof of Theorem 5.8
We will give a proof by contradiction, with this rough logic: Suppose and let . Since is by assumption fibre-conical, we start by showing that and . Using involutivity, we then show that we can translate along in the direction of the fibres which means that all conormal covectors of every irreducible component of lie in . This means vanishes on all of , hence on . This gives us the setup of Theorem 5.7, which implies a contradiction to our assumption.
We define the (radical) vanishing ideal consisting of all polynomials that vanish at every point , and , consisting of the functions with for every .
Because is fibre-conical, the -homogeneous components of any
polynomial in again lie in : if and
, then for all ,
and a polynomial in with infinitely many roots is zero, so every
and therefore .
Let be written in its -homogeneous components form. In particular, . Since has -degree , it is a polynomial in alone, so . Let . For we therefore have , so every vanishes at , i.e. . , since is closed [hartshorne, Prop. I.1.2], hence , or .
Furthermore, is the projection of and is therefore per definition of closed: since for ( and , we have so , and , since if , then also , so .
Lastly, assume there exists a , then and , which would put , which is wrong by our assumption. Therefore, .
For the next argument, we also refer to [Hotta 2008, Ch. E.2]. Let . Then . The Hamiltonian derivation of applied to is
Since , , so must have at least one minimal prime. The ideal is radical (by construction since it is a vanishing ideal) in the Noetherian ring , so has only finitely many minimal primes and [hartshorne, Cor. I.1.4 and Prop. I.1.5]. We choose and set . Since is prime, , so by Theorem 2.11; moreover is closed and irreducible since is prime [hartshorne, Cor. I.1.4], and since primeness implies is radical, by Theorem 2.11. Additionally, , hence , as we have already established .
Furthermore, if , we choose , then , and if , choose for every elements . Then does not lie in since no factor lies in but does lie in every other since they are ideals and one of the factors lies in them. On the other hand, if , then since , and since for all . Therefore, i.e. We now fix such a For a , Since and since , so and since also . For , the polynomial therefore vanishes on , and vanishes on all of , so and therefore , which means that vanishes on all of . Since this is true for every , we have for all , so As before, we can argue that since is an ideal, it is closed under -linear combinations, so for and , also , or put differently, for all . At a smooth point of , by Definition 5.4, the conormal space is
/-- **Theorem 5.8.** Let `K` be algebraically closed of characteristic zero and `W ⊆ T*X` Zariski closed, fibre-conical and involutive. If `P ∈ K[ξ_1, …, ξ_n]` with `P(1, 0, …, 0) ≠ 0` vanishes on `W` and `W ∩ π⁻¹(H) = ∅`, then `W = ∅`. Proved along Subsection 5.3. -/ theorem theorem_5_8 (hm : 0 < m) (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (P : MvPolynomial (Fin m) k) (haxis : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P ≠ 0) (hPW : ∀ q ∈ W, MvPolynomial.eval (fun i => q (.inr i)) P = 0) (hH : W ∩ {q | q (.inl ⟨0, hm⟩) = 0} = ∅) : W = ∅ := by -- Suppose `W ≠ ∅`. by_contra hne have hW : W.Nonempty := Set.nonempty_iff_ne_empty.2 hne -- `B` is radical with a minimal prime `𝔮`; `Y_0 = V_K(𝔮)`. obtain ⟨hrad, -, -, 𝔮, h𝔮⟩ := step_minimal_primes W hclosed hcone hW obtain ⟨hprime, -, -, -, havoid⟩ := step_component hm W hclosed hcone hH 𝔮 h𝔮 -- The separator `s`, then `P` vanishes on `N(Y_0)`. obtain ⟨s, hs𝔮, hs⟩ := step_separator W 𝔮 hrad h𝔮 have hPN := step_P_vanishes_on_conormal_cone W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs P hPW -- Theorem 5.7(i) contradicts `P(1, 0, …, 0) ≠ 0`. exact haxis (step_theorem_5_7_applies hm 𝔮 hprime havoid P (step_P_vanishes_on_N 𝔮 hprime P hPN))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
More Lean proofs and definitions
/-- Paragraph "Because `W` is fibre-conical": the `ξ`-homogeneous components `g_d` of every `g ∈ J` lie in `J`. The proof is the paper's: `g(y, aξ) = Σ a^d g_d(y, ξ)` vanishes for all `a ∈ K^×`, so every coefficient does. -/ theorem step_homogeneous_components (W : Set (PhaseVar m → k)) (hcone : IsFibreConical W) : ∀ g ∈ MvPolynomial.vanishingIdeal k W, ∀ d : ℕ, (DirectSum.decompose (orderDecomposition (k := k) (n := m)) g d : SymbolRing k m) ∈ MvPolynomial.vanishingIdeal k W := fun _ hg d => vanishingIdeal_isHomogeneous_of_isFibreConical W hcone d hg
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: `g_0 ∈ B`, hence `g(y, 0) = g_0(y) = 0` for `y ∈ Y = V_K(B)`, and `Y × {0} ⊆ V_K(J) = W` because `W` is closed. -/ theorem step_zero_section (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) : ∀ y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W), Sum.elim y (0 : Fin m → k) ∈ W := by intro y hy have hhom : (MvPolynomial.vanishingIdeal k W).IsHomogeneous (orderDecomposition (k := k) (n := m)) := fun d _ hg => step_homogeneous_components W hcone _ hg d have hzero := zeroSection_commonZero_of_isHomogeneous (MvPolynomial.vanishingIdeal k W) hhom y hy have hpoint : Sum.elim y (0 : Fin m → k) = zeroSectionPoint y := by funext i; rcases i with i | i <;> rfl rw [← zeroLocus_vanishingIdeal_of_algebraic_closed W hclosed, hpoint] exact hzero
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: `Y = π(W)`. -/ theorem step_projection (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) : {y : Fin m → k | ∃ ξ : Fin m → k, Sum.elim y ξ ∈ W} = MvPolynomial.zeroLocus k (exclusionBaseIdeal W) := by ext y constructor · rintro ⟨ξ, hξ⟩ f hf have h := hf (Sum.elim y ξ) hξ simpa [baseLift, MvPolynomial.eval_rename, Function.comp_def] using h · intro hy exact ⟨0, step_zero_section W hclosed hcone y hy⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Paragraph "Lastly": `W ∩ π⁻¹(H) = ∅` gives `Y ∩ H = ∅`. -/ theorem step_Y_avoids_H (hm : 0 < m) (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hH : W ∩ {q | q (.inl ⟨0, hm⟩) = 0} = ∅) : ∀ y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W), y ⟨0, hm⟩ ≠ 0 := by intro y hy hy0 have hmem : Sum.elim y (0 : Fin m → k) ∈ W ∩ {q | q (.inl ⟨0, hm⟩) = 0} := ⟨step_zero_section W hclosed hcone y hy, hy0⟩ rw [hH] at hmem exact hmem omit [IsAlgClosed k] [CharZero k] in
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Paragraph "For the next argument": for `f ∈ B`, `H_f g = Σ_i ∂f/∂x_i ∂g/∂ξ_i = {f, g}` (the bracket of Definition 2.9, `Section2.poisson`), and involutivity gives `H_f^m g ∈ J` for all `g ∈ J`, `m ≥ 0`. -/ theorem step_hamiltonian (W : Set (PhaseVar m → k)) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (f : MvPolynomial (Fin m) k) (hf : f ∈ exclusionBaseIdeal W) (g : SymbolRing k m) (hg : g ∈ MvPolynomial.vanishingIdeal k W) : Stafford38.Paper.Section2.poisson k (baseLift f) g = ∑ i : Fin m, baseLift (MvPolynomial.pderiv i f) * MvPolynomial.pderiv (Sum.inr i : PhaseVar m) g ∧ ∀ j : ℕ, hamiltonIter f j g ∈ MvPolynomial.vanishingIdeal k W := by refine ⟨?_, fun j => ?_⟩ · rw [Stafford38.Paper.Section2.poisson_eq] simp [poissonBracket, pderiv_baseLift_base, pderiv_baseLift_fibre] · induction j with | zero => exact hg | succ j hj => exact hinv _ hf _ hj
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: for `y ∈ Y`, `f ∈ B`, `g ∈ J`, the polynomial `φ(t) = g(y, t df_y)` satisfies `φ^{(j)}(0) = H_f^j g(y, 0) = 0` for all `j`. -/ theorem step_phi_derivatives (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W)) (f : MvPolynomial (Fin m) k) (hf : f ∈ exclusionBaseIdeal W) (g : SymbolRing k m) (hg : g ∈ MvPolynomial.vanishingIdeal k W) (j : ℕ) : Polynomial.eval 0 ((Polynomial.derivative^[j]) (fibreLinePolynomial y (differentialAt y f) g)) = 0 := by rw [iterate_derivative_fibreLinePolynomial, eval_zero_fibreLinePolynomial] have hpoint : zeroSectionPoint y = Sum.elim y (0 : Fin m → k) := by funext i; rcases i with i | i <;> rfl rw [hpoint] exact (step_hamiltonian W hinv f hf g hg).2 j _ (step_zero_section W hclosed hcone y hy)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: since `char K = 0`, `φ^{(j)}(0) = j! c_j = 0` forces `φ = 0`, so `(y, t df_y) ∈ V_K(J) = W` for every `t ∈ K`. -/ theorem step_line_mem (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W)) (f : MvPolynomial (Fin m) k) (hf : f ∈ exclusionBaseIdeal W) (t : k) : Sum.elim y (t • differentialAt y f) ∈ W := by rw [← zeroLocus_vanishingIdeal_of_algebraic_closed W hclosed] intro g hg have hphi : fibreLinePolynomial y (differentialAt y f) g = 0 := polynomial_eq_zero_of_eval_iterate_derivative_zero _ (step_phi_derivatives W hclosed hcone hinv y hy f hf g hg) have h := eval_fibreLinePolynomial y (differentialAt y f) g t rw [hphi, Polynomial.eval_zero] at h have hpt : Sum.elim y (t • differentialAt y f) = Sum.elim y (fun i => t * differentialAt y f i) := rfl rw [hpt] exact h.symm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph, last sentence: `{y} × span{df_y : f ∈ B} ⊆ W` for `y ∈ Y`, since `Σ c_a df_a = d(Σ c_a f_a)` with `Σ c_a f_a ∈ B`, at `t = 1`. -/ theorem step_span_mem (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W)) (ξ : Fin m → k) (hξ : ξ ∈ Submodule.span k (Set.range fun f : exclusionBaseIdeal W => differentialAt y f.1)) : Sum.elim y ξ ∈ W := by obtain ⟨f, hf, rfl⟩ := exists_mem_differentialAt_eq_of_mem_span _ y ξ hξ simpa using step_line_mem W hclosed hcone hinv y hy f hf 1
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Paragraph "Since `V_K(B) = Y = π(W) ≠ ∅`": `B` is radical and proper (by `step_projection`), has finitely many minimal primes `𝔮_1, …, 𝔮_r`, at least one, and `B = ⋂ 𝔮_i`. -/ theorem step_minimal_primes (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hne : W.Nonempty) : (exclusionBaseIdeal W).IsRadical ∧ (exclusionBaseIdeal W).minimalPrimes.Finite ∧ sInf (exclusionBaseIdeal W).minimalPrimes = exclusionBaseIdeal W ∧ (exclusionBaseIdeal W).minimalPrimes.Nonempty := by have hrad : (exclusionBaseIdeal W).IsRadical := (vanishingIdeal_isRadical W).comap _ have hproper : exclusionBaseIdeal W ≠ ⊤ := by intro htop obtain ⟨q, hq⟩ := hne have hy : (fun i => q (.inl i)) ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W) := by rw [← step_projection W hclosed hcone] refine ⟨fun i => q (.inr i), ?_⟩ have hsplit : Sum.elim (fun i => q (.inl i)) (fun i => q (.inr i)) = q := by funext i; cases i <;> rfl rwa [hsplit] have := hy 1 (htop ▸ trivial) simp at this refine ⟨hrad, Ideal.finite_minimalPrimes_of_isNoetherianRing _ _, ?_, Set.nonempty_coe_sort.mp (Ideal.nonempty_minimalPrimes hproper)⟩ rw [Ideal.sInf_minimalPrimes, hrad.radical]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: for a minimal prime `𝔮` of `B`, `Y_0 = V_K(𝔮)` is nonempty, closed and irreducible (`𝔮` prime), `𝔮 = I(Y_0)`, `Y_0 ⊆ Y` and `Y_0 ∩ H = ∅`. -/ theorem step_component (hm : 0 < m) (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hH : W ∩ {q | q (.inl ⟨0, hm⟩) = 0} = ∅) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) : 𝔮.IsPrime ∧ (MvPolynomial.zeroLocus k 𝔮).Nonempty ∧ MvPolynomial.vanishingIdeal k (MvPolynomial.zeroLocus k 𝔮) = 𝔮 ∧ MvPolynomial.zeroLocus k 𝔮 ⊆ MvPolynomial.zeroLocus k (exclusionBaseIdeal W) ∧ ∀ y ∈ MvPolynomial.zeroLocus k 𝔮, y ⟨0, hm⟩ ≠ 0 := by have hprime : 𝔮.IsPrime := h𝔮.1.1 have hI : MvPolynomial.vanishingIdeal k (MvPolynomial.zeroLocus k 𝔮) = 𝔮 := MvPolynomial.IsPrime.vanishingIdeal_zeroLocus 𝔮 have hsub : MvPolynomial.zeroLocus k 𝔮 ⊆ MvPolynomial.zeroLocus k (exclusionBaseIdeal W) := MvPolynomial.zeroLocus_anti_mono h𝔮.1.2 refine ⟨hprime, ?_, hI, hsub, fun y hy => step_Y_avoids_H hm W hclosed hcone hH y (hsub hy)⟩ rw [Set.nonempty_iff_ne_empty] intro hempty rw [hempty, MvPolynomial.vanishingIdeal_empty] at hI exact hprime.ne_top hI.symm omit [IsAlgClosed k] [CharZero k] in
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Paragraph "Furthermore": there is a separator `s ∉ 𝔮` with `s𝔮 ⊆ B`, `s = s_2 ⋯ s_r` with `s_i ∈ 𝔮_i ∖ 𝔮`. -/ theorem step_separator (W : Set (PhaseVar m → k)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (hrad : (exclusionBaseIdeal W).IsRadical) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) : ∃ s ∉ 𝔮, ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W := by classical set B := exclusionBaseIdeal W have hfin := Ideal.finite_minimalPrimes_of_isNoetherianRing (MvPolynomial (Fin m) k) B set T := hfin.toFinset.erase 𝔮 with hT have hchoice : ∀ Q ∈ T, ∃ s ∈ Q, s ∉ 𝔮 := by intro Q hQ rw [hT, Finset.mem_erase, hfin.mem_toFinset] at hQ by_contra hcon push Not at hcon exact hQ.1 (le_antisymm hcon (h𝔮.2 hQ.2.1 hcon)) choose s hsQ hsP using hchoice refine ⟨∏ Q ∈ T.attach, s Q.1 Q.2, ?_, ?_⟩ · intro hs obtain ⟨Q, -, hQ⟩ := (h𝔮.1.1.prod_mem_iff).1 hs exact hsP Q.1 Q.2 hQ · intro h hh rw [← hrad.radical, ← Ideal.sInf_minimalPrimes, Ideal.mem_sInf] intro Q hQ by_cases hQP : Q = 𝔮 · subst hQP exact Ideal.mul_mem_left _ _ hh · have hQT : Q ∈ T := by rw [hT, Finset.mem_erase, hfin.mem_toFinset] exact ⟨hQP, hQ⟩ refine Ideal.mul_mem_right _ _ ?_ exact Ideal.mem_of_dvd _ (Finset.dvd_prod_of_mem (fun Q : T ↦ s Q.1 Q.2) (Finset.mem_attach _ ⟨Q, hQT⟩)) (hsQ Q hQT)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: for `h ∈ 𝔮` and `y ∈ Y_0 ∩ {s ≠ 0}`, `(y, dh_y) ∈ W`, because `d(sh)_y = s(y) dh_y` with `sh ∈ B`, and `W` is fibre-conical. -/ theorem step_dh_on_open (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) (s : MvPolynomial (Fin m) k) (hs : ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W) (h : MvPolynomial (Fin m) k) (hh : h ∈ 𝔮) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k 𝔮) (hsy : MvPolynomial.eval y s ≠ 0) : Sum.elim y (differentialAt y h) ∈ W := by have hyY : y ∈ MvPolynomial.zeroLocus k (exclusionBaseIdeal W) := MvPolynomial.zeroLocus_anti_mono h𝔮.1.2 hy have hhy : MvPolynomial.eval y h = 0 := hy h hh have hd : differentialAt y (s * h) = MvPolynomial.eval y s • differentialAt y h := by funext i simp [differentialAt, hhy] have hsh : Sum.elim y (differentialAt y (s * h)) ∈ W := step_span_mem W hclosed hcone hinv y hyY _ (Submodule.subset_span ⟨⟨s * h, hs h hh⟩, rfl⟩) rw [hd] at hsh have := ((definition_5_6 W).1 hcone y _ ⟨hsh, rfl⟩ (MvPolynomial.eval y s)⁻¹ (inv_ne_zero hsy)).1 rwa [smul_smul, inv_mul_cancel₀ hsy, one_smul] at this
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: `(y, dh_y) ∈ W` for all `y ∈ Y_0`. For `g ∈ J`, `G(x) = g(x, dh_x)` vanishes on `Y_0 ∩ {s ≠ 0}`, so `sG ∈ I(Y_0) = 𝔮`, and `G ∈ 𝔮` as `s ∉ 𝔮`. -/ theorem step_dh_on_component (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) (s : MvPolynomial (Fin m) k) (hs𝔮 : s ∉ 𝔮) (hs : ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W) (h : MvPolynomial (Fin m) k) (hh : h ∈ 𝔮) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k 𝔮) : Sum.elim y (differentialAt y h) ∈ W := by rw [← zeroLocus_vanishingIdeal_of_algebraic_closed W hclosed] intro g hg have hG : gradientGraphSubstitution h g ∈ 𝔮 := by apply mem_of_vanishes_on_principal_open 𝔮 h𝔮.1.1 hs𝔮 intro z hz hsz rw [eval_gradientGraphSubstitution] exact hg _ (step_dh_on_open W hclosed hcone hinv 𝔮 h𝔮 s hs h hh z hz hsz) have h0 : MvPolynomial.eval y (gradientGraphSubstitution h g) = 0 := hy _ hG rw [eval_gradientGraphSubstitution] at h0 exact h0
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: `{y} × span{dh_y : h ∈ 𝔮} ⊆ W` for every `y ∈ Y_0`. -/ theorem step_span_component (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) (s : MvPolynomial (Fin m) k) (hs𝔮 : s ∉ 𝔮) (hs : ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W) (y : Fin m → k) (hy : y ∈ MvPolynomial.zeroLocus k 𝔮) (ξ : Fin m → k) (hξ : ξ ∈ Submodule.span k (Set.range fun h : 𝔮 => differentialAt y h.1)) : Sum.elim y ξ ∈ W := by obtain ⟨h, hh, rfl⟩ := exists_mem_differentialAt_eq_of_mem_span 𝔮 y ξ hξ exact step_dh_on_component W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs h hh y hy
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: at smooth points `y` of `Y_0`, `{y} × N*_y Y_0 ⊆ W`, with `N*_y Y_0 = span{df_y : f ∈ 𝔮}` (Definition 5.4(c), `𝔮 = I(Y_0)`). -/ theorem step_conormal_subset (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) (s : MvPolynomial (Fin m) k) (hs𝔮 : s ∉ 𝔮) (hs : ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W) (y : Fin m → k) (hy : SmoothAffinePoint 𝔮 y) (ξ : Fin m → k) (hξ : coordinateCovector ξ ∈ affineConormalSpace y 𝔮) : Sum.elim y ξ ∈ W := step_span_component W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs y (smoothAffinePoint_mem_zeroLocus 𝔮 hy) ξ ((definition_5_4_c 𝔮 y ξ).1 hξ)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Same paragraph: `P`, which depends on `ξ` only, vanishes on `N(Y_0)` (Definition 5.4(d)). -/ theorem step_P_vanishes_on_conormal_cone (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (h𝔮 : 𝔮 ∈ (exclusionBaseIdeal W).minimalPrimes) (s : MvPolynomial (Fin m) k) (hs𝔮 : s ∉ 𝔮) (hs : ∀ h ∈ 𝔮, s * h ∈ exclusionBaseIdeal W) (P : MvPolynomial (Fin m) k) (hPW : ∀ q ∈ W, MvPolynomial.eval (fun i => q (.inr i)) P = 0) : P ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection 𝔮) := by intro ξ hξ obtain ⟨y, hy, hspan⟩ := (definition_5_4_d 𝔮 ξ).1 hξ have hyξ := step_conormal_subset W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs y hy ξ ((definition_5_4_c 𝔮 y ξ).2 hspan) simpa using hPW _ hyξ
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- `P` vanishes on the printed cone `N(Y_0) = ⋃_{y ∈ Y_sm} N*_y Y_0` over the Jacobian smooth locus (Definition 5.4). `step_P_vanishes_on_conormal_cone` argues over Mathlib's smooth points, which are exactly the Jacobian smooth points (`definition_5_4_b_iff`), so the two cones are equal (`conormalCone_paper_eq`). -/ theorem step_P_vanishes_on_N (𝔮 : Ideal (MvPolynomial (Fin m) k)) (hprime : 𝔮.IsPrime) (P : MvPolynomial (Fin m) k) (hP : P ∈ MvPolynomial.vanishingIdeal k (smoothConormalFibreProjection 𝔮)) : ∀ ξ ∈ conormalCone ⟨𝔮, hprime⟩, MvPolynomial.eval ξ P = 0 := by rw [conormalCone_paper_eq ⟨𝔮, hprime⟩] intro ξ hξ simpa using (MvPolynomial.mem_vanishingIdeal_iff.1 hP) ξ hξ
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- Final sentence: `Y_0` is nonempty, closed, irreducible (Definition 5.4(a), `I(Y_0) = 𝔮` prime) and avoids `H`, and `P` vanishes on `N(Y_0)`, so Theorem 5.7(i) gives `P(1, 0, …, 0) = 0`. -/ theorem step_theorem_5_7_applies (hm : 0 < m) (𝔮 : Ideal (MvPolynomial (Fin m) k)) (hprime : 𝔮.IsPrime) (havoid : ∀ y ∈ MvPolynomial.zeroLocus k 𝔮, y ⟨0, hm⟩ ≠ 0) (P : MvPolynomial (Fin m) k) (hP : ∀ ξ ∈ conormalCone ⟨𝔮, hprime⟩, MvPolynomial.eval ξ P = 0) : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P = 0 := by -- `Y_0 = V_K(𝔮)` is nonempty, closed and irreducible, with `I(Y_0) = 𝔮` obtain ⟨hne, hY, hpt⟩ := irreducibleClosed_zeroLocus_of_isPrime (k := k) 𝔮 refine theorem_5_7_i_printed hm _ hne hY ?_ P (hpt ▸ hP) ext y simp only [Set.mem_inter_iff, Set.mem_ofPred_eq, Set.mem_empty_iff_false, iff_false, not_and] exact fun hy h0 => havoid y hy h0
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- **Theorem 5.7(i).** Let `Y ⊆ X = K^n` be nonempty, closed and irreducible (Definition 5.4(a)) with `Y ∩ H = ∅`, `H = {x_1 = 0}`, and let `N(Y) = ⋃_{y ∈ Y_sm} N*_y Y` be the conormal cone over the smooth locus. Every polynomial `P ∈ K[ξ_1, …, ξ_n]` vanishing on `N(Y)` satisfies `P(1, 0, …, 0) = 0`. Derived from `theorem_5_7_i_prime` with `𝔭 = I(Y)`. -/ theorem theorem_5_7_i_printed {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ} (hm : 0 < m) (Y : Set (Fin m → k)) (hne : Y.Nonempty) (hY : IsIrreducibleClosed Y) (hH : Y ∩ {y | y ⟨0, hm⟩ = 0} = ∅) (P : MvPolynomial (Fin m) k) (hP : ∀ ξ ∈ conormalCone (idealPoint Y hne hY), MvPolynomial.eval ξ P = 0) : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P = 0 := by refine theorem_5_7_i_prime hm (idealPoint Y hne hY) (fun y hy h0 => ?_) P hP change y ∈ MvPolynomial.zeroLocus k (MvPolynomial.vanishingIdeal k Y) at hy rw [zeroLocus_vanishingIdeal_of_isZariskiClosed hY.1] at hy exact (Set.eq_empty_iff_forall_notMem.mp hH) y ⟨hy, h0⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- For a prime `𝔭`, `Y = V_K(𝔭)` is nonempty, closed and irreducible (Definition 5.4(a)), with `I(Y) = 𝔭` (Nullstellensatz). -/ theorem irreducibleClosed_zeroLocus_of_isPrime {k : Type u} [Field k] [IsAlgClosed k] {m : ℕ} (𝔭 : Ideal (MvPolynomial (Fin m) k)) [h𝔭 : 𝔭.IsPrime] : ∃ (hne : (MvPolynomial.zeroLocus k 𝔭).Nonempty) (hY : IsIrreducibleClosed (MvPolynomial.zeroLocus k 𝔭)), idealPoint (MvPolynomial.zeroLocus k 𝔭) hne hY = ⟨𝔭, h𝔭⟩ := by have hI : MvPolynomial.vanishingIdeal k (MvPolynomial.zeroLocus k 𝔭) = 𝔭 := MvPolynomial.IsPrime.vanishingIdeal_zeroLocus 𝔭 have hne : (MvPolynomial.zeroLocus k 𝔭).Nonempty := by rw [Set.nonempty_iff_ne_empty] intro h0 rw [h0, MvPolynomial.vanishingIdeal_empty] at hI exact h𝔭.ne_top hI.symm have hY : IsIrreducibleClosed (MvPolynomial.zeroLocus k 𝔭) := (definition_5_4_a hne ⟨𝔭, rfl⟩).2 (by rw [hI]; exact h𝔭) exact ⟨hne, hY, PrimeSpectrum.ext hI⟩
Variables and notation
Source ↗open MvPolynomialSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.GeometrySource ↗open Stafford38.Geometry.GenericPointKaehlerConormal (quotientDerivation quotientDerivation_mk mkDerivation_eq_sum_pderiv)Source ↗
universe uSource ↗
open Stafford38.Geometry.ProjectiveConormalDirections/-- `B = I(W) ∩ K[x]`. -/ abbrev exclusionBaseIdeal (W : Set (PhaseVar m → k)) : Ideal (MvPolynomial (Fin m) k) := (MvPolynomial.vanishingIdeal k W).comap (baseLift (k := k) (n := m)) omit [IsAlgClosed k] [CharZero k] in
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
Comparison notes
The Lean proof of theorem_5_8 follows the 17 printed steps of Subsection 5.3, one step declaration each, and closes by calling Theorem 5.7(i) as printed (theorem_5_7_i_printed, through step_theorem_5_7_applies) for Y₀ = V_K(𝔮), which is nonempty, closed and irreducible with I(Y₀) = 𝔮 by the Nullstellensatz and Definition 5.4(a) (irreducibleClosed_zeroLocus_of_isPrime). One extra declaration, step_P_vanishes_on_N, passes from the cone over Mathlib's smooth locus to the printed cone over the Jacobian smooth locus (Definition 5.4(b)). The steps, in order: ξ-homogeneous components (step_homogeneous_components), the zero section Y × {0} ⊆ W and Y = π(W), Y ∩ H = ∅, the Hamiltonian derivations H_f = {f, ·} and the polynomial φ(t) = g(y, t df_y), {y} × span{df_y} ⊆ W, the minimal primes of B and the component Y₀, the separator s, (y, dh_y) ∈ W on Y₀, {y} × N*_y Y₀ ⊆ W, P vanishes on N(Y₀), and Theorem 5.7(i). In Lean, B = I(W) ∩ K[x] is exclusionBaseIdeal W.
Review notes
6. Proof of Theorem 1.1
Proposition 6.1 — Char(Q) ∩ π⁻¹(H) = ∅
Lean ():
Section6.proposition_6_1
/-- **Proposition 6.1.** `Char(Q) ∩ V(x₁) = ∅`, where `V(x₁) ⊆ Spec k[x, ξ]` is the set of primes containing `x₁`. -/ theorem proposition_6_1 [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) : Section2.Char k (idealI k N dt) ∩ Vx₁ k n = ∅ := by rw [step_6_1_char_eq_support] exact step_6_1_support_avoids hN hform
More Lean proofs and definitions
/-- `V(x₁) ⊆ Spec k[x, ξ]`, the primes containing `x₁`. -/ abbrev Vx₁ (k : Type u) [Field k] (n : ℕ) : Set (CotangentBundle k (n + 1)) := V (Ideal.span {x₁Symbol k n})
/-- `M = gr Q` is finite over `k[x, ξ]`: the filtration `Q_m = qF_m` is good (Definition 2.4; Proposition 3.6). -/ theorem step_6_1_finite {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Module.Finite (SymbolRing k (n + 1)) (grQ k N dt) := grCyclic_finite_of_good k (idealI k N dt)
/-- Since `M` is finite, `Supp(M/x₁M) = Supp(M) ∩ V(x₁)` (Stacks 00L3). -/ theorem step_6_1_support_quotient {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Module.support (SymbolRing k (n + 1)) (QuotSMulTop (x₁Symbol k n) (grQ k N dt)) = Module.support (SymbolRing k (n + 1)) (grQ k N dt) ∩ Vx₁ k n := by have := step_6_1_finite (N := N) dt rw [Module.support_quotSMulTop, Vx₁, V, Ideal.span, PrimeSpectrum.zeroLocus_span]
/-- By Corollary 4.11, `M/x₁M = 0`, so `Supp(M) ∩ V(x₁) = ∅`. -/ theorem step_6_1_support_avoids [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) : Module.support (SymbolRing k (n + 1)) (grQ k N dt) ∩ Vx₁ k n = ∅ := by have hU := Stafford38.Paper.Section4.corollary_4_11_quotSMulTop (k := k) hN hform rw [← step_6_1_support_quotient] exact Module.support_eq_empty
/-- Since `M` is finitely generated, `Char(Q) = V(Ann M) = Supp(M)` (Stacks 00L2). -/ theorem step_6_1_char_eq_support {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Section2.Char k (idealI k N dt) = Module.support (SymbolRing k (n + 1)) (grQ k N dt) := by have := step_6_1_finite (N := N) dt rw [Section2.Char_eq, Module.support_eq_zeroLocus]
/-- Corollary 4.11 in the form `M/x₁M = 0` over `k[x,ξ]` (`QuotSMulTop`), as consumed by the support-avoidance argument of Section 6. -/ theorem corollary_4_11_quotSMulTop (hN : 0 < N) (hform : IsMonicNormalForm k N d) : Subsingleton (QuotSMulTop (symbolCoordinate k n) (GrQ k n N d)) := by have hsurj := corollary_4_11 hN hform change Subsingleton (GrQ k n N d ⧸ symbolCoordinate k n • (⊤ : Submodule (Symbol k n) _)) rw [Submodule.Quotient.subsingleton_iff, eq_top_iff] intro z _ obtain ⟨w, rfl⟩ := hsurj z exact Submodule.smul_mem_pointwise_smul w _ ⊤ trivial
Variables and notation
Source ↗open scoped PointwiseSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section3 (IsMonicNormalForm isPBWMonicAt_of_isMonicNormalForm)Source ↗
variable {k : Type*} [Field k] [CharZero k] [Algebra ℚ k] {n N : ℕ} {d : PresentedWeyl k (n + 1)}
/-- `gr Q` is finitely generated over `k[x, ξ]` because `Q_m = qF_m` is good. -/ theorem grCyclic_finite_of_good (I : RightIdeal (PresentedWeyl k n)) : Module.Finite (SymbolRing k n) (grCyclic k I) := have := (qFilt_isGoodFiltration k I).finite Module.Finite.equiv (grCyclicEquivS k I).symm
/-- `Char(A_n/I) = V(Ann_{k[x,ξ]} gr Q)` with `gr Q` the repository's `grCyclic`. -/ theorem Char_eq (I : RightIdeal (PresentedWeyl k n)) : Char k I = V (Module.annihilator (SymbolRing k n) (grCyclic k I)) := by rw [Char, CharOf, annihilator_qFilt]
Comparison notes
proposition_6_1: for d̃ in the form (3.1) with N > 0, Char(Q) ∩ V(x₁) = ∅ in Spec k[x, ξ], V(x₁) the primes containing x₁. The proof follows the printed sentences: M = gr Q is finite (step_6_1_finite: Q_m = qF_m is good, grCyclic_finite_of_good), Supp(M/x₁M) = Supp(M) ∩ V(x₁) (step_6_1_support_quotient, Stacks 00L3), M/x₁M = 0 by Corollary 4.11 (step_6_1_support_avoids) and Char(Q) = Supp(M) (step_6_1_char_eq_support with Char_eq, Stacks 00L2).
Review notes
Lemma 6.2 — Base change of the canonical quotient and its characteristic variety
/-- **Lemma 6.2.** Let `K/k` be a field extension, `𝓘_K = d̃A_n(K) + x₁^N d̃A_n(K)` and `Q_K = Q ⊗_k K = A_n(K)/𝓘_K` with the order filtration `Q_{K,i} = qF_iA_n(K)`. Then `gr Q_K ≅ gr Q ⊗_k K` (as `K`-modules; Lean writes the factor `K` on the left, `K ⊗_k gr Q`), `Ann_{K[x,ξ]} gr Q_K = (Ann_{k[x,ξ]} gr Q) K[x,ξ]`, and `Q = 0 ⇔ Q_K = 0`. In particular, `Char(Q_K)` is the preimage of `Char(Q)` under `Spec K[x,ξ] → Spec k[x,ξ]`. -/ theorem lemma_6_2 {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Nonempty (grQ K N (dtK K dt) ≃ₗ[K] K ⊗[k] grQ k N dt) ∧ Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt)) = (Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt)).map (σ k K n) ∧ (Subsingleton (Q k N dt) ↔ Subsingleton (Q K N (dtK K dt))) ∧ Section2.Char K (idealI K N (dtK K dt)) = PrimeSpectrum.comap (σ k K n) ⁻¹' Section2.Char k (idealI k N dt) := ⟨⟨(step_6_2_graded (K := K) dt).symm⟩, step_6_2_annihilator dt, step_6_2_vanishing dt, step_6_2_char dt⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
More Lean proofs and definitions
/-- Proof, first sentence: by the PBW basis, `𝓘_K = 𝓘 ⊗_k K` inside `A_n(K) = A_n(k) ⊗_k K`. -/ theorem step_6_2_ideal {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Submodule.map (pbwTensorEquiv (k := k) (K := K) (n + 1)).toLinearMap ((rightIdealKSubmodule k (idealI k N dt)).baseChange K) = rightIdealKSubmodule K (idealI K N (dtK K dt)) := pbwTensorEquiv_map_canonicalRightIdeal n N dt
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]
/-- Consequently `Q ⊗_k K ≅ Q_K`: tensoring `0 → 𝓘 → A_n(k) → Q → 0` with `K` and identifying the image of `𝓘 ⊗_k K` with `𝓘_K` by `step_6_2_ideal`. -/ def step_6_2_quotient {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Q k N dt ⊗[k] K ≃ₗ[k] Q K N (dtK K dt) := by let J := rightIdealKSubmodule k (idealI k N dt) let eK := pbwTensorEquiv (k := k) (K := K) (n + 1) let JK := rightIdealKSubmodule K (idealI K N (dtK K dt)) let f := (TensorProduct.comm k (PresentedWeyl k (n + 1)) K).trans (LinearEquiv.restrictScalars k eK) let D := LinearMap.range (TensorProduct.map J.subtype (LinearMap.id : K →ₗ[k] K)) have hden : Submodule.map f.toLinearMap D = JK.restrictScalars k := by have hf : f.toLinearMap = (LinearEquiv.restrictScalars k eK).toLinearMap.comp (TensorProduct.comm k (PresentedWeyl k (n + 1)) K).toLinearMap := by ext z rfl rw [hf, Submodule.map_comp, Stafford38.PaperQuotientDescent.comm_map_tensorSubmodule_range] calc Submodule.map (LinearEquiv.restrictScalars k eK).toLinearMap (Submodule.restrictScalars k (J.baseChange K)) = Submodule.restrictScalars k (Submodule.map eK.toLinearMap (J.baseChange K)) := (Submodule.restrictScalars_map eK.toLinearMap (J.baseChange K)).symm _ = Submodule.restrictScalars k JK := by rw [step_6_2_ideal] exact (TensorProduct.congr (filteredRightQuotientEquivRightQuotient k (idealI k N dt)).symm (LinearEquiv.refl k K)).trans (((TensorProduct.quotientTensorEquiv K J).trans (Submodule.Quotient.equiv D (JK.restrictScalars k) f hden)).trans ((Submodule.Quotient.restrictScalarsEquiv k JK).trans (LinearEquiv.restrictScalars k (filteredRightQuotientEquivRightQuotient K (idealI K N (dtK K dt))))))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]
/-- **Lemma 6.2, proof, first sentence.** By the PBW basis `F_iA_n(K) = F_iA_n(k) ⊗_k K` and `𝓘_K = 𝓘 ⊗_k K`, so `Q_{K,i} = Q_i ⊗_k K`: `θ` maps `K ⊗_k Q_i` onto `Q_{K,i} = qF_iA_n(K)`. -/ theorem step_6_2_pieces (dt : PresentedWeyl k (n + 1)) (i : ℕ) : ((Qpiece k N dt i).baseChange K).map (theta62 (N := N) dt).toLinearMap = Qpiece K N (dtK K dt) i := by set ι := presentedWeylScalarExtension (k := k) (K := K) (n + 1) set F := orderPiece k (n + 1) i have hF : orderPiece K (n + 1) i = Submodule.span K (ι '' F) := by rw [← pbwTensorEquiv_map_orderPiece (k := k) (K := K), Submodule.baseChange_eq_span, Submodule.map_span, Submodule.map_coe, ← Set.image_comp] congr 1 ext z simp only [Set.mem_image, Function.comp_apply, TensorProduct.mk_apply, LinearEquiv.coe_coe, pbwTensorEquiv_tmul_one, SetLike.mem_coe] rfl have hR : Qpiece K N (dtK K dt) i = Submodule.span K ((fun a => (Submodule.Quotient.mk (ι a) : Q K N (dtK K dt))) '' F) := by change (orderPiece K (n + 1) i).map (orbitMap (qClass K (idealI K N (dtK K dt)))) = _ rw [hF, Submodule.map_span, ← Set.image_comp] congr 1 ext x simp only [Set.mem_image, Function.comp_apply, orbitMap_qClass] have hL : Qpiece k N dt i = Submodule.span k ((fun a => (Submodule.Quotient.mk a : Q k N dt)) '' F) := by calc Qpiece k N dt i = F.map (orbitMap (qClass k (idealI k N dt))) := rfl _ = (Submodule.span k (F : Set (PresentedWeyl k (n + 1)))).map (orbitMap (qClass k (idealI k N dt))) := by rw [Submodule.span_eq] _ = _ := by rw [Submodule.map_span] congr 1 ext x simp only [Set.mem_image, orbitMap_qClass] rw [hR, hL, Submodule.baseChange_span, Submodule.map_span, ← Set.image_comp, ← Set.image_comp] congr 1 ext x simp [theta62_tmul, ι]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- **Lemma 6.2, proof, second sentence.** As `K` is flat over `k`, `Q_{K,i}/Q_{K,i-1} = (Q_i/Q_{i-1}) ⊗_k K`. -/ def step_6_2_graded_pieces (dt : PresentedWeyl k (n + 1)) (i : ℕ) : K ⊗[k] grPiece (Qpiece k N dt) i ≃ₗ[K] grPiece (Qpiece K N (dtK K dt)) i := tensorPieceEquiv (theta62 (N := N) dt) (Qpiece k N dt i) (lowerPiece (Qpiece k N dt) i) (by cases i with | zero => exact bot_le | succ i => exact (cyclicFiltration_order_isCompatible k _).mono (Nat.le_succ i)) _ _ (step_6_2_pieces dt i) (lowerPiece_Qpiece_baseChange dt i)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- Same sentence: summing over `i`, `gr Q_K = gr Q ⊗_k K` (Definition 2.3, `gr`). -/ def step_6_2_sum (dt : PresentedWeyl k (n + 1)) : K ⊗[k] gr (Qpiece k N dt) ≃ₗ[K] gr (Qpiece K N (dtK K dt)) := (TensorProduct.directSumRight k K K (fun i => grPiece (Qpiece k N dt) i)).trans (DFinsupp.mapRange.linearEquiv fun i => step_6_2_graded_pieces dt i)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- `gr Q_K ≅ K ⊗_k gr Q` (as `K`-modules), assembled from the pieces through `grCyclicEquiv`. -/ def step_6_2_graded (dt : PresentedWeyl k (n + 1)) : K ⊗[k] grQ k N dt ≃ₗ[K] grQ K N (dtK K dt) := (LinearEquiv.baseChange k K _ _ (grCyclicEquiv k (idealI k N dt))).trans ((step_6_2_sum dt).trans (grCyclicEquiv K (idealI K N (dtK K dt))).symm)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- **Lemma 6.2, proof, third sentence.** `gr Q` is finitely generated (cyclic, generated by `q̄`) over `k[x, ξ]`, and `k[x, ξ] → K[x, ξ] = K ⊗_k k[x, ξ]` is flat, so the annihilator commutes with the base change (Stacks 07T8): through `gr Q_K ≅ K ⊗_k gr Q`, `Ann_{K[x,ξ]} gr Q_K = (Ann_{k[x,ξ]} gr Q) K[x,ξ]`. (For the cyclic module the 07T8 argument is spelled out: write `g ∈ K[x, ξ]` as `∑ b_λ σ(f_λ)` for a `k`-basis `b` of `K`.) -/ theorem step_6_2_annihilator (dt : PresentedWeyl k (n + 1)) : Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt)) = (Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt)).map (σ k K n) := by classical let I := idealI k N dt let IK := idealI K N (dtK K dt) apply le_antisymm · intro g hg let b := Module.Basis.ofVectorSpace k K let t := (MvPolynomial.algebraTensorAlgEquiv k K).symm g let c := TensorProduct.equivFinsuppOfBasisLeft b t have ht : t = c.sum fun l f => b l ⊗ₜ[k] f := by rw [← (TensorProduct.equivFinsuppOfBasisLeft b).symm_apply_apply t] change (TensorProduct.equivFinsuppOfBasisLeft b).symm.toLinearMap c = _ rw [TensorProduct.equivFinsuppOfBasisLeft_symm] simp [Finsupp.lsum_apply] rfl have hg' : g = c.sum fun l f => b l • σ k K n f := by rw [← (MvPolynomial.algebraTensorAlgEquiv k K).apply_symm_apply g] change MvPolynomial.algebraTensorAlgEquiv k K t = _ rw [ht, Finsupp.sum, map_sum] simp [Finsupp.sum, MvPolynomial.algebraTensorAlgEquiv_tmul] have hzero : step_6_2_graded (K := K) (N := N) dt (c.sum fun l f => b l ⊗ₜ[k] (f • qBar k I)) = 0 := by rw [Finsupp.sum, map_sum, Finset.sum_congr rfl fun l _ => smul_qBar_K_eq (N := N) dt (b l) (c l), ← Finset.sum_smul] have := Module.mem_annihilator.mp hg (qBar K IK) rw [hg', Finsupp.sum] at this exact this rw [← map_zero (step_6_2_graded (K := K) (N := N) dt)] at hzero have hc := (step_6_2_graded (K := K) (N := N) dt).injective hzero have hmap : (c.mapRange (fun f => f • qBar k I) (zero_smul _ _)) = 0 := by apply TensorProduct.sum_tmul_basis_left_eq_zero b rw [Finsupp.sum_mapRange_index (fun _ => TensorProduct.tmul_zero _ _)] exact hc have hann : ∀ l, c l ∈ Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt) := by intro l have hl := congrArg (· l) hmap simp only [Finsupp.mapRange_apply, Finsupp.coe_zero, Pi.zero_apply] at hl rw [Module.mem_annihilator] intro m obtain ⟨f, rfl⟩ := grCyclic_generated k I m rw [smul_comm, hl, smul_zero] rw [hg', Finsupp.sum] refine Submodule.sum_mem _ fun l _ => ?_ rw [Algebra.smul_def] exact Ideal.mul_mem_left _ _ (Ideal.mem_map_of_mem _ (hann l)) · rw [Ideal.map_le_iff_le_comap] intro f hf rw [Ideal.mem_comap, Module.mem_annihilator] intro y obtain ⟨t, rfl⟩ := (step_6_2_graded (K := K) (N := N) dt).surjective y induction t using TensorProduct.induction_on with | zero => rw [map_zero, smul_zero] | tmul c m => obtain ⟨h, rfl⟩ := grCyclic_generated k I m rw [smul_qBar_K_eq, smul_smul, mul_comm, ← smul_smul, ← step_6_2_graded_one_tmul_smul, Module.mem_annihilator.mp hf, TensorProduct.tmul_zero, map_zero, smul_zero] | add x y hx hy => rw [map_add, smul_add, hx, hy, add_zero]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- `Char(Q_K)` is the preimage of `Char(Q)` under `Spec K[x,ξ] → Spec k[x,ξ]`. -/ theorem step_6_2_char {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Section2.Char K (idealI K N (dtK K dt)) = PrimeSpectrum.comap (σ k K n) ⁻¹' Section2.Char k (idealI k N dt) := by rw [Section2.Char_eq, Section2.Char_eq, V, V, PrimeSpectrum.preimage_comap_zeroLocus, step_6_2_annihilator, Ideal.map, PrimeSpectrum.zeroLocus_span]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- `Q = 0 ⇔ Q_K = 0`: `Q ⊗_k K = 0` implies `Q = 0` because `K` is a nonzero free (so faithfully flat) `k`-module. -/ theorem step_6_2_vanishing {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Subsingleton (Q k N dt) ↔ Subsingleton (Q K N (dtK K dt)) := by constructor · intro h have : Subsingleton (Q k N dt ⊗[k] K) := inferInstance exact (step_6_2_quotient (K := K) dt).toEquiv.symm.subsingleton · intro h have : Subsingleton (Q k N dt ⊗[k] K) := (step_6_2_quotient (K := K) dt).toEquiv.subsingleton exact Module.FaithfullyFlat.rTensor_reflects_triviality k K (Q k N dt)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
Comparison notes
lemma_6_2 states the four conclusions for a field extension K/k: gr Q_K ≅ K ⊗_k gr Q as K-modules, Ann_{K[x,ξ]} gr Q_K is the extension of Ann_{k[x,ξ]} gr Q along k[x, ξ] → K[x, ξ], Q = 0 ⇔ Q_K = 0, and Char(Q_K) is the preimage of Char(Q). The proof follows the printed sentences: 𝓘_K = 𝓘 ⊗_k K by the PBW basis (step_6_2_ideal, step_6_2_quotient), Q_{K,i} = Q_i ⊗_k K (step_6_2_pieces), the graded pieces by flatness (step_6_2_graded_pieces), the sum over i (step_6_2_sum, step_6_2_graded), the annihilator by flat base change (step_6_2_annihilator, Stacks 07T8), the preimage of Char (step_6_2_char) and Q ⊗_k K = 0 ⇒ Q = 0 (step_6_2_vanishing).
Review notes
Proof §6 — Proof of Theorem 1.1: Gabber, exclusion, descent, transport
Let be nonzero of Bernstein degree and let be its monic normal form. If , then and every nonzero is a unit anyways, so let . As in Section 4 we write
Let , and let
- is Zariski closed, since it is the zero set of an ideal of polynomials.
- The module is graded by the order filtration, which assigns degree to the and degree to the ; so is homogeneous in . Hence if and , then for every , where denotes the -homogeneous component of degree of . So , i.e. is fibre-conical.
- By Gabber's theorem (Theorem 2.10) is involutive.
- By eq. (3.5), , hence by Lemma 6.2, so vanishes on .
- By eq. (3.4), .
- By Lemma 6.2, is the preimage of under , and is the preimage of . Since preimages commute with intersections, Proposition 6.1 gives . By Theorem 2.11(a) and Definition 2.5, a closed point lies in if and only if , i.e. the closed points of form . Hence the closed points of are , and so .
We apply the inverse of the automorphism of Remark 3.3:
Lean: Paper.theorem_1_1,
Paper.theorem_1_1_fixedSource,
Section6.theorem_1_1_paper,
Section6.theorem_6_Q_eq_zero
Lean ():
Section6.lemma_6_2
Lastly, as Stafford already showed, from follows cyclicity of the torsion right module:
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
/-- **Theorem 1.1, moreover.** If `n ≥ 1`, one may choose `F = ℓ^{deg_B d}` for a linear symplectic coordinate `ℓ = Φ_g^{-1}(x₁)`. -/ theorem theorem_1_1_fixedSource : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u} := Section6.theorem_1_1_fixedSource_paper
/-- **Proof of Theorem 1.1.** `n = 0` is `step_rank_zero`. For `n ≥ 1`, Section 3 (`section_3_reduction`) either finds `deg_B d = 0` and a unit, or the normal form `d̃ = c⁻¹Φ_g(d)` with `N = deg_B d ≥ 1`, for which `Q = 0` (`theorem_6_Q_eq_zero`) gives `1 = d̃R + x₁^N d̃S` and, transported by `Φ_g^{-1}` (Remark 3.3), `1 = d(c⁻¹Φ_g^{-1}(R)) + F d(c⁻¹Φ_g^{-1}(S))` with `F = Φ_g^{-1}(x₁)^N`. -/ theorem theorem_1_1_paper : Stafford38Challenge.UniversalStatement.{u} := by intro k _ _ n d hd cases n with | zero => exact step_rank_zero d hd | succ n => rcases section_3_reduction k hd with ⟨-, h⟩ | ⟨hN, g, c, -, hform, hdeg, himp⟩ · exact h · obtain ⟨ell, R, S, -, h⟩ := himp (theorem_6_Q_eq_zero hN hform hdeg) exact ⟨ell ^ bernsteinDegree k d, R, S, h⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
More Lean proofs and definitions
/-- **Theorem 1.1, "moreover".** For `n ≥ 1` one may take `F = Φ_g^{-1}(x₁)^{deg_B d}` with a linear `ω`-preserving `g`; for `deg_B d = 0` any linear coordinate works, since then `F = 1`. -/ theorem theorem_1_1_fixedSource_paper : Stafford38FixedSourceChallenge.UniversalFixedSourceStatement.{u} := by intro k _ _ n d hd rw [← bernsteinDegree_eq] rcases section_3_reduction k hd with ⟨hN0, -⟩ | ⟨hN, g, c, -, hform, hdeg, himp⟩ · obtain ⟨c, -, -, hcert⟩ := step_scalar_case k hd hN0 refine ⟨_, algebraMap k _ c⁻¹, 0, phiG_symm_x₁_isLinearWeylCoordinate k 1, ?_⟩ rw [hN0, pow_zero] exact hcert · exact himp (theorem_6_Q_eq_zero hN hform hdeg)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- `n = 0`: `A_0(k) = k`, and every nonzero `d` is a unit. -/ theorem step_rank_zero (d : PresentedWeyl k 0) (hd : d ≠ 0) : ∃ F R S : PresentedWeyl k 0, (1 : PresentedWeyl k 0) = d * R + F * d * S := by let e := presentedIteratedEquiv k 0 let c : k := e d have hc : c ≠ 0 := by intro hczero change e d = 0 at hczero apply hd calc d = e.symm (e d) := (e.symm_apply_apply d).symm _ = e.symm 0 := congrArg e.symm hczero _ = 0 := map_zero e.symm refine ⟨0, e.symm (c⁻¹ : k), 0, ?_⟩ have hmul : c * c⁻¹ = (1 : k) := mul_inv_cancel₀ hc have h : d * e.symm (c⁻¹ : k) = 1 := by calc d * e.symm (c⁻¹ : k) = e.symm c * e.symm (c⁻¹ : k) := by congr 1 exact (e.symm_apply_apply d).symm _ = e.symm (c * c⁻¹) := (map_mul e.symm _ _).symm _ = e.symm 1 := congrArg e.symm hmul _ = 1 := map_one e.symm simpa using h.symm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Section 3 for the final assembly.** For `0 ≠ d ∈ A_n(k)`, either `deg_B d = 0` and `d` is a unit with `1 = dR + FdS` (opening paragraph), or `N = deg_B d ≥ 1` and Proposition 3.2 gives the normal form `d̃ = c⁻¹Φ_g(d)` of (3.1) with `deg_B d̃ = N`, for which `Q = A/𝓘 = 0` yields `1 = dR + ℓ^N dS` with a linear symplectic coordinate `ℓ` (Remark 3.3). -/ theorem section_3_reduction [CharZero k] {n : ℕ} {d : PresentedWeyl k (n + 1)} (hd : d ≠ 0) : (bernsteinDegree k d = 0 ∧ Stafford.Reduction.Stafford38 d) ∨ (1 ≤ bernsteinDegree k d ∧ ∃ (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k), c ≠ 0 ∧ IsMonicNormalForm k (bernsteinDegree k d) (c⁻¹ • phiG k g d) ∧ bernsteinDegree k (c⁻¹ • phiG k g d) = bernsteinDegree k d ∧ (Subsingleton (Q k (bernsteinDegree k d) (c⁻¹ • phiG k g d)) → ∃ ell R S : PresentedWeyl k (n + 1), Stafford38FixedSourceChallenge.IsLinearWeylCoordinate k n ell ∧ (1 : PresentedWeyl k (n + 1)) = d * R + ell ^ bernsteinDegree k d * d * S)) := by by_cases hN : bernsteinDegree k d = 0 · obtain ⟨c, -, -, hcert⟩ := step_scalar_case k hd hN exact Or.inl ⟨hN, 1, _, 0, hcert⟩ · obtain ⟨g, c, hc, hform, hdeg⟩ := proposition_3_2 k hd (Nat.one_le_iff_ne_zero.mpr hN) exact Or.inr ⟨Nat.one_le_iff_ne_zero.mpr hN, g, c, hc, hform, hdeg, remark_3_3_of_Q_eq_zero k g c d⟩
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Body of the proof of Theorem 1.1.** For `d̃` in the normal form (3.1) with `N = deg_B d̃ ≥ 1`, `Q = A_n(k)/(d̃A_n(k) + x₁^N d̃A_n(k)) = 0`. With `K = k̄`: `W = ∅` by Theorem 5.8, so `gr Q_K = 0`, `Q_K = 0`, and `Q = 0` by Lemma 6.2. -/ theorem theorem_6_Q_eq_zero [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : Subsingleton (Q k N dt) := by let K := AlgebraicClosure k have : CharZero K := charZero_of_injective_algebraMap (algebraMap k K).injective have hW := step_W_empty (K := K) hN hform hdeg exact ((lemma_6_2 (K := K) (N := N) dt).2.2.1).2 (step_QK_zero (step_grQK_zero hW))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- `W = V_K(Ann_{K[x,ξ]} gr Q_K) ⊆ K^n × (K^n)^*`. -/ abbrev W (K : Type v) [Field K] [Algebra k K] {n : ℕ} (N : ℕ) (dt : PresentedWeyl k (n + 1)) : Set (PhaseVar (n + 1) → K) := VK K (Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt)))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]
/-- `W` is the set of closed points of `Char(Q_K)` (Theorem 2.12(a)), with vanishing ideal `I(W) = √(Ann gr Q_K)` (Theorem 2.12(c)). -/ theorem step_W_closed_points [IsAlgClosed K] {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : (∀ q, q ∈ W K N dt ↔ closedPoint K q ∈ Section2.Char K (idealI K N (dtK K dt))) ∧ vanishingIdealOf K (W K N dt) = (Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt))).radical := ⟨fun q => by rw [Section2.Char_eq]; exact (theorem_2_12 K _).1.2.2 q, (theorem_2_12 K _).2.2⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]
/-- **Bullet 1.** `W` is Zariski closed, the zero set of an ideal. -/ theorem step_W_closed {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : ∃ L : Ideal (SymbolRing K (n + 1)), W K N dt = MvPolynomial.zeroLocus K L := ⟨_, VK_eq_zeroLocus K _⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]
/-- "The module `gr Q_K` is graded by the order filtration, which assigns degree `0` to the `x_i` and degree `1` to the `ξ_i`; so `Ann gr Q_K` is homogeneous in `ξ`": a homogeneous `g_d` of degree `d` maps the degree-`0` generator `q̄` into degree `d`, so `g q̄ = 0` forces `g_d q̄ = 0` for every component, and `gr Q_K` is generated by `q̄`. -/ theorem step_annihilator_homogeneous {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : (Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt))).IsHomogeneous (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight (n + 1))) := by classical let I := idealI K N (dtK K dt) let 𝒜 := MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight (n + 1)) let x0 := orderPieceToQuotientGraded K I 0 (orderPieceOne (n := n + 1) K) have hq : qBar K I = orderAssociatedGradedOf K I 0 x0 := rfl -- `π_e`, the degree-`e` component of `gr Q_K = ⊕_e Q_{K,e}/Q_{K,e-1}`. let π : ∀ e : ℕ, OrderAssociatedGradedModule K I →+ QuotientOrderGradedPiece K I e := fun e => { toFun := fun z => (show Stafford38.CharacteristicFilteredQuotientGraded.QuotientOrderAssociatedGraded K I from z) e map_zero' := rfl map_add' := fun _ _ => rfl } have hπ : ∀ e j (x : QuotientOrderGradedPiece K I j), π e (orderAssociatedGradedOf K I j x) = if h : j = e then h ▸ x else 0 := by intro e j x change (DirectSum.of _ j x : DirectSum ℕ _) e = _ rw [DirectSum.of_apply] -- The degree-`(0 + d)` component of `g q̄` is `g_d q̄`. have hcomp : ∀ (g : SymbolRing K (n + 1)) (d : ℕ), ((DirectSum.decompose 𝒜 g d : SymbolRing K (n + 1))) • qBar K I = orderAssociatedGradedOf K I (0 + d) (π (0 + d) (g • qBar K I)) := by intro g d rw [hq, smul_orderAssociatedGradedOf_eq_of_homogeneousAction] congr 1 conv_rhs => rw [← DirectSum.sum_support_decompose 𝒜 g] rw [Finset.sum_smul, map_sum] simp_rw [smul_orderAssociatedGradedOf_eq_of_homogeneousAction, hπ] by_cases hd : d ∈ (DirectSum.decompose 𝒜 g).support · rw [Finset.sum_eq_single d] · simp · intro i _ hid exact dite_eq_right_iff.mpr fun h => absurd h (by omega) · exact fun h => absurd hd h · have hact : CharacteristicFilteredQuotientGraded.quotientOrderHomogeneousAction K I (0 : 𝒜 d) x0 = 0 := by have h0 := smul_orderAssociatedGradedOf_eq_of_homogeneousAction K I (0 : 𝒜 d) x0 rw [Submodule.coe_zero, zero_smul] at h0 exact DirectSum.of_injective (β := QuotientOrderGradedPiece K I) (0 + d) (h0.symm.trans (map_zero (DirectSum.of (QuotientOrderGradedPiece K I) (0 + d))).symm) rw [DFinsupp.notMem_support_iff.mp hd, hact, Finset.sum_eq_zero] intro i hi exact dite_eq_right_iff.mpr fun h => absurd (show i = d by omega ▸ hi) hd intro d g hg rw [Module.mem_annihilator] at hg ⊢ intro m obtain ⟨h, rfl⟩ := grCyclic_generated K I m rw [smul_comm, hcomp, hg, map_zero] have h0 : orderAssociatedGradedOf K I (0 + d) 0 = (0 : OrderAssociatedGradedModule K I) := map_zero _ rw [h0] exact smul_zero h
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 2.** `W` is fibre-conical: for `g ∈ Ann gr Q_K` and `(y, ξ) ∈ W`, `g(y, tξ) = ∑_d t^d g_d(y, ξ) = 0`, as every `ξ`-homogeneous component `g_d` lies in `Ann gr Q_K`. -/ theorem step_W_fibreConical [IsAlgClosed K] [CharZero K] {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : IsFibreConical (W K N dt) := by have hhom := step_annihilator_homogeneous (K := K) (N := N) dt rw [W, VK_eq_zeroLocus] intro q hq a _ let y : Fin (n + 1) → K := fun i => q (.inl i) let ξ : Fin (n + 1) → K := fun i => q (.inr i) have hqsplit : q = Sum.elim y ξ := by funext i rcases i with i | i <;> rfl have hline (f : SymbolRing K (n + 1)) (hf : f ∈ Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt))) : fibreLinePolynomial y ξ f = 0 := by apply Polynomial.ext intro d rw [coeff_fibreLinePolynomial_eq_eval_weightedHomogeneousComponent, ← hqsplit] exact (MvPolynomial.mem_zeroLocus_iff.mp hq) _ (hhom d hf) intro f hf have heval := eval_fibreLinePolynomial y ξ f a rw [hline f hf, Polynomial.eval_zero] at heval simpa [y, ξ] using heval.symm
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 3.** `I(W)` is involutive, by Gabber's theorem (Theorem 2.11) and `I(W) = √(Ann gr Q_K)`. -/ theorem step_W_involutive [IsAlgClosed K] [CharZero K] {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : IsInvolutiveSet K (W K N dt) := by rw [IsInvolutiveSet, (step_W_closed_points (K := K) (N := N) dt).2] exact theorem_2_11_cyclic K (n + 1) (idealI K N (dtK K dt))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 4.** By (3.5) (Proposition 3.6), `P ∈ Ann_{k[x,ξ]} gr Q`, hence `P ∈ Ann_{K[x,ξ]} gr Q_K` by Lemma 6.2, so `P` vanishes on `W`. Here `σ_N^ord(d̃) = P(ξ)` with `P ∈ k[ξ]` (Lemma 3.5), regarded in `K[ξ]`. -/ theorem step_W_P_vanishes {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) {P : MvPolynomial (Fin (n + 1)) k} (hP : orderSymbol k N dt = MvPolynomial.rename Sum.inr P) : ∀ q ∈ W K N dt, MvPolynomial.eval (fun i => q (.inr i)) (MvPolynomial.map (algebraMap k K) P) = 0 := by intro q hq have hAnn := proposition_3_6 k hform hdeg have hAnnK : σ k K n (orderSymbol k N dt) ∈ Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt)) := by rw [(lemma_6_2 (K := K) (N := N) dt).2.1] exact Ideal.mem_map_of_mem _ hAnn have h := hq _ hAnnK rw [hP, σ, AlgHom.toRingHom_eq_coe, AlgHom.coe_toRingHom, symbolScalarExtension_apply, MvPolynomial.map_rename, MvPolynomial.eval_rename] at h exact h
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 5.** By (3.4) (Lemma 3.5), `P(1, 0, …, 0) = 1 ≠ 0`. -/ theorem step_W_P_axis {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) {P : MvPolynomial (Fin (n + 1)) k} (hP : orderSymbol k N dt = MvPolynomial.rename Sum.inr P) : MvPolynomial.eval (fun i : Fin (n + 1) => if i = ⟨0, Nat.succ_pos n⟩ then (1 : K) else 0) (MvPolynomial.map (algebraMap k K) P) = 1 := by have h := lemma_3_5_eval_axis k hform hdeg rw [hP, MvPolynomial.eval_rename] at h have hpt : (fun i : Fin (n + 1) => if i = ⟨0, Nat.succ_pos n⟩ then (1 : K) else 0) = algebraMap k K ∘ (axisPoint k (.inr (0 : Fin (n + 1))) ∘ Sum.inr) := by funext i by_cases hi : i = 0 <;> simp [axisPoint, hi] have hcomp := MvPolynomial.eval₂_comp_left (algebraMap k K) (RingHom.id k) (axisPoint k (.inr (0 : Fin (n + 1))) ∘ Sum.inr) P rw [RingHom.comp_id] at hcomp rw [hpt, MvPolynomial.eval_map, ← hcomp] change algebraMap k K (MvPolynomial.eval _ P) = 1 rw [h, map_one]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- `V(x₁) ⊆ Spec K[x,ξ]` is the preimage of `V(x₁) ⊆ Spec k[x,ξ]`. -/ theorem step_Vx₁_preimage {n : ℕ} : Vx₁ K n = PrimeSpectrum.comap (σ k K n) ⁻¹' Vx₁ k n := by ext p simp only [Set.mem_preimage, Vx₁, mem_V_iff, Ideal.span_singleton_le_iff_mem, PrimeSpectrum.comap_asIdeal, Ideal.mem_comap] change _ ↔ MvPolynomial.map (algebraMap k K) (MvPolynomial.X _) ∈ p.asIdeal rw [MvPolynomial.map_X]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 6, first part.** `Char(Q_K)` and `V(x₁)` are the preimages of `Char(Q)` and `V(x₁)` under `Spec K[x,ξ] → Spec k[x,ξ]` (Lemma 6.2), and preimages commute with intersections, so Proposition 6.1 gives `Char(Q_K) ∩ V(x₁) = ∅`. -/ theorem step_char_K_avoids [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) : Section2.Char K (idealI K N (dtK K dt)) ∩ Vx₁ K n = ∅ := by rw [(lemma_6_2 (K := K) (N := N) dt).2.2.2, step_Vx₁_preimage (k := k), ← Set.preimage_inter, proposition_6_1 hN hform, Set.preimage_empty]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 6, second part.** By Definition 2.6 a closed point `𝔪_{(y,ξ)}` lies in `V(x₁)` iff `y₁ = 0`, so the closed points of `V(x₁)` form `π⁻¹(H) = H × (K^n)^*`. -/ theorem step_closedPoint_mem_Vx₁ {n : ℕ} (q : PhaseVar (n + 1) → K) : closedPoint K q ∈ Vx₁ K n ↔ q (.inl (0 : Fin (n + 1))) = 0 := by rw [Vx₁, mem_V_iff, Ideal.span_singleton_le_iff_mem] change x₁Symbol K n ∈ maxIdealAt K q ↔ _ rw [mem_maxIdealAt_iff, MvPolynomial.eval_X]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Bullet 6.** The closed points of `Char(Q_K) ∩ V(x₁)` are `W ∩ π⁻¹(H)`, hence `W ∩ π⁻¹(H) = ∅` (`π = proj`, Definition 2.5). -/ theorem step_W_avoids_H [IsAlgClosed K] [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) : W K N dt ∩ proj ⁻¹' {y : Fin (n + 1) → K | y ⟨0, Nat.succ_pos n⟩ = 0} = ∅ := by ext q simp only [Set.mem_inter_iff, Set.mem_preimage, Set.mem_ofPred_eq, Set.mem_empty_iff_false, iff_false] rintro ⟨hqW, hq1⟩ have hmem : closedPoint K q ∈ Section2.Char K (idealI K N (dtK K dt)) ∩ Vx₁ K n := ⟨((step_W_closed_points (K := K) (N := N) dt).1 q).1 hqW, (step_closedPoint_mem_Vx₁ q).2 hq1⟩ rw [step_char_K_avoids hN hform] at hmem exact hmem
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Theorem 5.8 applies:** `W = ∅`. -/ theorem step_W_empty [IsAlgClosed K] [CharZero K] [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : W K N dt = ∅ := by obtain ⟨-, P, hP, -⟩ := lemma_3_5 k hform hdeg have hinv : Stafford38.Characteristic.PostScalarExtensionPoisson.IsInvolutive (MvPolynomial.vanishingIdeal K (W K N dt)) := (isInvolutive_iff K _).1 (step_W_involutive (K := K) (N := N) dt) exact Stafford38.Paper.Section5.theorem_5_8 (Nat.succ_pos n) (W K N dt) (step_W_closed dt) (step_W_fibreConical dt) hinv (MvPolynomial.map (algebraMap k K) P) (by rw [step_W_P_axis hform hdeg hP]; exact one_ne_zero) (step_W_P_vanishes hform hdeg hP) (step_W_avoids_H hN hform)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- As `K` is algebraically closed, `W = ∅` gives `1 ∈ Ann gr Q_K` (Theorem 2.12(b)), so `gr Q_K = 0`. -/ theorem step_grQK_zero [IsAlgClosed K] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hW : W K N dt = ∅) : Subsingleton (grQ K N (dtK K dt)) := by have h1 := (theorem_2_12 K _).2.1.1 hW refine ⟨fun a b => ?_⟩ rw [← one_smul (SymbolRing K (n + 1)) a, ← one_smul (SymbolRing K (n + 1)) b, Module.mem_annihilator.mp h1, Module.mem_annihilator.mp h1] omit [Field K] [Algebra k K] in
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- An exhaustive filtration `M_0 ⊆ M_1 ⊆ …` with `M_{-1} = 0` and `gr M = ⊕_m M_m/M_{m-1} = 0` has `M = 0`: by induction on `m`, `M_m = M_{m-1} = 0`. -/ theorem step_filtration_zero {F : Type v} [Field F] {M : Type*} [AddCommGroup M] [Module F M] (Mf : ℕ → Submodule F M) (hex : ∀ x : M, ∃ i, x ∈ Mf i) (hgr : Subsingleton (Section2.gr Mf)) : Subsingleton M := by classical have hpiece : ∀ m (x : Mf m), x ∈ (lowerPiece Mf m).comap (Mf m).subtype := fun m x => by rw [← Submodule.Quotient.mk_eq_zero] apply DirectSum.of_injective (β := grPiece Mf) m rw [map_zero] exact Subsingleton.elim _ _ have hzero : ∀ m, Mf m = ⊥ := by intro m induction m with | zero => rw [eq_bot_iff] intro x hx simpa [lowerPiece] using hpiece 0 ⟨x, hx⟩ | succ m ih => rw [eq_bot_iff] intro x hx have h := hpiece (m + 1) ⟨x, hx⟩ change x ∈ Mf m at h rw [ih] at h exact h refine ⟨fun a b => ?_⟩ obtain ⟨i, hi⟩ := hex a obtain ⟨j, hj⟩ := hex b rw [hzero, Submodule.mem_bot] at hi hj rw [hi, hj]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- "Since the filtration of `Q_K` is exhaustive with `Q_{K,-1} = 0`, induction on `m` implies `Q_{K,m} = 0` for all `m`, so `Q_K = 0`": `gr Q_K` is the associated graded module of `Q_{K,m} = qF_mA_n(K)` (`grCyclicEquiv`). -/ theorem step_QK_zero {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (h : Subsingleton (grQ K N (dtK K dt))) : Subsingleton (Q K N (dtK K dt)) := step_filtration_zero _ (cyclicFiltration_order_isCompatible K _).exhaustive ((grCyclicEquiv K (idealI K N (dtK K dt))).toEquiv.symm.subsingleton)
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicEmptySupportVanishingSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicInitialIdealHomogeneousSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
local instance orderGradedAlgebraInstance {n : ℕ} : GradedAlgebra (MvPolynomial.weightedHomogeneousSubmodule K (@orderWeight n)) :=
/-- **Theorem 5.8.** Let `K` be algebraically closed of characteristic zero and `W ⊆ T*X` Zariski closed, fibre-conical and involutive. If `P ∈ K[ξ_1, …, ξ_n]` with `P(1, 0, …, 0) ≠ 0` vanishes on `W` and `W ∩ π⁻¹(H) = ∅`, then `W = ∅`. Proved along Subsection 5.3. -/ theorem theorem_5_8 (hm : 0 < m) (W : Set (PhaseVar m → k)) (hclosed : ∃ L : Ideal (SymbolRing k m), W = MvPolynomial.zeroLocus k L) (hcone : IsFibreConical W) (hinv : IsInvolutive (MvPolynomial.vanishingIdeal k W)) (P : MvPolynomial (Fin m) k) (haxis : MvPolynomial.eval (fun i : Fin m => if i = ⟨0, hm⟩ then (1 : k) else 0) P ≠ 0) (hPW : ∀ q ∈ W, MvPolynomial.eval (fun i => q (.inr i)) P = 0) (hH : W ∩ {q | q (.inl ⟨0, hm⟩) = 0} = ∅) : W = ∅ := by -- Suppose `W ≠ ∅`. by_contra hne have hW : W.Nonempty := Set.nonempty_iff_ne_empty.2 hne -- `B` is radical with a minimal prime `𝔮`; `Y_0 = V_K(𝔮)`. obtain ⟨hrad, -, -, 𝔮, h𝔮⟩ := step_minimal_primes W hclosed hcone hW obtain ⟨hprime, -, -, -, havoid⟩ := step_component hm W hclosed hcone hH 𝔮 h𝔮 -- The separator `s`, then `P` vanishes on `N(Y_0)`. obtain ⟨s, hs𝔮, hs⟩ := step_separator W 𝔮 hrad h𝔮 have hPN := step_P_vanishes_on_conormal_cone W hclosed hcone hinv 𝔮 h𝔮 s hs𝔮 hs P hPW -- Theorem 5.7(i) contradicts `P(1, 0, …, 0) ≠ 0`. exact haxis (step_theorem_5_7_applies hm 𝔮 hprime havoid P (step_P_vanishes_on_N 𝔮 hprime P hPN))
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.Characteristic.PostScalarExtensionPoissonSource ↗open Stafford38.Geometry.AffineConormalSpanSource ↗open Stafford38.Geometry.CoisotropicTranslationSource ↗open Stafford38.Geometry.ConormalPrincipalOpenDensitySource ↗open Stafford38.Geometry.FibreConicalVanishingIdealSource ↗open Stafford38.Geometry.GeneralCoisotropicSetsSource ↗open Stafford38.Geometry.GeneralConormalContainmentSource ↗open Stafford38.Geometry.ProjectiveConormalDirectionsSource ↗open Stafford38.Geometry.SmoothAffineConormalSource ↗universe uSource ↗
variable {k : Type u} [Field k] [IsAlgClosed k] [CharZero k] {m : ℕ}
/-- **Lemma 6.2.** Let `K/k` be a field extension, `𝓘_K = d̃A_n(K) + x₁^N d̃A_n(K)` and `Q_K = Q ⊗_k K = A_n(K)/𝓘_K` with the order filtration `Q_{K,i} = qF_iA_n(K)`. Then `gr Q_K ≅ gr Q ⊗_k K` (as `K`-modules; Lean writes the factor `K` on the left, `K ⊗_k gr Q`), `Ann_{K[x,ξ]} gr Q_K = (Ann_{k[x,ξ]} gr Q) K[x,ξ]`, and `Q = 0 ⇔ Q_K = 0`. In particular, `Char(Q_K)` is the preimage of `Char(Q)` under `Spec K[x,ξ] → Spec k[x,ξ]`. -/ theorem lemma_6_2 {n N : ℕ} (dt : PresentedWeyl k (n + 1)) : Nonempty (grQ K N (dtK K dt) ≃ₗ[K] K ⊗[k] grQ k N dt) ∧ Module.annihilator (SymbolRing K (n + 1)) (grQ K N (dtK K dt)) = (Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt)).map (σ k K n) ∧ (Subsingleton (Q k N dt) ↔ Subsingleton (Q K N (dtK K dt))) ∧ Section2.Char K (idealI K N (dtK K dt)) = PrimeSpectrum.comap (σ k K n) ⁻¹' Section2.Char k (idealI k N dt) := ⟨⟨(step_6_2_graded (K := K) dt).symm⟩, step_6_2_annihilator dt, step_6_2_vanishing dt, step_6_2_char dt⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.Weyl.PresentedScalarExtensionSource ↗open Stafford38.Weyl.FilteredScalarLiftingSource ↗open Stafford38.Paper.Section2Source ↗open Stafford38.Paper.Section3Source ↗open scoped TensorProductSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylLeadingSymbolSource ↗universe u vSource ↗
variable {k : Type u} {K : Type v} [Field k] [Field K] [Algebra k K]Source ↗
variable {n N : ℕ}
/-- **Proposition 6.1.** `Char(Q) ∩ V(x₁) = ∅`, where `V(x₁) ⊆ Spec k[x, ξ]` is the set of primes containing `x₁`. -/ theorem proposition_6_1 [CharZero k] {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hN : 0 < N) (hform : IsMonicNormalForm k N dt) : Section2.Char k (idealI k N dt) ∩ Vx₁ k n = ∅ := by rw [step_6_1_char_eq_support] exact step_6_1_support_avoids hN hform
/-- Theorem 2.11 for the cyclic module `M = A_n(K)/I` with `M_m = qF_m`, the case used in the proof of Theorem 1.1 (for `Q_K`): for this filtration `gr M` is `grCyclic` as a `K[x, ξ]`-module. -/ theorem theorem_2_11_cyclic (K : Type u) [Field K] [CharZero K] (n : ℕ) (I : RightIdeal (PresentedWeyl K n)) : IsInvolutive K (Module.annihilator (SymbolRing K n) (grCyclic K I)).radical := by have : Module.Finite (PresentedWeyl K n)ᵐᵒᵖ (RightQuotient I) := Stafford38.Paper.Support.Gabber.rightQuotient_finite I rw [← Stafford38.Paper.Support.Gabber.annihilator_qFilt] exact theorem_2_11 K n (Stafford38.Paper.Support.Gabber.qFilt K I) (qFilt_isGoodFiltration K I)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Theorem 2.12 (Hilbert's Nullstellensatz).** Let `K` be algebraically closed and `J ⊆ K[z_1, …, z_N]` an ideal (the variables indexed by any finite type `σ`, e.g. `PhaseVar n` for `K[x, ξ]`). (a) The maximal ideals of `K[z]` are exactly the `𝔪_a`, `a ↦ 𝔪_a` is injective, and `V_K(J)` corresponds to the closed points of `V(J)`; (b) `V_K(J) = ∅` iff `1 ∈ J`; (c) `I(V_K(J)) = √J`. -/ theorem theorem_2_12 [IsAlgClosed K] [Finite σ] (J : Ideal (MvPolynomial σ K)) : ((∀ M : Ideal (MvPolynomial σ K), M.IsMaximal ↔ ∃ a, M = maxIdealAt K a) ∧ Function.Injective (maxIdealAt K (σ := σ)) ∧ ∀ a, a ∈ VK K J ↔ closedPoint K a ∈ V J) ∧ (VK K J = ∅ ↔ (1 : MvPolynomial σ K) ∈ J) ∧ vanishingIdealOf K (VK K J) = J.radical := by have hc : vanishingIdealOf K (VK K J) = J.radical := by rw [vanishingIdealOf, VK_eq_zeroLocus, MvPolynomial.vanishingIdeal_zeroLocus_eq_radical] refine ⟨⟨fun M => ?_, ?_, mem_VK_iff_closedPoint_mem_V K J⟩, ⟨fun h => ?_, fun h => ?_⟩, hc⟩ · simp only [maxIdealAt_eq_vanishingIdeal] exact MvPolynomial.isMaximal_iff_eq_vanishingIdeal_singleton · intro a b hab funext i have hmem : MvPolynomial.X i - MvPolynomial.C (a i) ∈ maxIdealAt K b := hab ▸ Ideal.subset_span ⟨i, rfl⟩ rw [mem_maxIdealAt_iff] at hmem simpa [sub_eq_zero, eq_comm] using hmem · have h1 : (1 : MvPolynomial σ K) ∈ J.radical := by rw [← hc, h, mem_vanishingIdealOf] simp exact (Ideal.radical_eq_top.mp ((Ideal.eq_top_iff_one _).mpr h1)).symm ▸ Submodule.mem_top · ext a simp only [Set.mem_empty_iff_false, iff_false] intro ha simpa using ha 1 h
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe u vSource ↗
variable (K : Type u) [Field K] {σ : Type v}
/-- **Proposition 3.6.** Equip `Q` with the order filtration `Q_m = qF_m` induced by the cyclic generator `q = [1]`, and write `P = σ_N^ord(d̃)`. Then `P ∈ Ann_{k[x,ξ]} gr Q` (eq. (3.5)). -/ theorem proposition_3_6 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : orderSymbol k N dt ∈ Module.annihilator (SymbolRing k (n + 1)) (grQ k N dt) := by obtain ⟨⟨hF, -⟩, -⟩ := lemma_3_5 k hform hdeg rw [Module.mem_annihilator] intro m obtain ⟨f, rfl⟩ := (step_3_6_generated k dt).2.2 m rw [smul_comm, step_3_6_kills k hF, smul_zero]
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Lemma 3.5.** Let `d̃ = p^N + ∑_{j<N} p^j a_j` with `a_j ∈ C`, `deg_B a_j ≤ N - j` as in (3.1), and `N = deg_B d̃`. Then `d̃` has order `N`, and its order principal symbol is a polynomial in `ξ` alone, `σ_N^ord(d̃) = P(ξ)` with `P ∈ k[ξ_1, …, ξ_n]`, `deg P = N` (eq. (3.3)); `P` is homogeneous, monic of degree `N` in `ξ₁`, and `P(1, 0, …, 0) = 1` (eq. (3.4)). -/ theorem lemma_3_5 {n N : ℕ} {dt : PresentedWeyl k (n + 1)} (hform : IsMonicNormalForm k N dt) (hdeg : bernsteinDegree k dt = N) : (dt ∈ orderPiece k (n + 1) N ∧ ∀ j < N, dt ∉ orderPiece k (n + 1) j) ∧ ∃ P : MvPolynomial (Fin (n + 1)) k, orderSymbol k N dt = MvPolynomial.rename Sum.inr P ∧ P.IsHomogeneous N ∧ P.totalDegree = N ∧ IsMonicInXi1 k N P ∧ MvPolynomial.eval (fun i => if i = 0 then 1 else 0) P = 1 := by obtain ⟨hF, hfibre, hhom⟩ := step_3_5_fibre k hdeg obtain ⟨P, hsP, hP⟩ := step_3_5_fibre_polynomial k hfibre hhom have hpure := step_3_5_pure_coefficient k hform have hne : orderSymbol k N dt ≠ 0 := by intro h rw [h] at hpure simp at hpure have hcoeff : MvPolynomial.coeff (Finsupp.single 0 N) P = 1 := by have h := MvPolynomial.coeff_rename_mapDomain (Sum.inr : Fin (n + 1) → Fin (n + 1) ⊕ Fin (n + 1)) Sum.inr_injective P (Finsupp.single 0 N) rw [Finsupp.mapDomain_single, ← hsP] at h simpa [MvPolynomial.coeff] using h.symm.trans hpure obtain ⟨hmonic, htot, heval⟩ := step_3_5_monic k hP hcoeff exact ⟨step_3_5_order k hF hne, P, hsP, hP, htot, hmonic, heval⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicAssociatedGradedModuleSource ↗open Stafford38.CharacteristicFilteredQuotientSource ↗open Stafford38.CharacteristicInitialIdealSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.Geometry.ConormalAxisContradictionSource ↗open Stafford38.WeylAssociatedGradedSource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Proposition 3.2.** Let `0 ≠ d ∈ A_n(k)` with `N := deg_B d ≥ 1`. There are an automorphism `Φ_g` of `A_n(k)` and `c ∈ k^×` with `d̃ := c⁻¹ Φ_g(d) = p^N + ∑_{j<N} p^j a_j`, `a_j ∈ C`, `deg_B a_j ≤ N - j` (eq. (3.1)), and `deg_B d̃ = N`. -/ theorem proposition_3_2 [CharZero k] {n : ℕ} {d : PresentedWeyl k (n + 1)} (hd : d ≠ 0) (hN : 1 ≤ bernsteinDegree k d) : ∃ (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k), c ≠ 0 ∧ IsMonicNormalForm k (bernsteinDegree k d) (c⁻¹ • phiG k g d) ∧ bernsteinDegree k (c⁻¹ • phiG k g d) = bernsteinDegree k d := by obtain ⟨hpiece, hne, hhom, -, -⟩ := bernsteinSymbol_ne_zero k hd obtain ⟨v, hv, hv0⟩ := step_3_2_nonvanishing k hhom hne hN obtain ⟨h, hpres, hhe⟩ := step_3_2_h k hv0 (.inr (0 : Fin (n + 1))) obtain ⟨g, -, hmul⟩ := step_3_2_transpose k h hpres obtain ⟨hpiece', heval, hdeg'⟩ := step_3_2_symbol k g hpiece (.inr (0 : Fin (n + 1))) rw [hmul, hhe] at heval hdeg' obtain ⟨hpiece'', hdeg'', hval⟩ := step_3_2_normalize k hpiece' (hdeg' hv) _ hv heval exact ⟨g, _, hv, step_3_2_monic k hpiece'' hval, hdeg''⟩
Variables and notation
Source ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicHomogeneousChartSource ↗open Stafford38.CharacteristicLinearActionSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.WeylFiltrationSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylLeadingSymbolSource ↗open Stafford38.WeylMonicNormalizationSource ↗open Stafford38.WeylPBWMonicBridgeSource ↗open Stafford38.WeylPBWSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
/-- **Remark 3.3.** Since `Φ_g` is a ring automorphism and `c` a central unit, `1 = d̃R + x₁^N d̃S` for `d̃ = c⁻¹Φ_g(d)` transports back to `1 = d (c⁻¹Φ_g^{-1}(R)) + F d (c⁻¹Φ_g^{-1}(S))` with `F = Φ_g^{-1}(x₁)^N`. -/ theorem remark_3_3 {n N : ℕ} (g : Matrix.symplecticGroup (Fin (n + 1)) k) (c : k) (d R S : PresentedWeyl k (n + 1)) (h : (1 : PresentedWeyl k (n + 1)) = (c⁻¹ • phiG k g d) * R + x₁ k n ^ N * (c⁻¹ • phiG k g d) * S) : (1 : PresentedWeyl k (n + 1)) = d * (c⁻¹ • (phiG k g).symm R) + (phiG k g).symm (x₁ k n) ^ N * d * (c⁻¹ • (phiG k g).symm S) := by have ht := congrArg (phiG k g).symm h rw [map_one] at ht rw [ht, map_add, map_mul, map_mul, map_mul, map_pow, map_smul, AlgEquiv.symm_apply_apply, smul_mul_assoc, mul_smul_comm, mul_smul_comm, smul_mul_assoc, mul_smul_comm]
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylEulerResidueSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗open Stafford38.Paper.Section2Source ↗universe uSource ↗
variable (k : Type u) [Field k]
theorem universalStatement : UniversalStatement.{u} := UniversalAssembly.universalStatement_of_canonicalSupportVanishing FoundationClosure.canonicalSupportVanishingViaGeneralCoisotropic
theorem universalFixedSourceStatement : FixedSource.UniversalFixedSourceStatement.{u} := FixedSource.universalFixedSourceStatement_of_canonicalSupportVanishing FoundationClosure.canonicalSupportVanishingViaGeneralCoisotropic
/-- Stability under every nonzero scalar dilation in each cotangent fibre. -/ def IsFibreConical (W : Set (PhaseVar n → k)) : Prop := ∀ q ∈ W, ∀ a : k, a ≠ 0 → Sum.elim (fun i => q (.inl i)) (fun i => a * q (.inr i)) ∈ W
/-- **Definition 2.10 (involutive closed set).** A closed set `W ⊆ K^{2n}` of closed points is involutive if `I(W)` is. -/ def IsInvolutiveSet {n : ℕ} (W : Set (PhaseVar n → k)) : Prop := IsInvolutive k (vanishingIdealOf k W)
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
/-- **Definition 2.10 (involutive ideal).** `{J, J} ⊆ J`. -/ def IsInvolutive {n : ℕ} (J : Ideal (SymbolRing k n)) : Prop := ∀ f ∈ J, ∀ g ∈ J, poisson k f g ∈ J
Variables and notation
Source ↗open LinearMap (BilinForm)Source ↗
open ModuleSource ↗open Stafford38.CharacteristicSource ↗open Stafford38.CharacteristicSymplecticCompletionSource ↗open Stafford38.EulerSurjectivitySource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylSymplecticSource ↗universe u v wSource ↗
variable (k : Type u) [Field k]
Comparison notes
Section6.theorem_1_1_paper is the printed proof. n = 0 is step_rank_zero. For n ≥ 1, section_3_reduction gives the scalar case (deg_B d = 0, Section 3) or the normal form d̃ = c⁻¹Φ_g(d) of Proposition 3.2 together with the transport of Remark 3.3. theorem_6_Q_eq_zero proves Q = 0 with K = k̄: W is the set of closed points of Char(Q_K) with I(W) = √Ann gr Q_K (step_W_closed_points, Theorem 2.11), then one declaration per bullet: W closed (step_W_closed), fibre-conical (step_annihilator_homogeneous, step_W_fibreConical), involutive by Gabber (step_W_involutive, Theorem 2.10 for Q_K), P vanishes on W (step_W_P_vanishes, Proposition 3.6 and Lemma 6.2), P(1, 0, …, 0) = 1 (step_W_P_axis, Lemma 3.5), and W ∩ π⁻¹(H) = ∅ (step_Vx₁_preimage, step_char_K_avoids with Proposition 6.1 and Lemma 6.2, step_closedPoint_mem_Vx₁, step_W_avoids_H). Theorem 5.8 gives W = ∅ (step_W_empty), Theorem 2.11(b) gr Q_K = 0 (step_grQK_zero), the exhaustive filtration Q_K = 0 (step_filtration_zero, step_QK_zero), and Lemma 6.2 Q = 0. theorem_1_1_fixedSource_paper is the same proof for the “moreover” clause, F = Φ_g⁻¹(x₁)^{deg_B d}.
Review notes
Corollary 6.3 — Finitely generated torsion modules are cyclic and stably cyclic
is torsion and therefore Specifically, let us denote with and the annihilators of and , i.e.
and are elements in , since and
Because , also , i.e. is an element of .
On the other hand, can be generated by and . We can check this by finding such that for all we can write
Since can be generated by and and is an element of , it follows that can be generated by , i.e. is cyclic.
For stable cyclicity we see that since were arbitrary generators, for we can set and choose , then generates . For we simply note that by the right module version of [Stafford 1978, Theorem 3.7], is already proven to be stably two-generated, therefore, T is stably cyclic. ∎
Paper.corollary_6_3,
Paper.corollary_6_3_stablyCyclic,
Section6.corollary_6_3_of;
Mathlib-only statements Stafford38Challenge.PaperCorollary_6_3,
Stafford38Challenge.PaperCorollary_6_3_stablyCyclic
/-- **Corollary 6.3** (Stafford 1978). Every finitely generated torsion right `A_n(k)`-module is cyclic. -/ def PaperCorollary_6_3 : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (WeylAlg k n)ᵐᵒᵖ M] [Module.Finite (WeylAlg k n)ᵐᵒᵖ M], Stafford38.TorsionCyclicity.IsRightTorsion (A := WeylAlg k n) (M := M) → IsCyclicModule (WeylAlg k n) M
/-- **Corollary 6.3** (Stafford 1978). Every finitely generated torsion right `A_n(k)`-module is cyclic. Proved as printed by `Section6.corollary_6_3_of` from the cited [Sta78, Thm. 3.7] (`Section6.stafford_1978_theorem_3_7`), Theorem 1.1 and the right Ore condition (`Section6.weyl_rightOreCondition`). -/ theorem corollary_6_3 (k : Type u) [Field k] [CharZero k] (n : ℕ) (T : Type v) [AddCommGroup T] [Module (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] [Module.Finite (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] (hT : Section2.IsTorsion (Stafford38Challenge.WeylAlg k n) T) : Section2.IsCyclic (Stafford38Challenge.WeylAlg k n) T := (Section6.corollary_6_3_of (Section6.weyl_rightOreCondition (k := k) n) (fun d hd => theorem_1_1 k n d hd) hT (Section6.stafford_1978_theorem_3_7 k n T hT)).1
/-- **Proof of Corollary 6.3**, given the cited theorem [Sta78, Thm. 3.7] (`h37`: `T` is generated by two elements and stably two-generated). Take generators `t₁, t₂` (`h37`), `φ(r, s) = t₁r + t₂s` onto `T`, a common annihilator `d = d₁r₁ = d₂r₂` (Ore condition), `1 = dR + FdS`; then `(a, 1 - Fa) ∈ M = ker φ` and `(a, 1 - Fa), (1, -F)` generate `A ⊕ A`, so `T` is generated by `φ(1, -F) = t₁ - t₂F`. The closing remark, stable cyclicity: for `l = 1` choose `f_1 = -F` (`step_6_3_pair`); for `l ≥ 2` it is the stable two-generation of `h37`. -/ theorem corollary_6_3_of (hOre : RightOreCondition A) (hone : ∀ d : A, d ≠ 0 → ∃ F R S : A, (1 : A) = d * R + F * d * S) (htorsion : Stafford38.Paper.Section2.IsTorsion A T) (h37 : (∃ t₁ t₂ : T, Submodule.span Aᵐᵒᵖ ({t₁, t₂} : Set T) = ⊤) ∧ IsStablyGenerated A T 2) : Stafford38.Paper.Section2.IsCyclic A T ∧ IsStablyGenerated A T 1 := by obtain ⟨⟨t₁, t₂, ht⟩, h2⟩ := h37 obtain ⟨d, hd, h₁, h₂⟩ := step_6_3_common_annihilator hOre htorsion t₁ t₂ obtain ⟨F, R, S, hcert⟩ := hone d hd refine ⟨(Stafford38.Paper.Section2.isCyclic_iff_span A T).2 ⟨_, (step_6_3_pair t₁ t₂ d F R S h₁ h₂ hcert).trans ht⟩, ?_⟩ intro l hl a ha rcases Nat.lt_or_ge l 2 with hl2 | hl2 · obtain rfl : l = 1 := by omega obtain ⟨d', hd', h₁', h₂'⟩ := step_6_3_common_annihilator hOre htorsion (a 0) (a 1) obtain ⟨F', R', S', hcert'⟩ := hone d' hd' refine ⟨fun _ => -F', ?_⟩ have hrange : (Set.range fun i : Fin 1 => a i.castSucc + (MulOpposite.op (-F') : Aᵐᵒᵖ) • a (Fin.last 1)) = {a 0 - (MulOpposite.op F' : Aᵐᵒᵖ) • a 1} := by ext z simp [sub_eq_add_neg, Fin.last] exact eq_comm have hpair2 : ({a 0, a 1} : Set T) = Set.range a := by ext z simp [Fin.exists_fin_two, eq_comm] rw [hrange, step_6_3_pair (a 0) (a 1) d' F' R' S' h₁' h₂' hcert', hpair2, ha] · exact h2 l hl2 a ha
More Lean proofs and definitions
theorem corollary_6_3 : PaperCorollary_6_3 := by intro k _ _ n M _ _ _ hM exact Stafford38.Paper.corollary_6_3 k n M hM
/-- **Corollary 6.3, stable cyclicity** (tex lines 75, 108, 1619). Every finitely generated torsion right `A_n(k)`-module is stably cyclic. -/ def PaperCorollary_6_3_stablyCyclic : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (WeylAlg k n)ᵐᵒᵖ M] [Module.Finite (WeylAlg k n)ᵐᵒᵖ M], Stafford38.TorsionCyclicity.IsRightTorsion (A := WeylAlg k n) (M := M) → IsStablyCyclic (WeylAlg k n) M
/-- Stafford's *stably cyclic* [Sta78, p. 435], as quoted in the proof of Corollary 6.3 (tex line 1619): "given `l ≥ 1` and `a_1, …, a_{l+1} ∈ T` that generate `T`, there exist `f_i ∈ A_n` such that `a_i + a_{l+1} f_i`, `i = 1, …, l`, also generate `T`". -/ def IsStablyCyclic (A : Type u) (T : Type v) [Ring A] [AddCommGroup T] [Module Aᵐᵒᵖ T] : Prop := ∀ l ≥ 1, ∀ a : Fin (l + 1) → T, Submodule.span Aᵐᵒᵖ (Set.range a) = ⊤ → ∃ f : Fin l → A, Submodule.span Aᵐᵒᵖ (Set.range fun i : Fin l => a i.castSucc + (MulOpposite.op (f i) : Aᵐᵒᵖ) • a (Fin.last l)) = ⊤
/-- **Corollary 6.3, stable cyclicity** (closing paragraph of the printed proof, tex 1619). Every finitely generated torsion right `A_n(k)`-module is stably cyclic: for `l = 1` take `f₁ = -F`, for `l ≥ 2` use the stable two-generation of the cited [Sta78, Thm. 3.7] (`Section6.corollary_6_3_of`, second component). -/ theorem corollary_6_3_stablyCyclic (k : Type u) [Field k] [CharZero k] (n : ℕ) (T : Type v) [AddCommGroup T] [Module (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] [Module.Finite (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] (hT : Section2.IsTorsion (Stafford38Challenge.WeylAlg k n) T) : Section6.IsStablyGenerated (Stafford38Challenge.WeylAlg k n) T 1 := (Section6.corollary_6_3_of (Section6.weyl_rightOreCondition (k := k) n) (fun d hd => theorem_1_1 k n d hd) hT (Section6.stafford_1978_theorem_3_7 k n T hT)).2
theorem corollary_6_3_stablyCyclic : PaperCorollary_6_3_stablyCyclic := by intro k _ _ n M _ _ _ hM exact Stafford38.Paper.corollary_6_3_stablyCyclic k n M hM
/-- "This map is surjective, since `t₁, t₂` generate all of `T`": the image of `φ` is `t₁A + t₂A`. -/ theorem step_6_3_phi_range (t₁ t₂ : T) : LinearMap.range (phi63 (A := A) t₁ t₂) = Submodule.span Aᵐᵒᵖ ({t₁, t₂} : Set T) := by apply le_antisymm · rintro z ⟨⟨a, b⟩, rfl⟩ exact Submodule.add_mem _ (Submodule.smul_mem _ _ (Submodule.subset_span (by simp))) (Submodule.smul_mem _ _ (Submodule.subset_span (by simp))) · apply Submodule.span_le.2 intro z hz simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at hz rcases hz with rfl | rfl · exact ⟨(1, 0), by simp [phi63]⟩ · exact ⟨(0, 1), by simp [phi63]⟩
/-- "`T` is torsion": `t₁d₁ = t₂d₂ = 0` with `d₁, d₂ ≠ 0`. "`A_n` is a right Ore domain": `d₁r₁ = d₂r₂ ≠ 0`, and `d := d₁r₁ = d₂r₂` is a common annihilator. -/ theorem step_6_3_common_annihilator (hOre : RightOreCondition A) (htorsion : Stafford38.Paper.Section2.IsTorsion A T) (t₁ t₂ : T) : ∃ d : A, d ≠ 0 ∧ (MulOpposite.op d : Aᵐᵒᵖ) • t₁ = 0 ∧ (MulOpposite.op d : Aᵐᵒᵖ) • t₂ = 0 := by obtain ⟨d₁, hd₁, h₁⟩ := htorsion t₁ obtain ⟨d₂, hd₂, h₂⟩ := htorsion t₂ obtain ⟨r₁, r₂, hr, hne⟩ := hOre hd₁ hd₂ refine ⟨d₁ * r₁, hne, ?_, ?_⟩ · rw [MulOpposite.op_mul, mul_smul, h₁, smul_zero] · rw [hr, MulOpposite.op_mul, mul_smul, h₂, smul_zero]
/-- "`(d, 0)` and `(0, d)` are elements in `M = ker φ`", hence `(d, 0)S + (0, d)R = (dS, dR) = (dS, 1 - FdS) =: (a, 1 - Fa) ∈ M`; and `φ(1, -F) = t₁ - t₂F`. -/ theorem step_6_3_kernel (t₁ t₂ : T) (d F R S : A) (h₁ : (MulOpposite.op d : Aᵐᵒᵖ) • t₁ = 0) (h₂ : (MulOpposite.op d : Aᵐᵒᵖ) • t₂ = 0) (hcert : (1 : A) = d * R + F * d * S) : ((d, 0) : A × A) ∈ LinearMap.ker (phi63 t₁ t₂) ∧ ((0, d) : A × A) ∈ LinearMap.ker (phi63 t₁ t₂) ∧ (MulOpposite.op S : Aᵐᵒᵖ) • ((d, 0) : A × A) + (MulOpposite.op R : Aᵐᵒᵖ) • (0, d) = (d * S, 1 - F * (d * S)) ∧ ((d * S, 1 - F * (d * S)) : A × A) ∈ LinearMap.ker (phi63 t₁ t₂) ∧ phi63 t₁ t₂ (1, -F) = t₁ - (MulOpposite.op F : Aᵐᵒᵖ) • t₂ := by have hd0 : ((d, 0) : A × A) ∈ LinearMap.ker (phi63 t₁ t₂) := by change (MulOpposite.op d : Aᵐᵒᵖ) • t₁ + (MulOpposite.op (0 : A) : Aᵐᵒᵖ) • t₂ = 0 rw [h₁, MulOpposite.op_zero, zero_smul, add_zero] have h0d : ((0, d) : A × A) ∈ LinearMap.ker (phi63 t₁ t₂) := by change (MulOpposite.op (0 : A) : Aᵐᵒᵖ) • t₁ + (MulOpposite.op d : Aᵐᵒᵖ) • t₂ = 0 rw [h₂, MulOpposite.op_zero, zero_smul, add_zero] have hcomb : (MulOpposite.op S : Aᵐᵒᵖ) • ((d, 0) : A × A) + (MulOpposite.op R : Aᵐᵒᵖ) • (0, d) = (d * S, 1 - F * (d * S)) := by simp only [Prod.smul_mk, MulOpposite.smul_eq_mul_unop, MulOpposite.unop_op, zero_mul, Prod.mk_add_mk, add_zero, zero_add, Prod.mk.injEq, true_and] rw [hcert] noncomm_ring refine ⟨hd0, h0d, hcomb, ?_, ?_⟩ · rw [← hcomb] exact Submodule.add_mem _ (Submodule.smul_mem _ _ hd0) (Submodule.smul_mem _ _ h0d) · change (MulOpposite.op (1 : A) : Aᵐᵒᵖ) • t₁ + (MulOpposite.op (-F) : Aᵐᵒᵖ) • t₂ = _ simp [sub_eq_add_neg]
/-- "`A_n ⊕ A_n` can be generated by `(a, 1 - Fa)` and `(1, -F)`": `(x, y) = (a, 1 - Fa)u + (1, -F)v` with `u = y + Fx` and `v = x - a(y + Fx)`. -/ theorem step_6_3_generators (a F : A) : (∀ x y : A, ((x, y) : A × A) = (MulOpposite.op (y + F * x) : Aᵐᵒᵖ) • (a, 1 - F * a) + (MulOpposite.op (x - a * (y + F * x)) : Aᵐᵒᵖ) • (1, -F)) ∧ Submodule.span Aᵐᵒᵖ ({(a, 1 - F * a), (1, -F)} : Set (A × A)) = ⊤ := by have hid : ∀ x y : A, ((x, y) : A × A) = (MulOpposite.op (y + F * x) : Aᵐᵒᵖ) • (a, 1 - F * a) + (MulOpposite.op (x - a * (y + F * x)) : Aᵐᵒᵖ) • (1, -F) := by intro x y simp only [Prod.smul_mk, MulOpposite.smul_eq_mul_unop, MulOpposite.unop_op, Prod.mk_add_mk, one_mul, Prod.mk.injEq] constructor <;> noncomm_ring refine ⟨hid, top_unique ?_⟩ rintro ⟨x, y⟩ - rw [hid x y] exact Submodule.add_mem _ (Submodule.smul_mem _ _ (Submodule.subset_span (by simp))) (Submodule.smul_mem _ _ (Submodule.subset_span (by simp)))
/-- "Since `A_n ⊕ A_n` can be generated by `(a, 1 - Fa)` and `(1, -F)` and `(a, 1 - Fa) ∈ M`, `T ≅ (A_n ⊕ A_n)/M` can be generated by `φ(1, -F) = t₁ - t₂F`": the image `t₁A + t₂A` of `φ` is generated by `t₁ - t₂F`. (For `l = 1` this is also the stable-cyclicity statement with `f₁ = -F`.) -/ theorem step_6_3_pair (t₁ t₂ : T) (d F R S : A) (h₁ : (MulOpposite.op d : Aᵐᵒᵖ) • t₁ = 0) (h₂ : (MulOpposite.op d : Aᵐᵒᵖ) • t₂ = 0) (hcert : (1 : A) = d * R + F * d * S) : Submodule.span Aᵐᵒᵖ ({t₁ - (MulOpposite.op F : Aᵐᵒᵖ) • t₂} : Set T) = Submodule.span Aᵐᵒᵖ ({t₁, t₂} : Set T) := by obtain ⟨-, -, -, hzero, hgen⟩ := step_6_3_kernel t₁ t₂ d F R S h₁ h₂ hcert rw [← step_6_3_phi_range, ← Submodule.map_top, ← (step_6_3_generators (d * S) F).2, Submodule.map_span, Set.image_pair, LinearMap.mem_ker.mp hzero, hgen] apply le_antisymm · exact Submodule.span_mono (by simp) · apply Submodule.span_le.2 intro z hz simp only [Set.mem_insert_iff, Set.mem_singleton_iff] at hz rcases hz with rfl | rfl · exact Submodule.zero_mem _ · exact Submodule.subset_span rfl
/-- **[Sta78, Thm. 3.7]** (right-module version), as cited in the proof of Corollary 6.3: every finitely generated torsion right `A_n(k)`-module `T` is generated by two elements, and is stably two-generated (for `l ≥ 2` and generators `a_1, …, a_{l+1}` there are `f_i` with `a_i + a_{l+1}f_i`, `i ≤ l`, generating `T`). Proved here from Theorem 1.1, which does not use Corollary 6.3: `two_generators_of_identity` and `stably_generated_of_identity`. -/ theorem stafford_1978_theorem_3_7 (k : Type u) [Field k] [CharZero k] (n : ℕ) (T : Type v) [AddCommGroup T] [Module (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] [Module.Finite (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ T] (hT : Section2.IsTorsion (Stafford38Challenge.WeylAlg k n) T) : (∃ t₁ t₂ : T, Submodule.span (Stafford38Challenge.WeylAlg k n)ᵐᵒᵖ ({t₁, t₂} : Set T) = ⊤) ∧ IsStablyGenerated (Stafford38Challenge.WeylAlg k n) T 2 := ⟨two_generators_of_identity (weyl_rightOreCondition (k := k) n) (fun d hd => Stafford38.Paper.theorem_1_1 k n d hd) hT, fun l hl => stably_generated_of_identity (weyl_rightOreCondition (k := k) n) (fun d hd => Stafford38.Paper.theorem_1_1 k n d hd) hT l (by omega)⟩
/-- The two-generator part of the cited theorem [Sta78, Thm. 3.7] (right-module version), proved here from the identity `1 = dR + FdS` rather than by Stafford's argument: for a finite generating set `{x} ∪ s`, collapsing `s` to one generator `y` by repeating `step_6_3_pair` gives the two generators `t₁ = x`, `t₂ = y`. -/ theorem two_generators_of_identity (hOre : RightOreCondition A) (hone : ∀ d : A, d ≠ 0 → ∃ F R S : A, (1 : A) = d * R + F * d * S) (htorsion : Stafford38.Paper.Section2.IsTorsion A T) [Module.Finite Aᵐᵒᵖ T] : ∃ t₁ t₂ : T, Submodule.span Aᵐᵒᵖ ({t₁, t₂} : Set T) = ⊤ := by classical have hcollapse (s : Finset T) : ∃ z : T, Submodule.span Aᵐᵒᵖ (s : Set T) = Submodule.span Aᵐᵒᵖ ({z} : Set T) := by induction s using Finset.induction_on with | empty => exact ⟨0, by simp⟩ | @insert x s _ ih => obtain ⟨y, hy⟩ := ih obtain ⟨d, hd, hx, hyd⟩ := step_6_3_common_annihilator hOre htorsion x y obtain ⟨F, R, S, hcert⟩ := hone d hd refine ⟨x - (MulOpposite.op F : Aᵐᵒᵖ) • y, ?_⟩ rw [step_6_3_pair x y d F R S hx hyd hcert, Finset.coe_insert, Submodule.span_insert, hy, ← Submodule.span_union, Set.singleton_union] obtain ⟨s, hs⟩ := (Module.Finite.fg_top : (⊤ : Submodule Aᵐᵒᵖ T).FG) rcases s.eq_empty_or_nonempty with rfl | ⟨x, hx⟩ · exact ⟨0, 0, by rw [← hs]; simp⟩ · obtain ⟨y, hy⟩ := hcollapse (s.erase x) refine ⟨x, y, ?_⟩ rw [← hs, ← Finset.insert_erase hx, Finset.coe_insert, Submodule.span_insert, Submodule.span_insert, hy]
/-- Replacing `a_1` by `a_1 - a_{l+1}F` (`step_6_3_pair`) and keeping `a_2, …, a_l`: with `1 = dR + FdS` for a common annihilator `d` of `a_1` and `a_{l+1}`, the stable-generation property holds for every `l ≥ 1`. -/ theorem stably_generated_of_identity (hOre : RightOreCondition A) (hone : ∀ d : A, d ≠ 0 → ∃ F R S : A, (1 : A) = d * R + F * d * S) (htorsion : Stafford38.Paper.Section2.IsTorsion A T) : IsStablyGenerated A T 1 := by classical intro l hl a ha obtain ⟨d, hd, h₁, h₂⟩ := step_6_3_common_annihilator hOre htorsion (a 0) (a (Fin.last l)) obtain ⟨F, R, S, hcert⟩ := hone d hd let i₀ : Fin l := ⟨0, hl⟩ refine ⟨fun i => if i = i₀ then -F else 0, ?_⟩ have hpair := step_6_3_pair (a 0) (a (Fin.last l)) d F R S h₁ h₂ hcert rw [eq_top_iff, ← ha, Submodule.span_le] rintro _ ⟨j, rfl⟩ let U := Submodule.span Aᵐᵒᵖ (Set.range fun i : Fin l => a i.castSucc + (MulOpposite.op (if i = i₀ then -F else 0) : Aᵐᵒᵖ) • a (Fin.last l)) have hmem0 : a 0 - (MulOpposite.op F : Aᵐᵒᵖ) • a (Fin.last l) ∈ U := by refine Submodule.subset_span ⟨i₀, ?_⟩ simp only [↓reduceIte, MulOpposite.op_neg, neg_smul, sub_eq_add_neg] rfl have hpairU : Submodule.span Aᵐᵒᵖ ({a 0, a (Fin.last l)} : Set T) ≤ U := by rw [← hpair, Submodule.span_le, Set.singleton_subset_iff] exact hmem0 by_cases hj0 : j = 0 · subst hj0 exact hpairU (Submodule.subset_span (by simp)) by_cases hjl : j = Fin.last l · subst hjl exact hpairU (Submodule.subset_span (by simp)) · obtain ⟨i, rfl⟩ := Fin.exists_castSucc_eq.mpr hjl have hi : i ≠ i₀ := fun h => hj0 (by rw [h]; rfl) refine Submodule.subset_span ⟨i, ?_⟩ simp [hi]
/-- **`A_n(k)` is a right Ore domain** [Björk, Prop. 1.8.4]: nonzero `d₁, d₂` have `r₁, r₂` with `d₁r₁ = d₂r₂ ≠ 0`. -/ theorem weyl_rightOreCondition (n : ℕ) : AlgebraicAnalysis.OreRightIntersection.RightOreCondition (PresentedWeyl k n) := by intro a b ha hb by_contra hcon have hab : ∀ x y : PresentedWeyl k n, a * x = b * y → a * x = 0 := by intro x y hxy by_contra hne exact hcon ⟨x, y, hxy, hne⟩ obtain ⟨Na, hNa⟩ := exists_mem_bernsteinPiece k a obtain ⟨Nb, hNb⟩ := exists_mem_bernsteinPiece k b let N := max Na Nb have haN : a ∈ bernsteinPiece k n N := presentedWeightPiece_mono k bernsteinWeight (le_max_left Na Nb) hNa have hbN : b ∈ bernsteinPiece k n N := presentedWeightPiece_mono k bernsteinWeight (le_max_right Na Nb) hNb have h10 : (1 : PresentedWeyl k n) ≠ 0 := fun h => ha (by rw [← mul_one a, h, mul_zero]) have hgrow : ∀ j : ℕ, 2 ^ j ≤ Module.finrank k (bernsteinPiece k n (j * N)) := by intro j induction j with | zero => have hfin := (bernsteinPiece_finite_finrank_le (k := k) n 0).1 rw [pow_zero, zero_mul, Nat.one_le_iff_ne_zero] intro h0 rw [Submodule.finrank_eq_zero] at h0 have h1 := one_mem_bernsteinPiece_zero (k := k) n rw [h0, Submodule.mem_bot] at h1 exact h10 h1 | succ j ih => have h := two_mul_finrank_le ha hb hab haN hbN (j * N) rw [show (j + 1) * N = N + j * N by ring, pow_succ] omega obtain ⟨j, hj⟩ := exists_pow_lt_two_pow N (Fintype.card (PhaseVar n)) have hle := (bernsteinPiece_finite_finrank_le (k := k) n (j * N)).2 have := hgrow j omega
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
theorem corollary_6_3 : PaperCorollary_6_3 := paperCorollary_6_3_of fun k _ _ n M _ _ _ hM => Stafford38.TorsionCyclicity.weyl_isCyclic_of_isRightTorsion k n M hM
/-- Every finitely generated torsion right module over a Weyl algebra is cyclic. The main theorem supplies the coefficient identity used at each pair-replacement step in the finite-generator induction. -/ theorem weyl_isCyclic_of_isRightTorsion (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (Stafford38.WeylAlg k n)ᵐᵒᵖ M] [Module.Finite (Stafford38.WeylAlg k n)ᵐᵒᵖ M] (hM : IsRightTorsion (A := Stafford38.WeylAlg k n) (M := M)) : ∃ z : M, Submodule.span (Stafford38.WeylAlg k n)ᵐᵒᵖ ({z} : Set M) = ⊤ := Stafford38.PaperCyclicity.paper_torsion_module_is_cyclic (k := k) (n := n) (M := M) hM
Variables and notation
Source ↗universe u vSource ↗
variable {A : Type u} {M : Type v} [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
/-- Every finitely generated torsion right module over a characteristic-zero Weyl algebra is generated by one element. -/ def torsionCyclicStatement : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (WeylAlg k n)ᵐᵒᵖ M] [Module.Finite (WeylAlg k n)ᵐᵒᵖ M], IsRightTorsion (A := WeylAlg k n) (M := M) → ∃ z : M, Submodule.span (WeylAlg k n)ᵐᵒᵖ ({z} : Set M) = ⊤
def pairGeneratorStatement : Prop := ∀ (A : Type u) [Ring A] (M : Type v) [AddCommGroup M] [Module Aᵐᵒᵖ M] (x y : M) (d F R S : A), (MulOpposite.op d : Aᵐᵒᵖ) • x = 0 → (MulOpposite.op d : Aᵐᵒᵖ) • y = 0 → (1 : A) = d * R + F * d * S → Submodule.span Aᵐᵒᵖ ({x - (MulOpposite.op F : Aᵐᵒᵖ) • y} : Set M) = Submodule.span Aᵐᵒᵖ ({x, y} : Set M)
theorem pairGeneratorStatement_consumer : pairGeneratorStatement.{u, v} := by intro A _ M _ _ x y d F R S hx hy hcert exact Stafford38.TorsionCyclicity.span_adjusted_pair_eq_span_pair x y d F R S hx hy hcert
/-- **Definition 2.1 (cyclic).** `M = mA` for some `m ∈ M`. -/ def IsCyclic : Prop := ∃ m : M, ∀ x : M, ∃ a : A, x = MulOpposite.op a • m
Variables and notation
Source ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open AlgebraicAnalysis.OreRightPBWSource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗open Stafford38.WeylPaperRightMonicSource ↗universe u vSource ↗
variable (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M]
/-- Stafford's notion [Sta78, p. 435]: `T` is *stably `m`-generated* if for every `l ≥ m` and generators `a_1, …, a_{l+1}` of `T` there are `f_i ∈ A` such that `a_i + a_{l+1} f_i`, `i = 1, …, l`, generate `T`. Stably cyclic is `m = 1`. -/ def IsStablyGenerated (A : Type u) [Ring A] (T : Type v) [AddCommGroup T] [Module Aᵐᵒᵖ T] (m : ℕ) : Prop := ∀ l ≥ m, ∀ a : Fin (l + 1) → T, Submodule.span Aᵐᵒᵖ (Set.range a) = ⊤ → ∃ f : Fin l → A, Submodule.span Aᵐᵒᵖ (Set.range fun i : Fin l => a i.castSucc + (MulOpposite.op (f i) : Aᵐᵒᵖ) • a (Fin.last l)) = ⊤
Comparison notes
Stafford38Challenge.PaperCorollary_6_3 (Challenge.lean, Mathlib-only) states the corollary as printed, and Stafford38.Paper.corollary_6_3 proves it, for every field of characteristic zero and every n. The proof is Section6.corollary_6_3_of, the printed argument: two generators t₁, t₂ and φ(r, s) = t₁r + t₂s (step_6_3_phi_range), a common annihilator d by the right Ore condition (step_6_3_common_annihilator, weyl_rightOreCondition [Björk, Prop. 1.8.4], proved), (a, 1 − Fa) ∈ ker φ (step_6_3_kernel), generation of A_n ⊕ A_n by (a, 1 − Fa) and (1, −F) (step_6_3_generators), and T generated by t₁ − t₂F (step_6_3_pair); corollary_6_3_of also gives stable cyclicity. The input [Sta78, Thm. 3.7] (T is generated by two elements and stably two-generated) is stated as cited, Section6.stafford_1978_theorem_3_7, and proved in the repository by another route than Stafford's: from Theorem 1.1, which does not use Corollary 6.3 (two_generators_of_identity, stably_generated_of_identity). Solution.lean proves PaperCorollary_6_3 through Paper.corollary_6_3; AlternativeSolution.lean through the earlier TorsionCyclicity.weyl_isCyclic_of_isRightTorsion. Stafford38Challenge.PaperCorollary_6_3_stablyCyclic (Challenge.lean, Mathlib-only) states the stable cyclicity now included in the corollary statement: every finitely generated torsion right A_n(k)-module T is stably cyclic (IsStablyCyclic: for l ≥ 1 and generators a_1, …, a_{l+1} of T there are f_i with a_i + a_{l+1}f_i, i = 1, …, l, generating T). Stafford38.Paper.corollary_6_3_stablyCyclic proves it as the second component of Section6.corollary_6_3_of: f₁ = −F for l = 1, and the stable two-generation of [Sta78, Thm. 3.7] for l ≥ 2. Solution.lean proves PaperCorollary_6_3_stablyCyclic through Paper.corollary_6_3_stablyCyclic.
| # | Proof step | Lean |
|---|---|---|
| 1 | Literal paper generator certificate | Stafford38PaperPairChallenge.pairGeneratorStatement Stafford38PaperPairChallenge.pairGeneratorStatement_consumer |
Review notes
7. Generalizations and open questions
Section 7 — Consequences proved in Lean
Generalizations and open questions
The formalization in Lean discussed in Section 1.2 also proves the following consequences, which have no printed proof: the left-handed identity , via the formal adjoint, which implies Corollary 6.3 also for left torsion modules; the identity for finite-order differential operators on localizations of , including principal opens, Laurent-polynomial coefficients and the fraction field ; and, independently of the geometric part of Section 5, the identity with for , , in any -algebra containing elements with , provided each commutes with and . Lean:Section7.leftHandedIdentity;
Section7.corollary_6_3_left;
Section7.localizedDifferentialIdentity;
Section7.evolutionaryIdentity
/-- Stafford 3.8 is preserved by genuine right Ore localization. -/ theorem s38_rightOreLocalization {R : Type u} [Ring R] {S : Submonoid R} [OreLocalization.OreSet (oppositeSubmonoid S)] (hR : S38 R) : S38 (RightOreLocalization R S) := AlgebraicAnalysis.TwoGeneratorIdentity.of_rightOreLocalization hR
theorem leftHanded_of_universalStatement (h : UniversalStatement.{u}) : LeftHandedStatement.{u} := by intro k _ _ n d hd have htd : transpose k n d ≠ 0 := by intro hz apply hd simpa only [transpose_transpose, transpose_zero] using congrArg (transpose k n) hz rcases h k n (transpose k n d) htd with ⟨F, R, S, hone⟩ refine ⟨transpose k n R, transpose k n S, transpose k n F, ?_⟩ have ht := congrArg (transpose k n) hone have h1 : transpose k n (1 : WeylAlg k n) = 1 := by simp [transpose] rw [h1] at ht simpa [transpose_add, transpose_mul, transpose_transpose, mul_assoc] using ht
More Lean proofs and definitions
theorem s38_laurent_differential (B : Type u) [CommRing B] [Algebra (A (k := k) (n := n)) B] [Algebra k B] [IsScalarTower k (A (k := k) (n := n)) B] [IsLocalization (Submonoid.powers (allCoordinateProduct (k := k) (n := n))) B] : S38 (D (k := k) B) := by apply s38_localized_differential (Submonoid.powers (allCoordinateProduct (k := k) (n := n))) B intro d hd exact Stafford38.universalStatement (k := k) n d hd
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.DifferentialOperatorsSource ↗open Stafford38.LocalizedWeylActionSource ↗open Stafford38.LocalizedDifferentialClearingSource ↗open Stafford38.LocalizationCorollariesSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable {k : Type u} [Field k] [CharZero k] {n : ℕ}Source ↗
variable (S : Submonoid (MvPolynomial (Fin n) k))Source ↗
variable (B : Type u) [CommRing B]Source ↗
variable [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B]Source ↗
variable [IsScalarTower k (MvPolynomial (Fin n) k) B]Source ↗
variable [IsLocalization S B]
theorem s38_partial_laurent_differential (J : Finset (Fin n)) (B : Type u) [CommRing B] [Algebra (A (k := k) (n := n)) B] [Algebra k B] [IsScalarTower k (A (k := k) (n := n)) B] [IsLocalization (Submonoid.powers (J.prod (fun i => (MvPolynomial.X i : A (k := k) (n := n)))) ) B] : S38 (D (k := k) B) := s38_unconditional_localized_differential (Submonoid.powers (J.prod (fun i => (MvPolynomial.X i : A (k := k) (n := n))))) B
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.DifferentialOperatorsSource ↗open Stafford38.LocalizedWeylActionSource ↗open Stafford38.LocalizedDifferentialClearingSource ↗open Stafford38.LocalizationCorollariesSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable {k : Type u} [Field k] [CharZero k] {n : ℕ}Source ↗
variable (S : Submonoid (MvPolynomial (Fin n) k))Source ↗
variable (B : Type u) [CommRing B]Source ↗
variable [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B]Source ↗
variable [IsScalarTower k (MvPolynomial (Fin n) k) B]Source ↗
variable [IsLocalization S B]
theorem s38_fraction_ring_differential (B : Type u) [CommRing B] [Algebra (A (k := k) (n := n)) B] [Algebra k B] [IsScalarTower k (A (k := k) (n := n)) B] [IsFractionRing (A (k := k) (n := n)) B] : S38 (D (k := k) B) := by apply s38_localized_differential (nonZeroDivisors (A (k := k) (n := n))) B intro d hd exact Stafford38.universalStatement (k := k) n d hd
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.DifferentialOperatorsSource ↗open Stafford38.LocalizedWeylActionSource ↗open Stafford38.LocalizedDifferentialClearingSource ↗open Stafford38.LocalizationCorollariesSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable {k : Type u} [Field k] [CharZero k] {n : ℕ}Source ↗
variable (S : Submonoid (MvPolynomial (Fin n) k))Source ↗
variable (B : Type u) [CommRing B]Source ↗
variable [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B]Source ↗
variable [IsScalarTower k (MvPolynomial (Fin n) k) B]Source ↗
variable [IsLocalization S B]
/-- Exact paper-facing form. No commutativity among the coefficients is assumed: each displayed coefficient need only commute with `x` and `p`. -/ theorem evolutionaryCorollary (k : Type*) [Field k] [CharZero k] [Algebra k D] (hw : p * x = x * p + 1) {r : ℕ} (hr : 0 < r) (terms : List (D × ℕ)) (hx : ∀ t ∈ terms, t.1 * x = x * t.1) (hp : ∀ t ∈ terms, t.1 * p = p * t.1) {d : D} (hd : d = p ^ r - potentialSum x terms) : ∃ R S : D, 1 = d * R + x ^ r * d * S := by obtain ⟨R, S, h⟩ := evolution_stafford k hw hr terms hx hp exact ⟨R, S, by simpa [hd] using h.symm⟩
/-- The evolutionary certificate in the literal tensor algebra `B ⊗[k] A₁(k)`. This asserts no identification with a geometric ring of differential operators. -/ theorem tensorEvolutionaryCorollary {r : ℕ} (hr : 0 < r) (terms : List (B × ℕ)) {d : Dᵣ} (hd : d = ιW (presentedMomentum k 0) ^ r - potentialSum (ιW (presentedCoordinate k 0)) (terms.map fun t => (ιB t.1, t.2))) : ∃ R S : Dᵣ, 1 = d * R + ιW (presentedCoordinate k 0) ^ r * d * S := by let lifted : List (Dᵣ × ℕ) := terms.map fun t => (ιB t.1, t.2) have hcomm : ∀ (b : B) (w : W₁), ιB b * ιW w = ιW w * ιB b := by intro b w simp [Algebra.TensorProduct.includeLeft_apply, Algebra.TensorProduct.includeRight_apply, Algebra.TensorProduct.tmul_mul_tmul] have hx : ∀ t ∈ lifted, t.1 * ιW (presentedCoordinate k 0) = ιW (presentedCoordinate k 0) * t.1 := by rintro t ht rcases List.mem_map.mp ht with ⟨⟨b, n⟩, _, rfl⟩ exact hcomm b (presentedCoordinate k 0) have hp : ∀ t ∈ lifted, t.1 * ιW (presentedMomentum k 0) = ιW (presentedMomentum k 0) * t.1 := by rintro t ht rcases List.mem_map.mp ht with ⟨⟨b, n⟩, _, rfl⟩ exact hcomm b (presentedMomentum k 0) have hw : ιW (presentedMomentum k 0) * ιW (presentedCoordinate k 0) = ιW (presentedCoordinate k 0) * ιW (presentedMomentum k 0) + 1 := by calc _ = ιW (presentedMomentum k 0 * presentedCoordinate k 0) := (map_mul ιW _ _).symm _ = ιW (presentedCoordinate k 0 * presentedMomentum k 0 + 1) := congrArg ιW (presentedMomentum_mul_coordinate k 0) _ = _ := by rw [map_add, map_mul, map_one] exact evolutionaryCorollary (D := Dᵣ) (x := ιW (presentedCoordinate k 0)) (p := ιW (presentedMomentum k 0)) k hw hr lifted hx hp hd
Variables and notation
Source ↗open scoped TensorProductSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗variable {D : Type*} [Ring D] {x p : D}Source ↗
variable (k B : Type*) [Field k] [CharZero k] [Ring B] [Algebra k B]Source ↗
local notation "W₁" => PresentedWeyl k 1Source ↗
local notation "Dᵣ" => B ⊗[k] W₁Source ↗local notation "ιB" => (Algebra.TensorProduct.includeLeft : B →ₐ[k] Dᵣ)Source ↗
local notation "ιW" => (Algebra.TensorProduct.includeRight : W₁ →ₐ[k] Dᵣ)
abbrev S38 := AlgebraicAnalysis.TwoGeneratorIdentity
Comparison notes
The first paragraph of Section 7 lists three consequences without printed proofs; each is mapped separately below. The identity in every right Ore localization is not listed. Stafford38/Paper/Section7.lean proves them in the Mathlib-only form of Challenge.lean, and Solution.lean proves the compared theorems through them.
Review notes
Corollary — Left-handed identity via formal adjoint
The formalization in Lean discussed in Section 1.2 also proves the following consequences, which have no printed proof: the left-handed identity , via the formal adjoint, which implies Corollary 6.3 also for left torsion modules; the identity for finite-order differential operators on localizations of , including principal opens,
/-- **Section 7, left-handed identity** (tex lines 1629–1630): every nonzero `d ∈ A_n(k)` has `R, S, F` with `1 = Rd + SdF`. -/ def LeftHandedIdentity : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (d : WeylAlg k n), d ≠ 0 → ∃ R S F : WeylAlg k n, (1 : WeylAlg k n) = R * d + S * d * F
/-- **Left-handed identity.** Every nonzero `d ∈ A_n(k)` has `R, S, F` with `1 = Rd + SdF`. -/ theorem leftHandedIdentity : LeftHandedIdentity.{u} := Stafford38.LeftHandedCorollary.leftHanded_of_universalStatement Stafford38.Paper.theorem_1_1
More Lean proofs and definitions
theorem leftHandedIdentity : LeftHandedIdentity := Stafford38.Paper.Section7.leftHandedIdentity
theorem leftHanded_of_universalStatement (h : UniversalStatement.{u}) : LeftHandedStatement.{u} := by intro k _ _ n d hd have htd : transpose k n d ≠ 0 := by intro hz apply hd simpa only [transpose_transpose, transpose_zero] using congrArg (transpose k n) hz rcases h k n (transpose k n d) htd with ⟨F, R, S, hone⟩ refine ⟨transpose k n R, transpose k n S, transpose k n F, ?_⟩ have ht := congrArg (transpose k n) hone have h1 : transpose k n (1 : WeylAlg k n) = 1 := by simp [transpose] rw [h1] at ht simpa [transpose_add, transpose_mul, transpose_transpose, mul_assoc] using ht
/-- **Section 7, Corollary 6.3 for left torsion modules** (tex line 1630): every finitely generated torsion left `A_n(k)`-module is cyclic. -/ def LeftCorollary_6_3 : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (WeylAlg k n) M] [Module.Finite (WeylAlg k n) M], IsLeftTorsion (WeylAlg k n) M → IsLeftCyclicModule (WeylAlg k n) M
/-- Every element of a left module is killed by some nonzero scalar. -/ def IsLeftTorsion (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module A M] : Prop := ∀ m : M, ∃ d : A, d ≠ 0 ∧ d • m = 0
/-- A left module `M` is cyclic: `M = Am` for some `m ∈ M`. -/ def IsLeftCyclicModule (A : Type u) (M : Type v) [Ring A] [AddCommGroup M] [Module A M] : Prop := ∃ m : M, ∀ x : M, ∃ a : A, x = a • m
/-- **Corollary 6.3 for left torsion modules.** Every finitely generated torsion left `A_n(k)`-module is cyclic. The left analogue of the printed proof (`Section6.corollary_6_3_of` for the ring `A_n(k)ᵐᵒᵖ`, whose right modules are the left `A_n(k)`-modules), fed with the left-handed identity `1 = Rd + SdF` (`leftHandedIdentity`) and the left Ore condition; the two generators of the left version of [Sta78, Thm. 3.7] come from the same identity. -/ theorem corollary_6_3_left : LeftCorollary_6_3.{u, v} := by intro k _ _ n M _ _ _ hM let ψ : (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ →+* WeylAlg k n := (RingEquiv.opOp (WeylAlg k n)).symm.toRingHom let : Module (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ M := Module.compHom M ψ have hsmul : ∀ (a : WeylAlg k n) (m : M), (MulOpposite.op (MulOpposite.op a) : (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ) • m = a • m := fun _ _ => rfl have hone : ∀ d : (WeylAlg k n)ᵐᵒᵖ, d ≠ 0 → ∃ F R S : (WeylAlg k n)ᵐᵒᵖ, (1 : (WeylAlg k n)ᵐᵒᵖ) = d * R + F * d * S := by intro d hd obtain ⟨R, S, F, h⟩ := leftHandedIdentity k n d.unop (fun h0 => hd (MulOpposite.unop_injective (by rw [h0, MulOpposite.unop_zero]))) refine ⟨MulOpposite.op F, MulOpposite.op R, MulOpposite.op S, ?_⟩ apply MulOpposite.unop_injective simp only [MulOpposite.unop_one, MulOpposite.unop_add, MulOpposite.unop_mul, MulOpposite.unop_op, h, mul_assoc] have : Module.Finite (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ M := by obtain ⟨s, hs⟩ := (Module.Finite.fg_top : (⊤ : Submodule (WeylAlg k n) M).FG) refine ⟨⟨s, ?_⟩⟩ let N : Submodule (WeylAlg k n) M := { carrier := Submodule.span (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ (s : Set M) add_mem' := fun ha hb => Submodule.add_mem _ ha hb zero_mem' := Submodule.zero_mem _ smul_mem' := fun c x hx => by change c • x ∈ Submodule.span (WeylAlg k n)ᵐᵒᵖᵐᵒᵖ (s : Set M) rw [← hsmul] exact Submodule.smul_mem _ _ hx } have hN : (⊤ : Submodule (WeylAlg k n) M) ≤ N := by rw [← hs] exact Submodule.span_le.2 fun x hx => Submodule.subset_span hx exact eq_top_iff.2 fun x _ => hN Submodule.mem_top have hT : Stafford38.Paper.Section2.IsTorsion (WeylAlg k n)ᵐᵒᵖ M := by intro m obtain ⟨d, hd, hdm⟩ := hM m refine ⟨MulOpposite.op d, fun h => hd (MulOpposite.op_injective (by rw [h, MulOpposite.op_zero])), ?_⟩ rw [hsmul, hdm] have hOre := weyl_leftOreCondition k n obtain ⟨m, hm⟩ := (Section6.corollary_6_3_of hOre hone hT ⟨Section6.two_generators_of_identity hOre hone hT, fun l hl => Section6.stably_generated_of_identity hOre hone hT l (by omega)⟩).1 refine ⟨m, fun x => ?_⟩ obtain ⟨a, ha⟩ := hm x exact ⟨a.unop, ha⟩
theorem corollary_6_3_left : LeftCorollary_6_3 := Stafford38.Paper.Section7.corollary_6_3_left
/-- **Proof of Corollary 6.3**, given the cited theorem [Sta78, Thm. 3.7] (`h37`: `T` is generated by two elements and stably two-generated). Take generators `t₁, t₂` (`h37`), `φ(r, s) = t₁r + t₂s` onto `T`, a common annihilator `d = d₁r₁ = d₂r₂` (Ore condition), `1 = dR + FdS`; then `(a, 1 - Fa) ∈ M = ker φ` and `(a, 1 - Fa), (1, -F)` generate `A ⊕ A`, so `T` is generated by `φ(1, -F) = t₁ - t₂F`. The closing remark, stable cyclicity: for `l = 1` choose `f_1 = -F` (`step_6_3_pair`); for `l ≥ 2` it is the stable two-generation of `h37`. -/ theorem corollary_6_3_of (hOre : RightOreCondition A) (hone : ∀ d : A, d ≠ 0 → ∃ F R S : A, (1 : A) = d * R + F * d * S) (htorsion : Stafford38.Paper.Section2.IsTorsion A T) (h37 : (∃ t₁ t₂ : T, Submodule.span Aᵐᵒᵖ ({t₁, t₂} : Set T) = ⊤) ∧ IsStablyGenerated A T 2) : Stafford38.Paper.Section2.IsCyclic A T ∧ IsStablyGenerated A T 1 := by obtain ⟨⟨t₁, t₂, ht⟩, h2⟩ := h37 obtain ⟨d, hd, h₁, h₂⟩ := step_6_3_common_annihilator hOre htorsion t₁ t₂ obtain ⟨F, R, S, hcert⟩ := hone d hd refine ⟨(Stafford38.Paper.Section2.isCyclic_iff_span A T).2 ⟨_, (step_6_3_pair t₁ t₂ d F R S h₁ h₂ hcert).trans ht⟩, ?_⟩ intro l hl a ha rcases Nat.lt_or_ge l 2 with hl2 | hl2 · obtain rfl : l = 1 := by omega obtain ⟨d', hd', h₁', h₂'⟩ := step_6_3_common_annihilator hOre htorsion (a 0) (a 1) obtain ⟨F', R', S', hcert'⟩ := hone d' hd' refine ⟨fun _ => -F', ?_⟩ have hrange : (Set.range fun i : Fin 1 => a i.castSucc + (MulOpposite.op (-F') : Aᵐᵒᵖ) • a (Fin.last 1)) = {a 0 - (MulOpposite.op F' : Aᵐᵒᵖ) • a 1} := by ext z simp [sub_eq_add_neg, Fin.last] exact eq_comm have hpair2 : ({a 0, a 1} : Set T) = Set.range a := by ext z simp [Fin.exists_fin_two, eq_comm] rw [hrange, step_6_3_pair (a 0) (a 1) d' F' R' S' h₁' h₂' hcert', hpair2, ha] · exact h2 l hl2 a ha
/-- `A_n(k)` is a left Ore domain: the right Ore condition of `A_n(k)ᵐᵒᵖ`, obtained from the right Ore condition of `A_n(k)` through the transpose. -/ theorem weyl_leftOreCondition (k : Type u) [Field k] [CharZero k] (n : ℕ) : AlgebraicAnalysis.OreRightIntersection.RightOreCondition (WeylAlg k n)ᵐᵒᵖ := by intro a b ha hb have hne : ∀ c : (WeylAlg k n)ᵐᵒᵖ, c ≠ 0 → transpose k n c.unop ≠ 0 := by intro c hc h apply hc apply MulOpposite.unop_injective rw [MulOpposite.unop_zero] simpa only [transpose_transpose, transpose_zero] using congrArg (transpose k n) h obtain ⟨x, y, hxy, hx0⟩ := Section6.weyl_rightOreCondition (k := k) n (hne a ha) (hne b hb) have hxy' := congrArg (transpose k n) hxy simp only [transpose_mul, transpose_transpose] at hxy' refine ⟨MulOpposite.op (transpose k n x), MulOpposite.op (transpose k n y), ?_, ?_⟩ · apply MulOpposite.unop_injective simpa only [MulOpposite.unop_mul, MulOpposite.unop_op] using hxy' · intro h apply hx0 have h' := congrArg (fun c => transpose k n c.unop) h simpa only [MulOpposite.unop_mul, MulOpposite.unop_op, MulOpposite.unop_zero, transpose_mul, transpose_transpose, transpose_zero] using h'
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
Comparison notes
The paragraph adds that the left-handed identity gives Corollary 6.3 for left torsion modules. The paper states this consequence without a printed proof. Stafford38Challenge.LeftHandedIdentity (Challenge.lean, Mathlib-only) states the identity: for every field k of characteristic zero, every n and every nonzero d ∈ A_n(k) there are R, S, F with 1 = Rd + SdF. Section7.leftHandedIdentity proves it from Theorem 1.1 through the formal adjoint (LeftHandedCorollary.leftHanded_of_universalStatement). Stafford38Challenge.LeftCorollary_6_3 states that every finitely generated torsion left A_n(k)-module is cyclic. Section7.corollary_6_3_left proves it by the printed proof of Corollary 6.3 (Section6.corollary_6_3_of) for the ring A_n(k)ᵐᵒᵖ, whose right modules are the left A_n(k)-modules, with 1 = Rd + SdF in place of 1 = dR + FdS and the left Ore condition (Section7.weyl_leftOreCondition, from the right Ore condition through the transpose); the two generators and the stable two-generation come from the same identity. Solution.lean proves both compared theorems through these declarations.
Review notes
Corollary — Differential operators over polynomial localizations
formal adjoint, which implies Corollary 6.3 also for left torsion modules; the identity for finite-order differential operators on localizations of , including principal opens, Laurent-polynomial coefficients and the fraction field ; and, independently
/-- **Section 7, localizations** (tex lines 1630–1631): the identity for finite-order differential operators on every localization `B = S⁻¹k[x_1, …, x_n]`. -/ def LocalizedDifferentialIdentity : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (S : Submonoid (MvPolynomial (Fin n) k)) (B : Type v) [CommRing B] [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B] [IsScalarTower k (MvPolynomial (Fin n) k) B] [IsLocalization S B], HasDifferentialIdentity k B
/-- **Localizations.** For every localization `B` of `k[x_1, …, x_n]`, every nonzero finite-order differential operator `d` on `B` has finite-order `F, R, S` with `1 = dR + FdS`. Theorem 1.1 transfers to `D_k(S⁻¹k[x])` by clearing a left denominator of each operator (`s38_localized_differential`, for `Localization S`) and then along the isomorphism `Localization S ≅ B`. -/ theorem localizedDifferentialIdentity : LocalizedDifferentialIdentity.{u, v} := by intro k _ _ n S B _ _ _ _ _ have h := hasDifferentialIdentity_of (Stafford38.LocalizedDifferentialCorollaries.s38_localized_differential S (Localization S) fun d hd => Stafford38.Paper.theorem_1_1 k n d hd) exact hasDifferentialIdentity_of_algEquiv ((IsLocalization.algEquiv S (Localization S) B).restrictScalars k) h
More Lean proofs and definitions
/-- The identity `1 = dR + FdS` in the ring `D_k(B)` of finite-order `k`-linear differential operators on `B`: for every nonzero `d ∈ D_k(B)` there are `F, R, S ∈ D_k(B)` with `1 = dR + FdS`. -/ def HasDifferentialIdentity (k : Type u) (B : Type v) [CommRing k] [CommRing B] [Algebra k B] : Prop := ∀ d : Module.End k B, IsDifferentialOperator k B d → d ≠ 0 → ∃ F R S : Module.End k B, IsDifferentialOperator k B F ∧ IsDifferentialOperator k B R ∧ IsDifferentialOperator k B S ∧ (1 : Module.End k B) = d * R + F * d * S
/-- `P` is a `k`-linear differential operator of finite order on `B`. -/ def IsDifferentialOperator (k : Type u) (B : Type v) [CommRing k] [CommRing B] [Algebra k B] (P : Module.End k B) : Prop := ∃ N, IsDifferentialOperatorOfOrder k B N P
/-- `P` is a `k`-linear differential operator of order at most `N` on the commutative `k`-algebra `B` (Grothendieck): every iterated commutator `[…[[P, b₀], b₁], …, b_N]` with multiplication operators vanishes, where `[Q, b] = Q ∘ b − b ∘ Q`. -/ def IsDifferentialOperatorOfOrder (k : Type u) (B : Type v) [CommRing k] [CommRing B] [Algebra k B] (N : ℕ) (P : Module.End k B) : Prop := ∀ b : List B, b.length = N + 1 → b.foldl (fun Q c => Q * LinearMap.mulLeft k c - LinearMap.mulLeft k c * Q) P = 0
theorem localizedDifferentialIdentity : LocalizedDifferentialIdentity := Stafford38.Paper.Section7.localizedDifferentialIdentity
theorem principalOpenDifferentialIdentity : ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (f : MvPolynomial (Fin n) k), HasDifferentialIdentity k (Localization.Away f) := by intro k _ _ n f exact localizedDifferentialIdentity k n (Submonoid.powers f) (Localization.Away f)
theorem laurentDifferentialIdentity : ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ), HasDifferentialIdentity k (Localization.Away (∏ i, MvPolynomial.X i : MvPolynomial (Fin n) k)) := by intro k _ _ n exact principalOpenDifferentialIdentity k n _
theorem fractionFieldDifferentialIdentity : ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ), HasDifferentialIdentity k (FractionRing (MvPolynomial (Fin n) k)) := by intro k _ _ n exact localizedDifferentialIdentity k n (nonZeroDivisors _) (FractionRing _)
theorem s38_unconditional_localized_differential (S : Submonoid (A (k := k) (n := n))) (B : Type u) [CommRing B] [Algebra (A (k := k) (n := n)) B] [Algebra k B] [IsScalarTower k (A (k := k) (n := n)) B] [IsLocalization S B] : S38 (D (k := k) B) := by apply s38_localized_differential S B intro d hd exact Stafford38.universalStatement (k := k) n d hd
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.DifferentialOperatorsSource ↗open Stafford38.LocalizedWeylActionSource ↗open Stafford38.LocalizedDifferentialClearingSource ↗open Stafford38.LocalizationCorollariesSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable {k : Type u} [Field k] [CharZero k] {n : ℕ}Source ↗
variable (S : Submonoid (MvPolynomial (Fin n) k))Source ↗
variable (B : Type u) [CommRing B]Source ↗
variable [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B]Source ↗
variable [IsScalarTower k (MvPolynomial (Fin n) k) B]Source ↗
variable [IsLocalization S B]
theorem s38_localized_differential (hS : S38 (PresentedWeyl k n)) : S38 (D (k := k) B) := by apply s38_of_leftUnitClearing hS intro q hq obtain ⟨s, a, ha⟩ := localized_differential_left_denominator_clearing S B q let b : B := algebraMap (A (k := k) (n := n)) B (s : A) have hb : IsUnit b := IsLocalization.map_units B s refine ⟨a, multiplicationD (k := k) (B := B) b, multiplicationD_isUnit (k := k) (B := B) hb, ?_⟩ apply Subtype.ext change multiplication (k := k) b * (q : Module.End k B) = weylEnd S B a exact ha
Variables and notation
Source ↗open StaffordSource ↗open Stafford38.DifferentialOperatorsSource ↗open Stafford38.LocalizedWeylActionSource ↗open Stafford38.LocalizedDifferentialClearingSource ↗open Stafford38.LocalizationCorollariesSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable {k : Type u} [Field k] [CharZero k] {n : ℕ}Source ↗
variable (S : Submonoid (MvPolynomial (Fin n) k))Source ↗
variable (B : Type u) [CommRing B]Source ↗
variable [Algebra (MvPolynomial (Fin n) k) B] [Algebra k B]Source ↗
variable [IsScalarTower k (MvPolynomial (Fin n) k) B]Source ↗
variable [IsLocalization S B]Source ↗
include S in/-- The iterated-commutator order of `Stafford38Challenge` is the recursive order filtration `order N` of `AlgebraicAnalysis`. -/ theorem isDifferentialOperatorOfOrder_iff (N : ℕ) (P : Module.End k B) : IsDifferentialOperatorOfOrder k B N P ↔ P ∈ order (k := k) (R := B) N := by induction N generalizing P with | zero => rw [mem_order_zero_iff] constructor · intro h a exact h [a] rfl · intro h b hb match b, hb with | [a], _ => exact h a | succ N ih => rw [mem_order_succ_iff] constructor · intro h a rw [← ih] intro b hb exact h (a :: b) (by simp [hb]) · intro h b hb match b, hb with | a :: b, hb => exact (ih _).2 (h a) b (by simpa using hb)
theorem isDifferentialOperator_iff (P : Module.End k B) : IsDifferentialOperator k B P ↔ P ∈ algebra (k := k) (R := B) := by simp only [IsDifferentialOperator, isDifferentialOperatorOfOrder_iff, mem_algebra_iff]
/-- The identity in the subalgebra `D_k(B)` gives the Mathlib-only form. -/ theorem hasDifferentialIdentity_of (h : AlgebraicAnalysis.TwoGeneratorIdentity (algebra (k := k) (R := B))) : HasDifferentialIdentity k B := by intro d hd hd0 obtain ⟨F, R, S, hone⟩ := h ⟨d, (isDifferentialOperator_iff d).1 hd⟩ (fun h0 => hd0 (congrArg Subtype.val h0)) refine ⟨F, R, S, (isDifferentialOperator_iff _).2 F.2, (isDifferentialOperator_iff _).2 R.2, (isDifferentialOperator_iff _).2 S.2, ?_⟩ exact congrArg Subtype.val hone
/-- The identity in `D_k(B')` transfers along a `k`-algebra isomorphism `B' ≅ B`. -/ theorem hasDifferentialIdentity_of_algEquiv (e : B' ≃ₐ[k] B) (h : HasDifferentialIdentity k B') : HasDifferentialIdentity k B := by intro d hd hd0 let P := conjEnd e.symm d have hP : conjEnd e P = d := by ext x simp [P] have hP0 : P ≠ 0 := fun h0 => hd0 (by rw [← hP, h0, map_zero]) obtain ⟨F, R, S, hF, hR, hS, hone⟩ := h P (isDifferentialOperator_conjEnd e.symm hd) hP0 refine ⟨conjEnd e F, conjEnd e R, conjEnd e S, isDifferentialOperator_conjEnd e hF, isDifferentialOperator_conjEnd e hR, isDifferentialOperator_conjEnd e hS, ?_⟩ have := congrArg (conjEnd e) hone simp only [map_one, map_add, map_mul, hP] at this exact this
/-- Conjugation `P ↦ e ∘ P ∘ e⁻¹` of `k`-linear endomorphisms along `e`. -/ abbrev conjEnd (e : B' ≃ₐ[k] B) : Module.End k B' ≃+* Module.End k B := LinearEquiv.conjRingEquiv e.toLinearEquiv
theorem conjEnd_apply (e : B' ≃ₐ[k] B) (P : Module.End k B') (x : B) : conjEnd e P x = e (P (e.symm x)) := rfl
theorem conjEnd_mulLeft (e : B' ≃ₐ[k] B) (b : B') : conjEnd e (LinearMap.mulLeft k b) = LinearMap.mulLeft k (e b) := by ext x simp
theorem foldl_conjEnd (e : B' ≃ₐ[k] B) : ∀ (l : List B') (P : Module.End k B'), (l.map e).foldl (fun Q c => Q * LinearMap.mulLeft k c - LinearMap.mulLeft k c * Q) (conjEnd e P) = conjEnd e (l.foldl (fun Q c => Q * LinearMap.mulLeft k c - LinearMap.mulLeft k c * Q) P) | [], _ => rfl | b :: l, P => by rw [List.map_cons, List.foldl_cons, List.foldl_cons, ← foldl_conjEnd e l] congr 1 rw [map_sub, map_mul, map_mul, conjEnd_mulLeft]
theorem isDifferentialOperator_conjEnd (e : B' ≃ₐ[k] B) {P : Module.End k B'} (h : IsDifferentialOperator k B' P) : IsDifferentialOperator k B (conjEnd e P) := by obtain ⟨N, hN⟩ := h refine ⟨N, fun l hl => ?_⟩ have hl' : (l.map e.symm).map e = l := by clear hl induction l with | nil => rfl | cons b l ih => simp only [List.map_cons, AlgEquiv.apply_symm_apply, ih] rw [← hl', foldl_conjEnd, hN _ (by simpa using hl), map_zero]
/-- **Theorem 1.1.** For every field `k` of characteristic zero, every `n ≥ 0` and every nonzero `d ∈ A_n(k)` there is `F ∈ A_n(k)` with `A_n(k) = dA_n(k) + FdA_n(k)`, i.e. `1 = dR + FdS` for some `R, S`. -/ theorem theorem_1_1 : Stafford38Challenge.UniversalStatement.{u} := Section6.theorem_1_1_paper
Comparison notes
The paper states this consequence without a printed proof, including principal opens, Laurent-polynomial coefficients and the fraction field.
Stafford38Challenge.LocalizedDifferentialIdentity (Challenge.lean, Mathlib-only) states it for every localization B of k[x_1, …, x_n] at a submonoid S, k a field of characteristic zero: every nonzero k-linear differential operator d of finite order on B (IsDifferentialOperator: the iterated commutators with multiplication operators vanish) has F, R, S of finite order with 1 = dR + FdS (HasDifferentialIdentity). Section7.localizedDifferentialIdentity proves it from Theorem 1.1 by LocalizedDifferentialCorollaries.s38_localized_differential for Localization S and then along the isomorphism Localization S ≅ B (hasDifferentialIdentity_of_algEquiv); isDifferentialOperator_iff identifies this order with the order filtration used there. Solution.lean derives the compared theorems principalOpenDifferentialIdentity (B = k[x]_f), laurentDifferentialIdentity (f = x_1⋯x_n) and fractionFieldDifferentialIdentity (B = k(x_1, …, x_n)) from it.
Review notes
Corollary — Evolutionary version with coefficients commuting with a Weyl pair
Laurent-polynomial coefficients and the fraction field ; and, independently of the geometric part of Section 5, the identity with for , , in any -algebra containing elements with , provided each commutes with and .
/-- **Section 7, evolutionary identity** (tex lines 1631–1634): for `r ≥ 1`, in every `k`-algebra `D` with `px = xp + 1`, the element `d = p^r − ∑_j a_j x^{m_j}`, each `a_j` commuting with `x` and `p`, has `R, S` with `1 = dR + x^r d S`. -/ def EvolutionaryIdentity : Prop := ∀ (k : Type u) [Field k] [CharZero k] (D : Type v) [Ring D] [Algebra k D] (x p : D), p * x = x * p + 1 → ∀ r : ℕ, 1 ≤ r → ∀ (J : ℕ) (a : Fin J → D) (m : Fin J → ℕ), (∀ j, a j * x = x * a j) → (∀ j, a j * p = p * a j) → ∃ R S : D, (1 : D) = (p ^ r - ∑ j, a j * x ^ m j) * R + x ^ r * (p ^ r - ∑ j, a j * x ^ m j) * S
/-- **Evolutionary identity.** For `r ≥ 1` and `px = xp + 1` in a `k`-algebra, `d = p^r - ∑_j a_j x^{m_j}` with each `a_j` commuting with `x` and `p` has `1 = dR + x^r d S`. -/ theorem evolutionaryIdentity : EvolutionaryIdentity.{u, v} := by intro k _ _ D _ _ x p hw r hr J a m hx hp have hmem : ∀ t ∈ List.ofFn (fun j => (a j, m j)), ∃ j, t = (a j, m j) := by intro t ht obtain ⟨j, rfl⟩ := List.mem_ofFn.1 ht exact ⟨j, rfl⟩ obtain ⟨R, S, h⟩ := Stafford38.Evolution.evolutionaryCorollary k hw hr (List.ofFn fun j => (a j, m j)) (fun t ht => by obtain ⟨j, rfl⟩ := hmem t ht; exact hx j) (fun t ht => by obtain ⟨j, rfl⟩ := hmem t ht; exact hp j) rfl exact ⟨R, S, by rw [← potentialSum_ofFn x J a m]; exact h⟩
More Lean proofs and definitions
theorem evolutionaryIdentity : EvolutionaryIdentity := Stafford38.Paper.Section7.evolutionaryIdentity
/-- Exact paper-facing form. No commutativity among the coefficients is assumed: each displayed coefficient need only commute with `x` and `p`. -/ theorem evolutionaryCorollary (k : Type*) [Field k] [CharZero k] [Algebra k D] (hw : p * x = x * p + 1) {r : ℕ} (hr : 0 < r) (terms : List (D × ℕ)) (hx : ∀ t ∈ terms, t.1 * x = x * t.1) (hp : ∀ t ∈ terms, t.1 * p = p * t.1) {d : D} (hd : d = p ^ r - potentialSum x terms) : ∃ R S : D, 1 = d * R + x ^ r * d * S := by obtain ⟨R, S, h⟩ := evolution_stafford k hw hr terms hx hp exact ⟨R, S, by simpa [hd] using h.symm⟩
/-- `potentialSum` over the list of pairs `(a_j, m_j)` is `∑_j a_j x^{m_j}`. -/ theorem potentialSum_ofFn {D : Type v} [Ring D] (x : D) : ∀ (J : ℕ) (a : Fin J → D) (m : Fin J → ℕ), Stafford38.Evolution.potentialSum x (List.ofFn fun j => (a j, m j)) = ∑ j, a j * x ^ m j | 0, _, _ => by simp [Stafford38.Evolution.potentialSum] | J + 1, a, m => by rw [List.ofFn_succ, Fin.sum_univ_succ, Stafford38.Evolution.potentialSum, potentialSum_ofFn x J (fun j => a j.succ) (fun j => m j.succ)]
Comparison notes
The paper states this consequence without a printed proof.
Stafford38Challenge.EvolutionaryIdentity (Challenge.lean, Mathlib-only) states it with a finite sum: for a field k of characteristic zero, a k-algebra D with px = xp + 1, r ≥ 1 and d = p^r − Σ_{j<J} a_j x^{m_j}, each a_j commuting with x and p, there are R, S with 1 = dR + x^r dS. Section7.evolutionaryIdentity proves it from Evolution.evolutionaryCorollary, without Theorem 1.1; potentialSum_ofFn rewrites the sum in the list form of that theorem. The a_j need not commute with one another.
Review notes
Paragraph — Generalization (a): differential operators on smooth affine domains
• Let be a smooth finitely generated -algebra that is a domain,
and let be the ring of -linear differential operators of finite
order on . For every nonzero there are
with [Albert 2026, Albert 2026].
For this is the identity in .
Lean: GlobalStaffordChallenge.universalStatement;
statement GlobalStaffordChallenge.UniversalStatement
/-- **WP-18 terminal declaration**: the Global Stafford theorem, quantified over every characteristic-zero field `k` and every smooth integral affine `k`-algebra `A`. -/ theorem universalStatement : ∀ (k : Type u) [Field k] [CharZero k] (A : Type u) [CommRing A] [IsDomain A] [Algebra k A] [Algebra.Smooth k A], AlgebraicAnalysis.TwoGeneratorIdentity (algebra (k := k) (R := A)) := fun _k _ _ _A _ _ _ _ => GlobalStafford.PhaseI.twoGeneratorIdentity_of_inputs inputs
More Lean proofs and definitions
/-- Global Stafford: for every characteristic-zero field `k`, every smooth integral finitely generated `k`-algebra `A`, and every nonzero differential operator `d` on `A`, there are differential operators `F, R, S` on `A` with `1 = d * R + F * d * S`, the products being compositions of endomorphisms. -/ def UniversalStatement : Prop := ∀ (k : Type u) [Field k] [CharZero k] (A : Type u) [CommRing A] [IsDomain A] [Algebra k A] [Algebra.Smooth k A] (d : Module.End k A), IsDifferentialOperator d → d ≠ 0 → ∃ F R S : Module.End k A, IsDifferentialOperator F ∧ IsDifferentialOperator R ∧ IsDifferentialOperator S ∧ (1 : Module.End k A) = d * R + F * d * S
/-- The two-generator identity in a ring, with the distinguished factor on the left of the first product and in the middle of the second. -/ def TwoGeneratorIdentity (R : Type u) [Ring R] : Prop := ∀ d : R, d ≠ 0 → ∃ F r s : R, (1 : R) = d * r + F * d * s
/-- The algebra of all finite-order `k`-linear differential operators on `R`. -/ def algebra : Subalgebra k (End (k := k) (R := R)) where carrier := {P | ∃ n, P ∈ order n} zero_mem' := ⟨0, (order 0).zero_mem⟩ add_mem' := by rintro P Q ⟨m, hP⟩ ⟨n, hQ⟩ refine ⟨max m n, (order (max m n)).add_mem ?_ ?_⟩ · exact order_mono (Nat.le_max_left _ _) hP · exact order_mono (Nat.le_max_right _ _) hQ mul_mem' := by rintro P Q ⟨m, hP⟩ ⟨n, hQ⟩ exact ⟨m + n, mul_mem_order hP hQ⟩ one_mem' := by refine ⟨0, (mem_order_zero_iff_eq_multiplication _).2 ?_⟩ ext x simp [multiplication_apply] algebraMap_mem' := by intro c refine ⟨0, (mem_order_zero_iff_eq_multiplication _).2 ?_⟩ ext x simp [multiplication_apply]
Comparison notes
Review notes
Paragraph — Generalization (b): filtered simple algebras
• Let be an infinite-dimensional simple -algebra with an
exhaustive filtration by
finite-dimensional subspaces, , whose
associated graded ring is a commutative domain finitely generated over .
For all nonzero and every there are with
[Albert 2026]. The Bernstein filtration makes
such an algebra, and , gives .
Lean: SymplecticStaffordChallenge.universalStatement;
statement SymplecticStaffordChallenge.AffineSameDivisorStatement
/-- Target theorem A: the affine same-divisor identity in the original algebra. -/ def AffineSameDivisorStatement : Prop := ∀ (k : Type u) (U : Type v) (G : Type w) [Field k] [CharZero k] [Ring U] [Algebra k U] [IsSimpleRing U] [CommRing G] [IsDomain G] [Algebra k G] [Algebra.FiniteType k G] (𝒜 : ℕ → Submodule k G) [GradedAlgebra 𝒜], ¬ FiniteDimensional k U → ringKrullDim G < ⊤ → FilteredAlgebraData k U G 𝒜 → ∀ d a : U, d ≠ 0 → a ≠ 0 → ∀ B : U, ∃ H R S : U, 1 = d * R + (B + H * a) * d * S
theorem universalStatement : AffineSameDivisorStatement.{u, v, w} := by intro k U G _ _ _ _ _ _ _ _ _ 𝒜 _ _ hdim D d a hd ha B have : Infinite k := Infinite.of_injective (Nat.cast : ℕ → k) Nat.cast_injective have : IsNoetherianRing G := Algebra.FiniteType.isNoetherianRing k G let D' : SymplecticStafford.Paper.FilteredAlgebraData k U G 𝒜 := { F := D.F mono := D.mono exhaustive := D.exhaustive zero_piece := D.zero_piece finiteDimensional := D.finiteDimensional mul_mem := D.mul_mem gr := D.gr gr_surjective := D.gr_surjective gr_zero_eq_zero_iff := D.gr_zero_eq_zero_iff gr_succ_eq_zero_iff := D.gr_succ_eq_zero_iff gr_mul := D.gr_mul } exact D'.targetA hdim d a hd ha B
More Lean proofs and definitions
/-- Target theorem A: the affine same-divisor identity in the original algebra. -/ def AffineSameDivisorStatement : Prop := ∀ (k : Type u) (U : Type v) (G : Type w) [Field k] [CharZero k] [Ring U] [Algebra k U] [IsSimpleRing U] [CommRing G] [IsDomain G] [Algebra k G] [Algebra.FiniteType k G] (𝒜 : ℕ → Submodule k G) [GradedAlgebra 𝒜], ¬ FiniteDimensional k U → ringKrullDim G < ⊤ → FilteredAlgebraData k U G 𝒜 → ∀ d a : U, d ≠ 0 → a ≠ 0 → ∀ B : U, ∃ H R S : U, 1 = d * R + (B + H * a) * d * S
Comparison notes
Lean states the identity 1 = dR + (B + Ha)dS for a simple, infinite-dimensional k-algebra U with an exhaustive, multiplicative filtration by finite-dimensional pieces, F₀U = k, whose associated graded ring is a commutative, finitely generated domain encoded by graded symbol maps. Lean additionally assumes finite Krull dimension of the graded ring, which holds for every finitely generated commutative k-algebra. The paper gives no printed proof and cites the result as in preparation.
Review notes
Paragraph — Open questions and a terminating search
The proof of Theorem 1.1 shows without producing and . Over a computable field, the fixed choice of nevertheless gives a terminating algorithm: enumerate the countably many pairs of linear symplectic changes of generators and degree bounds, and solve the linear system with for of Bernstein degree at most . Theorem 1.1 guarantees that some pair succeeds. We cannot provide bounds for and in terms of , so the question for an efficient algorithm remains open. In (a) and (b), is not fixed, so this search does not apply.
No corresponding declaration.
Comparison notes
Discussion, not a theorem. The fixed choice F = Φ_g⁻¹(x₁)^{deg_B d} gives a terminating search over computable fields; bounds and efficiency remain open.
Review notes
A. Appendix A: AI-assisted research and formalization
Appendix A — AI-assisted research, formalization and verification provenance
AI-assisted research and formalization
AI agents assisted with proof discovery, counterexample searches, review, formalization and writing. The work began with computations of torsion generators in Macaulay2 and Maple. Direct attempts to construct universal cofactors produced special cases and reductions, but no general construction. The decisive step was to study the quotient and to show that its characteristic support is empty.
The first complete argument was written by AI agents in the style of a paper proof. Agents then formalized it in Lean, and the formalization changed parts of the argument where the original steps could not be checked as written. After the Lean proof was complete and published, the paper proof was rewritten from the checked Lean proof, so that every statement corresponds to a Lean declaration, and the Lean proof was then again aligned with the printed proof step by step. The authors reviewed this correspondence and substantially revised the text for clarity and readability together with the agents, until neither the human review nor the AI checks raised further objections.
For finding the first complete argument, we proceeded as follows: One controlling agent kept the authoritative state of the proof. Parallel agents worked on separate approaches and returned evidence together with its scope, and independent reviewer agents checked each result before it was accepted. The first author steered the work with short instructions, and the agents wrote the detailed task prompts for each other. Advice from consultations counted only after an independent check.
The working rules developed together with the proof into a reusable agent skill for open mathematical problems. Early versions emphasized falsification and keeping conditional results apart from established ones. Once a candidate proof existed, the emphasis moved to repairing the first invalid step of a promising argument, and to auditing proofs.
Research and review used OpenAI GPT models from GPT-5.5 to GPT-6.1, Anthropic Claude models from Haiku 4.5 to Opus 5, and locally hosted Qwen and DeepSeek models. GPT and Claude models, most recently Claude Opus 5.5, helped draft and revise the manuscript and the supplementary material. The authors are responsible for the final text.
The formal proof uses Lean 4, Mathlib and AlgebraicAnalysis v0.3.3 [Moura 2021, Community 2020, Albert 2026]. Conditional results were kept apart from proved ones until every hypothesis was discharged. Comparator checked that the proved theorems have exactly the statements of separate challenge files that use only Mathlib definitions, and both the Lean kernel and the independent checker NanoDa accepted the proofs [contributors, NanoDa, Registry]. Theorem 1.1 and generalization (a) of Section 7 are registered in the Palomar registry [Albert 2026, Albert 2026], and their Lean sources are archived [Albert 2026, Albert 2026].
Comparison notes
Review notes
Definitions and references
Challenge definitions
theorem universalStatement : UniversalStatement := Stafford38.Paper.theorem_1_1
theorem universalFixedSourceStatement : UniversalFixedSourceStatement := Stafford38.Paper.theorem_1_1_fixedSource
Definitions
Stafford38Challenge.PhaseVar
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
Stafford38Challenge.relation
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The Weyl commutator relation for the standard symplectic matrix. -/ abbrev relation {k : Type*} [Field k] {n : ℕ} (omega : Matrix (PhaseVar n) (PhaseVar n) k) (a b : FreeAlgebra k (PhaseVar n)) : Prop := Stafford.freeWeylRelation omega a b
Stafford38Challenge.WeylAlg
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
Stafford38Challenge.UniversalStatement
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Stafford's Conjecture 3.8 in its intended nonzero form. -/ def UniversalStatement : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (d : WeylAlg k n), d ≠ 0 → Stafford.Reduction.Stafford38 d
Stafford38Challenge.generator
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
Stafford38FixedSourceChallenge.oldIndex
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
Stafford38FixedSourceChallenge.freeOldMap
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Insert an old free-algebra word into the next-rank free algebra. -/ def freeOldMap (k : Type u) [Field k] (n : ℕ) : FreeAlgebra k (PhaseVar n) →ₐ[k] FreeAlgebra k (PhaseVar (n + 1)) := FreeAlgebra.lift k (fun i => FreeAlgebra.ι k (oldIndex i))
Stafford38FixedSourceChallenge.freeOrderedMonomial
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Ordered PBW words, with old pairs followed by the newest pair. -/ def freeOrderedMonomial (k : Type u) [Field k] : (n : ℕ) → (Fin n → ℕ) → (Fin n → ℕ) → FreeAlgebra k (PhaseVar n) | 0, _, _ => 1 | n + 1, a, p => freeOldMap k n (freeOrderedMonomial k n (fun i => a i.succ) (fun i => p i.succ)) * FreeAlgebra.ι k (.inl (0 : Fin (n + 1))) ^ a 0 * FreeAlgebra.ι k (.inr (0 : Fin (n + 1))) ^ p 0
Stafford38FixedSourceChallenge.orderedMonomial
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The image of an ordered PBW word in the shared Weyl quotient. -/ def orderedMonomial (k : Type u) [Field k] (n : ℕ) (a p : Fin n → ℕ) : WeylAlg k n := RingQuot.mkAlgHom k (relation (k := k) (n := n) (Matrix.J (Fin n) k)) (freeOrderedMonomial k n a p)
Stafford38FixedSourceChallenge.phaseDegree
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Total degree of a split coordinate/momentum exponent. -/ def phaseDegree {n : ℕ} (a p : Fin n → ℕ) : ℕ := (∑ i, a i) + ∑ i, p i
Stafford38FixedSourceChallenge.bernsteinPiece
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Span of the ordered Weyl words of total degree at most `N`. -/ def bernsteinPiece (k : Type u) [Field k] (n N : ℕ) : Submodule k (WeylAlg k n) := Submodule.span k {z | ∃ a p : Fin n → ℕ, phaseDegree a p ≤ N ∧ z = orderedMonomial k n a p}
Stafford38FixedSourceChallenge.bernsteinDegree
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The intrinsic Bernstein degree of a presented element: its least piece. -/ noncomputable def bernsteinDegree (k : Type u) [Field k] {n : ℕ} (d : WeylAlg k n) : ℕ := sInf {N : ℕ | d ∈ bernsteinPiece k n N}
Stafford.linearCombination
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Linear combination of an indexed family by the rows of a matrix. -/ def linearCombination (M : Matrix ι ι k) (z : ι → A) (i : ι) : A := ∑ j, algebraMap k A (M i j) * z j
Stafford38Challenge.standardForm
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
Stafford38FixedSourceChallenge.IsLinearWeylCoordinate
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- A source coordinate obtained from an invertible symplectic linear change of the canonical Weyl generators. -/ def IsLinearWeylCoordinate (k : Type u) [Field k] (n : ℕ) (ell : WeylAlg k (n + 1)) : Prop := ∃ (M N : Matrix (PhaseVar (n + 1)) (PhaseVar (n + 1)) k) (_hM : M * standardForm k (n + 1) * Matrix.transpose M = standardForm k (n + 1)) (_hN : N * standardForm k (n + 1) * Matrix.transpose N = standardForm k (n + 1)) (_hMN : M * N = 1) (_hNM : N * M = 1), ell = linearCombination k N (generator k (n + 1)) (.inl (0 : Fin (n + 1)))
Stafford38FixedSourceChallenge.UniversalFixedSourceStatement
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The exact fixed-source challenge, using the intrinsic degree. -/ abbrev UniversalFixedSourceStatement : Prop := UniversalFixedSourceStatementWithDegree.{u} bernsteinDegree
Stafford38.TorsionCyclicity.IsRightTorsion
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Every element of a right module is killed by some nonzero right scalar. -/ def IsRightTorsion {A : Type u} {M : Type v} [Ring A] [AddCommGroup M] [Module Aᵐᵒᵖ M] : Prop := ∀ m : M, ∃ d : A, d ≠ 0 ∧ (MulOpposite.op d : Aᵐᵒᵖ) • m = 0
Stafford38CorollaryChallenge.torsionCyclicStatement
Defined in This project (Stafford38) · CorollaryChallenge.lean
/-- Every finitely generated torsion right module over a characteristic-zero Weyl algebra is generated by one element. -/ def torsionCyclicStatement : Prop := ∀ (k : Type u) [Field k] [CharZero k] (n : ℕ) (M : Type v) [AddCommGroup M] [Module (WeylAlg k n)ᵐᵒᵖ M] [Module.Finite (WeylAlg k n)ᵐᵒᵖ M], IsRightTorsion (A := WeylAlg k n) (M := M) → ∃ z : M, Submodule.span (WeylAlg k n)ᵐᵒᵖ ({z} : Set M) = ⊤
Field — Commutative field axioms, including nontriviality via its parents
Defined in Mathlib · Mathlib/Algebra/Field/Defs.lean
/-- A `Field` is a `CommRing` with multiplicative inverses for nonzero elements. An instance of `Field K` includes maps `ratCast : ℚ → K` and `qsmul : ℚ → K → K`. Those two fields are needed to implement the `DivisionRing K → Algebra ℚ K` instance since we need to control the specific definitions for some special cases of `K` (in particular `K = ℚ` itself). See also note [forgetful inheritance]. If the field has positive characteristic `p`, our division by zero convention forces `ratCast (1 / p) = 1 / 0 = 0`. -/ @[stacks 09FD "first part"] class Field (K : Type u) extends CommRing K, DivisionRing K -- see Note [lower instance priority]
DivisionRing — Division, inverse and nontrivial ring axioms
Defined in Mathlib · Mathlib/Algebra/Field/Defs.lean
/-- A `DivisionRing` is a `Ring` with multiplicative inverses for nonzero elements. An instance of `DivisionRing K` includes maps `ratCast : ℚ → K` and `qsmul : ℚ → K → K`. Those two fields are needed to implement the `DivisionRing K → Algebra ℚ K` instance since we need to control the specific definitions for some special cases of `K` (in particular `K = ℚ` itself). See also note [forgetful inheritance]. Similarly, there are maps `nnratCast ℚ≥0 → K` and `nnqsmul : ℚ≥0 → K → K` to implement the `DivisionSemiring K → Algebra ℚ≥0 K` instance. If the division ring has positive characteristic `p`, our division by zero convention forces `ratCast (1 / p) = 1 / 0 = 0`. -/ class DivisionRing (K : Type*) extends Ring K, DivInvMonoid K, Nontrivial K, NNRatCast K, RatCast K where /-- For a nonzero `a`, `a⁻¹` is a right multiplicative inverse. -/ protected mul_inv_cancel : ∀ (a : K), a ≠ 0 → a * a⁻¹ = 1 /-- The inverse of `0` is `0` by convention. -/ protected inv_zero : (0 : K)⁻¹ = 0 protected nnratCast := NNRat.castRec /-- However `NNRat.cast` is defined, it must be equal to `a / b`. Do not use this lemma directly. Use `NNRat.cast_def` instead. -/ protected nnratCast_def (q : ℚ≥0) : (NNRat.cast q : K) = q.num / q.den := by intros; rfl /-- Scalar multiplication by a nonnegative rational number. Unless there is a risk of a `Module ℚ≥0 _` instance diamond, write `nnqsmul := _`. This will set `nnqsmul` to `(NNRat.cast · * ·)` thanks to unification in the default proof of `nnqsmul_def`. Do not use directly. Instead use the `•` notation. -/ protected nnqsmul : ℚ≥0 → K → K /-- However `qsmul` is defined, it must be propositionally equal to multiplication by `Rat.cast`. Do not use this lemma directly. Use `NNRat.smul_def` instead. -/ protected nnqsmul_def (q : ℚ≥0) (a : K) : nnqsmul q a = NNRat.cast q * a := by intros; rfl protected ratCast := Rat.castRec /-- However `Rat.cast q` is defined, it must be propositionally equal to `q.num / q.den`. Do not use this lemma directly. Use `Rat.cast_def` instead. -/ protected ratCast_def (q : ℚ) : (Rat.cast q : K) = q.num / q.den := by intros; rfl /-- Scalar multiplication by a rational number. Unless there is a risk of a `Module ℚ _` instance diamond, write `qsmul := _`. This will set `qsmul` to `(Rat.cast · * ·)` thanks to unification in the default proof of `qsmul_def`. Do not use directly. Instead use the `•` notation. -/ protected qsmul : ℚ → K → K /-- However `qsmul` is defined, it must be propositionally equal to multiplication by `Rat.cast`. Do not use this lemma directly. Use `Rat.cast_def` instead. -/ protected qsmul_def (a : ℚ) (x : K) : qsmul a x = Rat.cast a * x := by intros; rfl -- see Note [lower instance priority]
CommRing — Commutative unital ring axioms
Defined in Mathlib · Mathlib/Algebra/Ring/Defs.lean
/-- A commutative ring is a ring with commutative multiplication. -/ @[wikidata Q858656] class CommRing (α : Type u) extends Ring α, CommMonoid α
CharZero — Injectivity of the natural-number cast
Defined in Mathlib · Mathlib/Algebra/CharZero/Defs.lean
/-- Typeclass for monoids with characteristic zero. (This is usually stated on fields but it makes sense for any additive monoid with 1.) *Warning*: for a semiring `R`, `CharZero R` and `CharP R 0` need not coincide. * `CharZero R` requires an injection `ℕ ↪ R`; * `CharP R 0` asks that only `0 : ℕ` maps to `0 : R` under the map `ℕ → R`. For instance, endowing `{0, 1}` with addition given by `max` (i.e. `1` is absorbing), shows that `CharZero {0, 1}` does not hold and yet `CharP {0, 1} 0` does. This example is formalized in `Counterexamples/CharPZeroNeCharZero.lean`. -/ class CharZero (R) [AddMonoidWithOne R] : Prop where /-- An additive monoid with one has characteristic zero if the canonical map `ℕ → R` is injective. -/ cast_injective : Function.Injective (Nat.cast : ℕ → R)
RingQuot — Universal ring quotient by a generating relation
Defined in Mathlib · Mathlib/Algebra/RingQuot.lean
/-- The quotient of a ring by an arbitrary relation. -/ structure RingQuot (r : R → R → Prop) where toQuot : Quot (RingQuot.Rel r)
FreeAlgebra — Free associative algebra
Defined in Mathlib · Mathlib/Algebra/FreeAlgebra.lean
/-- If `α` is a type, and `R` is a commutative semiring, then `FreeAlgebra R α` is the free (unital, associative) `R`-algebra generated by `α`. This is an `R`-algebra equipped with a function `FreeAlgebra.ι R : α → FreeAlgebra R α` which has the following universal property: if `A` is any `R`-algebra, and `f : α → A` is any function, then this function is the composite of `FreeAlgebra.ι R` and a unique `R`-algebra homomorphism `FreeAlgebra.lift R f : FreeAlgebra R α →ₐ[R] A`. A typical element of `FreeAlgebra R α` is an `R`-linear combination of formal products of elements of `α`. For example if `x` and `y` are terms of type `α` and `a`, `b` are terms of type `R` then `(3 * a * a) • (x * y * x) + (2 * b + 1) • (y * x) + (a * b * b + 3)` is a "typical" element of `FreeAlgebra R α`. In particular if `α` is empty then `FreeAlgebra R α` is isomorphic to `R`, and if `α` has one term `t` then `FreeAlgebra R α` is isomorphic to the polynomial ring `R[t]`. If `α` has two or more terms then `FreeAlgebra R α` is not commutative. One can think of `FreeAlgebra R α` as the free non-commutative polynomial ring with coefficients in `R` and variables indexed by `α`. -/ def FreeAlgebra := Quot (FreeAlgebra.Rel R X)
Variables and notation
Source ↗open scoped MonoidAlgebraSource ↗variable (R X : Type*) [CommSemiring R]
Matrix.J — Standard symplectic matrix; fixes commutator signs
Defined in Mathlib · Mathlib/LinearAlgebra/SymplecticGroup.lean
/-- The matrix defining the canonical skew-symmetric bilinear form. -/ def J : Matrix (l ⊕ l) (l ⊕ l) R := Matrix.fromBlocks 0 (-1) 1 0
Variables and notation
Source ↗open MatrixSource ↗variable {l R : Type*}Source ↗
variable (l) [DecidableEq l] (R) [CommRing R]
Stafford38PaperPairChallenge.pairGeneratorStatement — Supplementary certificate calculation only; the main Challenge and Solution are unchanged.
Defined in This project (Stafford38) · PaperPairChallenge.lean
def pairGeneratorStatement : Prop := ∀ (A : Type u) [Ring A] (M : Type v) [AddCommGroup M] [Module Aᵐᵒᵖ M] (x y : M) (d F R S : A), (MulOpposite.op d : Aᵐᵒᵖ) • x = 0 → (MulOpposite.op d : Aᵐᵒᵖ) • y = 0 → (1 : A) = d * R + F * d * S → Submodule.span Aᵐᵒᵖ ({x - (MulOpposite.op F : Aᵐᵒᵖ) • y} : Set M) = Submodule.span Aᵐᵒᵖ ({x, y} : Set M)
Stafford.FreeWeyl — Generic matrix presentation; WeylAlg specializes this construction.
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The free algebra modulo the commutator relations encoded by `omega`. -/ abbrev FreeWeyl (k : Type*) [Field k] (ι : Type*) (omega : Matrix ι ι k) := RingQuot (freeWeylRelation omega)
Stafford.freeWeylRelation — Relations for the generic matrix presentation.
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The commutator relations presenting a Weyl-type algebra. -/ def freeWeylRelation (omega : Matrix ι ι k) (a b : FreeAlgebra k ι) : Prop := ∃ i j, a = FreeAlgebra.ι k i * FreeAlgebra.ι k j - FreeAlgebra.ι k j * FreeAlgebra.ι k i ∧ b = algebraMap k (FreeAlgebra k ι) (omega i j)
Stafford.freeWeylGenerator — Canonical quotient generator.
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- The canonical generators of the presented Weyl-type algebra. -/ def freeWeylGenerator (omega : Matrix ι ι k) (i : ι) : FreeWeyl k ι omega := RingQuot.mkAlgHom k (freeWeylRelation omega) (FreeAlgebra.ι k i)
Stafford.Reduction.Stafford38 — The certificate predicate for one element.
Defined in This project (Stafford38) · Stafford38/ChallengeDefinitions.lean
/-- Stafford 3.8 for one element, in written operator order. -/ def Stafford38 (e : A) : Prop := ∃ F R S : A, (1 : A) = e * R + F * e * S
Stafford38.WeylFiltration.bernsteinWeight — Weight used for the computational degree.
Defined in This project (Stafford38) · Stafford38/Weyl/Filtration.lean
/-- Every named generator has Bernstein weight one. -/ def bernsteinWeight {n : ℕ} (_ : PhaseVar n) : ℕ := 1
MvPolynomial.weightedTotalDegree — Imported from Mathlib.
Defined in Mathlib · Mathlib/RingTheory/MvPolynomial/WeightedHomogeneous.lean
/-- When `M` has a `⊥` element, we can define the weighted total degree of a multivariate polynomial as a function taking values in `M`. -/ def weightedTotalDegree (w : σ → M) (p : MvPolynomial σ R) : M := p.support.sup fun s => weight w s
Variables and notation
Source ↗open Set Function Finset Finsupp AddMonoidAlgebraSource ↗variable {R M : Type*} [CommSemiring R]Source ↗
variable {σ : Type*}Source ↗
variable [AddCommMonoid M]Source ↗
variable [SemilatticeSup M]Source ↗
variable [OrderBot M]
Finsupp.weight — Imported from Mathlib.
Defined in Mathlib · Mathlib/Data/Finsupp/Weight.lean
/-- The `weight` of the finitely supported function `f : σ →₀ R` with respect to `w : σ → M` is the sum `∑ i, f i • w i`. -/ noncomputable def weight : (σ →₀ R) →+ M := (Finsupp.linearCombination R w).toAddMonoidHom
AlgebraicAnalysis.ringCommutator — Imported from AlgebraicAnalysis.
Defined in AlgebraicAnalysis library · AlgebraicAnalysis/Commutator.lean
/-- The ring commutator, with the written multiplication order retained. -/ def ringCommutator (u v : A) : A := u * v - v * u
Stafford38.WeylPBW.presentedNormalFormLinearEquiv — PBW normal form used to compute the intrinsic degree.
Defined in This project (Stafford38) · Stafford38/Weyl/PBW.lean
/-- Normal-form coordinates for the quotient presentation, obtained by transport through the presented/iterated algebra equivalence. -/ def presentedNormalFormLinearEquiv (n : Nat) : PresentedWeyl k n ≃ₗ[k] SymbolRing k n := (presentedIteratedEquiv k n).toLinearEquiv.trans (iteratedNormalFormLinearEquiv k n)
Variables and notation
Source ↗open StaffordSource ↗open AlgebraicAnalysisSource ↗open AlgebraicAnalysis.OreDivisionSource ↗open Stafford38.CharacteristicSource ↗open AlgebraicAnalysis.OreAssociativitySource ↗open Stafford38.OreCoordinateStageSource ↗open Stafford38.OreIteratedPairStageSource ↗open Stafford38.OreLinearNormalFormSource ↗open Stafford38.OrePairStageSource ↗open Stafford38.OreScalarAlgebraSource ↗open Stafford38.WeylIteratedEquivalenceSource ↗universe uSource ↗
variable (k : Type u) [Field k]