Hypersurfaces of maximal contact: locality of the dynamic form #
[Kol07, Definition 78] ends with "Being a hypersurface of maximal contact is a local property."
The definition itself (IsDynamicMaximalContact,
Hironaka/Resolution/Algebraic/MaximalContact/Basic.lean) quantifies over every open X⁰ ⊆ X and
every smooth blow-up sequence of order m for (X⁰, I|_{X⁰}) with no divisorial part; this file
proves that the property can be checked on the members of an open cover X = ⋃ U_α
(isDynamicMaximalContact_iff_forall_opens):
- (⇒) (
IsDynamicMaximalContact.comap): an open subschemej : Y ⟶ U_αof a member of the cover is an open subschemej ≫ (U_α).ιofX(Mathlib'sIsOpenImmersion.comp), and the restricted data are the restricted data ofX(comap_comp); so the hypothesis onXapplies verbatim. Nothing about sequences is used. - (⇐) (
isDynamicMaximalContact_of_forall_opens): letj : Y ⟶ Xbe an open subscheme andSa smooth blow-up sequence of ordermfor(Y, I|_Y). The opensV_α := j⁻¹(U_α)coverY. Along each open immersion(V_α).ιthe pulled-back sequenceS.pullback (V_α).ιis again a smooth blow-up sequence of ordermfor the restricted data (the functoriality [Kol07, 34.1],IsOrderSeq.pullback; an open immersion is smooth of relative dimension0, and no deletion of empty blow-ups is needed, since an empty pulled-back centre satisfies the clauses of [Kol07, Definition 66] vacuously), andV_αis an open subscheme ofU_αthroughj.resLE : V_α ⟶ U_α(resLE_comp_ι,IsOpenImmersion.of_comp). The hypothesis onU_αtherefore puts every centre of the pulled-back sequence inside the strict transform ofH ∩ V_α. The centres of the pulled-back sequence are the inverse images of the centres (center_pullback) and its strict transforms are the inverse images of the strict transforms (strictTransformSeq_pullback, for the pull-back of [Kol07, Definition 30.1]) under the stage liftsh_i : (S.pullback (V_α).ι).stage i ⟶ S.stage i, which are open immersions (isOpenImmersion_pullbackStageHom) whose imagesΠ_i⁻¹(V_α)cover the stage (range_pullbackStageHom). Containment of closed subschemes is local: an inclusion of ideal sheaves holds once it holds aftercomapalong a jointly surjective family of open immersions (le_of_comap_le_of_forall_exists), which givesZ_i ⊆ H_ionY.
Kollár's standing hypotheses (X smooth over k of characteristic zero) enter only through
IsOrderSeq.pullback, which needs them to carry the order clause along the pull-back.
The direction (⇒) of the locality in [Kol07, Definition 78]: a hypersurface of maximal contact
for (X, I, m) restricts to one on every open subscheme g : Y ⟶ X, since an open subscheme of
Y is an open subscheme of X, and the restricted data restrict again.
The direction (⇐) of the locality in [Kol07, Definition 78]: if H ∩ U_α is a hypersurface of
maximal contact for (U_α, I|_{U_α}, m) for every member of an open cover
X = ⋃ U_α, then H is one for (X, I, m). Given an open j : Y ⟶ X and a sequence S of
order m on Y, pull S back to the pieces j⁻¹(U_α) of the induced cover of Y, read the
centres and strict transforms of the pulled-back sequences as inverse images under the stage
lifts, and descend the containment Z_i ⊆ H_i along those open immersions, which cover each
stage.
Kollár's "Being a hypersurface of maximal contact is a local property" ([Kol07, Definition 78]):
for an open cover X = ⋃ U_α, H is a hypersurface of maximal contact for (X, I, m) iff
H ∩ U_α is one for (U_α, I|_{U_α}, m) for every α.