Documentation

Hironaka.Resolution.Algebraic.OrderReduction.ClosedEmbedding

Step 2.4 of order reduction: closed embeddings #

Step 2.4 of the proof of [Kol07, Theorem 103] ([Kol07, 104, Step 2.4]) proves clause (3): for a smooth hypersurface τ : Y ↪ X and an ideal sheaf J ⊂ 𝒪_Y, nonzero on every irreducible component of Y, with τ_*(𝒪_Y/J) = 𝒪_X/I, the ideal I contains the local equations of Y and so has order 1, whence I = W(I); for E = ∅, Step 2.1 does nothing and Step 2.2 may take H = Y, so that clause (3) of Theorem 103 follows from clause (3) of [Kol07, Lemma 102]. This module proves that clause for the maximal-contact case, BO^Y_{n,1}(X, I, ∅) = τ_* BMO_{n−1,1}(Y, J, 1, ∅), in the form maxContactCase_bdData_eq_pushforward, following Kollár's three observations.

"I = W(I)." At the mark 1, BO^Y_{n,1}(X, I, ∅) is Step 2 on the tuned triple (X, W_{s(1)}(I), ∅) at the mark s(1); since s(1) = 1 and W_1(I) = I (tuningParam_one, W_tuningParam_one), it is Step 2 on (X, I, ∅) at the mark 1. The transport step2Seq_congr_mark generalises the mark and the tuned ideal and substitutes (maxContactCase_one_eq_step2Seq).

"Step 2.1 does nothing." For E = ∅ the Step 2.1 sequence is nil (card E.ι = 0, step21Seq_val_eq_nil), and the whole Step 2.2 data is generalised over a sequence S₁ with S₁ = nil and substituted (step22_closedEmbedding_aux), so that the Step 2.2 triple becomes (X, I, ∅ + Y) with Y at position 0 (card F_r.ι = 0, nth_append_last), and I = τ_*J transports to it through the identification e of the position-0 member with Y (map_comap_eqToHom).

"(103.3) follows from (102.3)." Step 2.2 at the mark is the boundary-clearing functor at s(1) on the re-tuned triple, again with s(1) = 1 and W_1 = id (functor_seq_congr_mark); for the boundary-clearing data Hironaka.BD.bdData at the mark 1 that is Hironaka.BD.dataFunctor, whose clause (3) is Hironaka.BD.eq_pushforward_of_quotient: τ_* B(Y, J, 1, (F − Y)|_Y + ⊤) on the Step 2.2 triple. Below the mark (max-ord I = 0, so I = 𝒪_X) both sides are empty: maxContactCase is nil by the remark after [Kol07, Theorem 68] and the Step 2.2 functor is nil (step22Functor_seq_of_maxOrd_lt), while clause (3) of Lemma 102 still identifies the pushforward with the empty value.

The globalised form of clause (3), for BO_{n,1} itself, is assembled in Hironaka/Resolution/Algebraic/Stage/ClosedEmbeddingHypersurface.lean.

theorem Hironaka.BO.step2Seq_congr_mark {k : Type u} [Field k] [CharZero k] {n : ℕ} (T : AlgebraicGeometry.Triple k) (hn : T.HasDimLE n) (bd : (m j : ℕ) → BDData n m j) {H : T.X.left.IdealSheafData} (hH : AlgebraicGeometry.IsSmoothDivisor H) (s : ℕ) (hs : s = 1) (I' : T.X.left.IdealSheafData) (hI' : I' = T.I) (hne' : I'.IsNonzeroEverywhere) (hn' : { X := T.X, smoothOfRelativeDimension := ⋯, I := I', isNonzeroEverywhere := hne', E := T.E, isSnc := ⋯ }.HasDimLE n) (hmax' : I'.maxOrd ≤ ↑s) (hs1 : 1 ≤ s) (hle' : AlgebraicGeometry.Scheme.IdealSheafData.IsMaximalContact (T.X.left ↘ AlgebraicGeometry.Spec ↧k) I' s H) (hmax : T.I.maxOrd ≤ ↑1) (hle : AlgebraicGeometry.Scheme.IdealSheafData.IsMaximalContact (T.X.left ↘ AlgebraicGeometry.Spec ↧k) T.I 1 H) :
step2Seq { X := T.X, smoothOfRelativeDimension := ⋯, I := I', isNonzeroEverywhere := hne', E := T.E, isSnc := ⋯ } hn' hmax' bd hs1 hH hle' = step2Seq T hn hmax bd ⋯ hH hle

Transport of Step 2 along the identification of the mark and of the tuned ideal at m = 1: s(1) = 1 and W_1(I) = I are propositional equalities, so the mark and the ideal are generalised and substituted (structure eta identifies {T with I := T.I} with T).

At the mark 1, BO^H_{n,1}(X, I, E) is Step 2 on (X, I, E) itself ([Kol07, 104, Step 2.4]: "in particular, I = W(I)"): the tuning W_{s(1)}(I) = W_1(I) = I is the identity (tuningParam_one, W_tuningParam_one).

theorem Hironaka.BO.functor_seq_congr_mark {k : Type u} [Field k] [CharZero k] {n : ℕ} (bd : (m j : ℕ) → BDData n m j) (j : ℕ) (T₂ : AlgebraicGeometry.Triple k) (s : ℕ) (hs : s = 1) (I' : T₂.X.left.IdealSheafData) (hI' : I' = T₂.I) (hne' : I'.IsNonzeroEverywhere) (hT' : AlgebraicGeometry.Triple.BDClass n s j { X := T₂.X, smoothOfRelativeDimension := ⋯, I := I', isNonzeroEverywhere := hne', E := T₂.E, isSnc := ⋯ }) (hT : AlgebraicGeometry.Triple.BDClass n 1 j T₂) :
((bd s j).functor k).seq { X := T₂.X, smoothOfRelativeDimension := ⋯, I := I', isNonzeroEverywhere := hne', E := T₂.E, isSnc := ⋯ } hT' = ((bd 1 j).functor k).seq T₂ hT

Transport of the boundary-clearing functor's value along the identification of the mark and of the tuned ideal at m = 1 (the re-tuning inside Step 2.2).

I = τ_*J transported along an identification of the hypersurface: for A = B and J on V(B), (J ∘ eqToHom) on V(A) pushes forward to the same ideal.

theorem Hironaka.BO.step21Seq_val_eq_nil {k : Type u} [Field k] [CharZero k] {n : ℕ} (T : AlgebraicGeometry.Triple k) (hn : T.HasDimLE n) (hE : IsEmpty T.E.ι) {m : ℕ} (hmax : T.I.maxOrd ≤ ↑m) (bd : (j : ℕ) → BDData n m j) :

For an empty boundary the Step 2.1 sequence is empty ([Kol07, 104, Step 2.4]: "if E = ∅ then Step 2.1 does nothing").

theorem Hironaka.BO.step22_closedEmbedding_aux {k : Type u} [Field k] [CharZero k] {n : ℕ} (T : AlgebraicGeometry.Triple k) {Y : T.X.left.IdealSheafData} (hE : IsEmpty T.E.ι) (J : Y.subscheme.IdealSheafData) (hJ : J.IsNonzeroEverywhere) (hIJ : T.I = J.map Y.subschemeι) (Dom : (k : Type u) → [inst : Field k] → [CharZero k] → MarkedTriple k → Prop) (B : (k : Type u) → [inst : Field k] → [inst_1 : CharZero k] → OrderGeSeqAssignment k (Dom k)) (hDom : ∀ (m : ℕ) (k : Type u) [inst : Field k] [inst_1 : CharZero k] (T' : MarkedTriple k), T'.HasDimLE (n - 1) → T'.m = IsLocalRing.tuningParam m → Dom k T') (hB : ∀ (k : Type u) [inst : Field k] [inst_1 : CharZero k] (T' : MarkedTriple k) (hT' : Dom k T'), ((B k).endTriple T' hT').I.maxOrd < ↑T'.m) (hsm : ∀ (k : Type u) [inst : Field k] [inst_1 : CharZero k], (B k).CommutesWithSmooth) (hbc : ∀ (k L : Type u) [inst : Field k] [inst_1 : CharZero k] [inst_2 : Field L] [inst_3 : CharZero L] (σ : k →+* L), (B k).CommutesWithBaseChange (B L) σ) (S₁ : T.X.left.BlowUpSequence) (hS₁ : S₁.IsOrderSeq (T.X.left ↘ AlgebraicGeometry.Spec ↧k) T.I T.E 1) (e₀ : S₁ = AlgebraicGeometry.Scheme.BlowUpSequence.nil T.X.left) (hsnc : ((S₁.exceptionalFamily T.E).append (S₁.strictTransformSeq Y (Fin.last S₁.length))).IsSnc) (j₂ : ℕ) (hj₂ : j₂ = 0) (hT₂ : AlgebraicGeometry.Triple.BDClass n 1 j₂ (step22TripleOfSeq T S₁ hS₁ hsnc)) (h₀ : 0 < Fintype.card (step22TripleOfSeq T S₁ hS₁ hsnc).E.ι) (e : ((step22TripleOfSeq T S₁ hS₁ hsnc).E.nth ⟨0, h₀⟩).subscheme = Y.subscheme) (hdim : (BD.hypersurfaceTriple (step22TripleOfSeq T S₁ hS₁ hsnc) h₀ (J.comap (CategoryTheory.eqToHom e)) ⋯).HasDimLE (n - 1)) :
(step22Functor (fun (m j : ℕ) => BD.bdData n m j Dom B ⋯ hB hsm hbc) 1 j₂ ⋯).seq (step22TripleOfSeq T S₁ hS₁ hsnc) hT₂ = ((B k).seq (BD.hypersurfaceTriple (step22TripleOfSeq T S₁ hS₁ hsnc) h₀ (J.comap (CategoryTheory.eqToHom e)) ⋯) ⋯).pushforward ((step22TripleOfSeq T S₁ hS₁ hsnc).E.nth ⟨0, h₀⟩).subschemeι

Generalised over the Step 2.1 sequence: for S₁ = nil, Step 2.2 with the boundary-clearing data at the mark 1 on the Step 2.2 triple of Y is the pushforward τ_* B(Y_r, J, 1, (F_r − Y_r)|_{Y_r} + ⊤) of clause (3) of [Kol07, Lemma 102] ([Kol07, 104, Step 2.4]: "in Step 2.2 we can choose H = Y; thus (103.3) follows from (102.3)"): at the mark through the re-tuning identity (functor_seq_congr_mark) and Hironaka.BD.eq_pushforward_of_quotient; below the mark both sides are empty.

theorem Hironaka.BO.step2Seq_eq_nil_aux {k : Type u} [Field k] [CharZero k] {n : ℕ} (T : AlgebraicGeometry.Triple k) {Y : T.X.left.IdealSheafData} (bd : (m j : ℕ) → BDData n m j) (S₁ : T.X.left.BlowUpSequence) (hS₁ : S₁.IsOrderSeq (T.X.left ↘ AlgebraicGeometry.Spec ↧k) T.I T.E 1) (e₀ : S₁ = AlgebraicGeometry.Scheme.BlowUpSequence.nil T.X.left) (hsnc : ((S₁.exceptionalFamily T.E).append (S₁.strictTransformSeq Y (Fin.last S₁.length))).IsSnc) (j₂ : ℕ) (hT₂ : AlgebraicGeometry.Triple.BDClass n 1 j₂ (step22TripleOfSeq T S₁ hS₁ hsnc)) (hlt : T.I.maxOrd < ↑1) :

Step 2 is empty below the mark for an empty boundary: Step 2.1 is nil and Step 2.2 is nil by the remark after [Kol07, Theorem 68].

theorem Hironaka.BO.maxContactCase_bdData_eq_pushforward {k : Type u} [Field k] [CharZero k] {n : ℕ} (T : AlgebraicGeometry.Triple k) {Y : T.X.left.IdealSheafData} (hY : AlgebraicGeometry.IsSmoothDivisor Y) (J : Y.subscheme.IdealSheafData) (hJ : J.IsNonzeroEverywhere) (hIJ : T.I = J.map Y.subschemeι) (hn : T.HasDimLE n) (hE : IsEmpty T.E.ι) (Dom : (k : Type u) → [inst : Field k] → [CharZero k] → MarkedTriple k → Prop) (B : (k : Type u) → [inst : Field k] → [inst_1 : CharZero k] → OrderGeSeqAssignment k (Dom k)) (hDom : ∀ (m : ℕ) (k : Type u) [inst : Field k] [inst_1 : CharZero k] (T' : MarkedTriple k), T'.HasDimLE (n - 1) → T'.m = IsLocalRing.tuningParam m → Dom k T') (hB : ∀ (k : Type u) [inst : Field k] [inst_1 : CharZero k] (T' : MarkedTriple k) (hT' : Dom k T'), ((B k).endTriple T' hT').I.maxOrd < ↑T'.m) (hsm : ∀ (k : Type u) [inst : Field k] [inst_1 : CharZero k], (B k).CommutesWithSmooth) (hbc : ∀ (k L : Type u) [inst : Field k] [inst_1 : CharZero k] [inst_2 : Field L] [inst_3 : CharZero L] (σ : k →+* L), (B k).CommutesWithBaseChange (B L) σ) :
(maxContactCase T ⋯ hY ⋯ fun (m j : ℕ) => BD.bdData n m j Dom B ⋯ hB hsm hbc) = (↑(step21Seq T hn ⋯ (fun (j : ℕ) => BD.bdData n 1 j Dom B ⋯ hB hsm hbc) (Fintype.card T.E.ι))).concat (((B k).seq (closedEmbeddingTriple T hn ⋯ (fun (m j : ℕ) => BD.bdData n m j Dom B ⋯ hB hsm hbc) hY ⋯ hE J hJ) ⋯).pushforward ((step22Triple T hn ⋯ (fun (m j : ℕ) => BD.bdData n m j Dom B ⋯ hB hsm hbc) ⋯ hY ⋯).E.nth ⟨0, ⋯⟩).subschemeι)

Clause (3) of [Kol07, Theorem 103] in the maximal-contact case ([Kol07, 104, Step 2.4]; clause (3) of [Kol07, Lemma 102]): for the boundary-clearing data bdData built from the inductive input B, E = ∅ and I = τ_*J, BO^Y_{n,1}(X, I, ∅) = τ_* B(Y_r, J, 1, (F_r − Y_r)|_{Y_r} + ⊤), the marked triple of clause (3) on the Step 2.2 triple of Y at the mark 1, Y_r at position 0, J transported, preceded by the empty Step 2.1 sequence.