Documentation

Hironaka.Resolution.Algebraic.Kol07.GoingUp

The going-up theorem for D-balanced ideals #

Kollár's going-up theorem [Kol07, Theorem 84]: for a D-balanced ideal I with max-ord I = m on a smooth variety X and a smooth hypersurface S ⊆ X, the push-forward to X of a smooth blow-up sequence of order ≥ m for (S, I|_S, m) is a smooth blow-up sequence of order m for (X, I). Its proof [Kol07, 90] is by induction on the length: assuming the statement for sequences of length < r, Corollary 89 for the prefix Π_{r−1} and the D-balanced property give cosupp(J_{r−1}, m) ⊆ cosupp(I_{r−1}, m), so the last blow-up also has order ≥ m.

This module proves the theorem with the boundary E of [Kol07, Corollary 85] carried along (IsDBalanced.isOrderSeq_pushforward) and in the printed form at E = ∅ (IsDBalanced.isOrderSeq_pushforward_of_not_subset_cosupp). The induction is run as a strong induction on the stage index: the clauses at the stages < l make the prefix B.take l a sequence of order ≥ m (isOrderGeSeq_take_of_forall_lt, Hironaka/Scheme/BlowUpSequence/Truncate.lean), to which Corollary 89 applies, and the results are carried from the prefix to stage l along the truncation lemmas. Together with the going-down theorem this gives the one-to-one correspondence of Corollary 85 between order reduction on X and on a hypersurface of maximal contact.

Two groups of auxiliary lemmas are proved on the way:

Congruences along an equality of sequences and of morphisms #

theorem Hironaka.Sequence.stage_congr {X : AlgebraicGeometry.Scheme} {S₁ S₂ : X.BlowUpSequence} (e : S₁ = S₂) (n : ℕ) (h₁ : n < S₁.length + 1) (h₂ : n < S₂.length + 1) :
S₁.stage ⟨n, h₁⟩ = S₂.stage ⟨n, h₂⟩
theorem Hironaka.Sequence.markedTransformSeq_congr_heq {X : AlgebraicGeometry.Scheme} {S₁ S₂ : X.BlowUpSequence} (e : S₁ = S₂) (J : X.IdealSheafData) (c n : ℕ) (h₁ : n < S₁.length + 1) (h₂ : n < S₂.length + 1) :
S₁.markedTransformSeq J c ⟨n, h₁⟩ ≍ S₂.markedTransformSeq J c ⟨n, h₂⟩
theorem Hironaka.Sequence.pullbackStageHom_congr_heq {X Y : AlgebraicGeometry.Scheme} {S₁ S₂ : X.BlowUpSequence} (e : S₁ = S₂) (h : Y ⟶ X) (n : ℕ) (h₁ : n < S₁.length + 1) (h₂ : n < S₂.length + 1) :
S₁.pullbackStageHom h ⟨n, h₁⟩ ≍ S₂.pullbackStageHom h ⟨n, h₂⟩

The stage lifts of a pullback depend only on the pulled-back morphism.

Transport of the stage lifts along an eqToHom on the source (the stage-lift form of pullback_eqToHom_comp).

The embeddings of the pushforward are the stage lifts of its restriction #

On the push-forward B = T.pushforward j along a closed immersion [Kol07, 30.3], the stage lift B.pullbackStageHom j n : (B.pullback j)_n ⟶ B_n of the restriction [Kol07, 30.2] is the stage inclusion T.pushforwardStageHom j n : T_n ⟶ B_n; heterogeneously, as B.pullback j = T holds only propositionally.

Corollary 89 on the push-forward #

The centers of a push-forward along S.subschemeι lie in the strict transforms of S (strictTransformSeq_le_center_pushforward with ker S.subschemeι = S).

[Kol07, Corollary 89] for the push-forward Π = T.pushforward S.subschemeι, read on the stages of T through the embeddings T.pushforwardStageHom, as the first line of [Kol07, 90] uses it: the restricted sequence Π|_S is T and its stage lifts are the stage inclusions (pullbackStageHom_pushforward_heq_mk); the two sets of Corollary 89 are transported pointwise along the equality of stages.

More congruences along equalities of schemes and sequences #

theorem Hironaka.Sequence.totalTransformSeq_congr_heq {X : AlgebraicGeometry.Scheme} {S₁ S₂ : X.BlowUpSequence} (e : S₁ = S₂) (E : X.DivisorFamily) (n : ℕ) (h₁ : n < S₁.length + 1) (h₂ : n < S₂.length + 1) :
theorem Hironaka.Sequence.comap_congr_heq {A B Y : AlgebraicGeometry.Scheme} (e : A = B) {φ : A ⟶ Y} {φ' : B ⟶ Y} (h : φ ≍ φ') (F : Y.DivisorFamily) :
F.comap φ ≍ F.comap φ'

comap of a family along heterogeneously equal morphisms (equal sources).

theorem Hironaka.Sequence.maxOrd_of_heq {A B : AlgebraicGeometry.Scheme} (e : A = B) {J : A.IdealSheafData} {J' : B.IdealSheafData} (h : J ≍ J') :

The maximal order of heterogeneously equal ideal sheaves.

Simple normal crossings along a sequence, in prefix form #

theorem Hironaka.Sequence.isSnc_totalTransformSeq_append_of_forall_lt {X : AlgebraicGeometry.Scheme} {k : Type u} [Field k] [PerfectField k] (f : X ⟶ AlgebraicGeometry.Spec ↧k) [AlgebraicGeometry.Smooth f] (S : X.BlowUpSequence) (J : X.IdealSheafData) (E : X.DivisorFamily) (hsm : S.IsSmooth f) (hE : (E.append J).IsSnc) (j : ℕ) (hsnc : ∀ (i : Fin S.length), ↑i < j → (S.totalTransformSeq E i.castSucc).HasSncWith (S.center i)) (hZ : ∀ (i : Fin S.length), S.strictTransformSeq J i.castSucc ≤ S.center i) (hj : j < S.length + 1) :

isSnc_totalTransformSeq_append_mk with the normal crossing hypothesis only at the stages < j: E_j + H_j is snc as soon as the centers before j have simple normal crossings with the E_i; the same induction, using one clause per step.

totalTransformSeq_comap_pullbackStageHom_mk with the normal crossing hypothesis only at the stages < j: E_j|_{H_j} = (E|_H)_j; the same induction.

The induction of Kollár's item 90, and Theorem 84 as printed #

The going-up theorem [Kol07, Theorem 84] with the boundary E of [Kol07, Corollary 85] carried along: for a D-balanced I with max-ord I = m, a smooth hypersurface S with E + S snc, and a smooth blow-up sequence T of order ≥ m for (S, I|_S, m, E|_S), the push-forward B = T.pushforward ι is a smooth blow-up sequence of order m for (X, I, E).

The proof is the induction of [Kol07, 90]. B is smooth (IsSmooth.pushforward); its normal crossing and order clauses are proved at every stage l by strong induction on l, the clauses at the stages < l making the prefix B.take l a sequence of order ≥ m (isOrderGeSeq_take_of_forall_lt). At stage l, with the stage inclusion g = j_l : S_l ↪ X_l: the order clause follows from max-ord I_l ≤ m (Lemma 61 iterated along the prefix) and cosupp(J_l, m) ⊆ j_l^{-1} cosupp(I_l, m), the latter from Corollary 89 on the prefix T.take l (whose push-forward is B.take l) and the D-balanced property (cosupp_markedTransformSeq_subset_iInter), carried to stage l of T and B along the truncation lemmas; the normal crossing clause lifts the simple normal crossings of Z_l with (E|_S)_l in S_l to E_l in X_l, using E_l + S_l snc and E_l|_{S_l} = (E|_S)_l from the prefix forms of the snc lemmas. Finally Remark 67 (max-ord I = m) turns order ≥ m into order m.

Kollár's Theorem 84 [Kol07, Theorem 84] as printed: the going-up theorem at E = ∅ (E + S = S alone is snc for a smooth hypersurface, isSnc_append_empty_of_isSmoothDivisor; ∅|_S = ∅, empty_comap). The hypothesis _hSc, that S is not contained in cosupp(I, m), is part of Kollár's statement of Theorem 84 and is kept so that the theorem reads as printed; the proof does not need it, because it goes through IsDBalanced.isOrderSeq_pushforward, whose Corollary 89 rests on the sheaf-level form of Lemma 62, which holds with no nonvanishing hypothesis on the restriction I|_S.