Assembly of BD_{n,m,j}: the tuning layer, the mark 0, and the data of clause (3) #
[Kol07, Lemma 102] asserts, for every m, j, a smooth blow-up sequence functor BD_{n,m,j} of
order m on triples with dim X ≤ n and max-ord I ≤ m, with clauses (1)–(3). Its proof first
reduces to a D-balanced ideal (replace I by W_{s(m)}(I), Tuned.lean), then builds the
sequence (Center.lean to Output.lean), proves its functoriality (FunctorialityPullback.lean,
FunctorialityBaseChange.lean), and ends with clause (3). This module defines the pieces of the
assembly and discharges the obligations they carry.
BD.zeroFunctor n j, the functor at the markm = 0. The classBDClass n 0 jis not empty (Kollár'smis a mark,≥ 1;BDClassdoes not exclude0):max-ord I ≤ 0makesIthe unit ideal, every component ofE^jlies incosupp(I, 0), soZ_{-1} = E^j,π_{-1}is the blow-up ofE^jand the birational transform ofE^jis empty. The functor isπ_{-1}with its empty blow-up deleted ([Kol07, 32]); no inductive input is needed.BD.tunedFunctor hm F hDom', the tuning device of Step 1 of the proof of [Kol07, Theorem 103] for Lemma 102's class: from a functorFof orders(m)on a class containing the tuned triples ofBDClass n m j, the functor of ordermonBDClass n m jwhose value isF(X, W_{s(m)}(I), E)whenmax-ord I = mand the empty sequence whenmax-ord I < m(the convention of [Kol07, Theorem 68]: "the casemax-ord I < mis trivial, that is,X_r = X"). It is the instance atC := BDClass n m jof the class-genericOrderSeqAssignment.ofTunedClass(whose instanceOrderSeqAssignment.ofTunedis the same device on the class of Theorem 103); the order condition isisOrderSeq_iff_tuned(Tuned.lean, [Kol07, Corollary 101] at this library's tuning parameter).BD.domain_tuned: the tuned triple of a triple ofBDClass n m jabove the mark lies in Lemma 102's standing classDomain n (s(m)) j(isDBalanced_tuned,maxOrd_tunedofHironaka/Resolution/Algebraic/OrderReduction/Tuned.lean).BD.dataFunctor n m j hm B hDom, the functorBD_{n,m,j}over one field, form ≥ 1:tunedFunctorapplied tofunctor n (s(m)) j B hDomofOutput.lean, for a marked functorBof the shape of [Kol07, Theorem 69] at marks(m)in dimensions≤ n − 1(the induction hypothesisBMO_{≤ n−1, s(m)}). Lemma 102's dataBDData n m j, the functor over every field with clauses (1)–(2), is assembled from it inAssemblyTuned.lean(bdData) and fromzeroFunctorinAssemblyZero.lean(bdDataZero).BD.hypersurfaceTriple T hj J hJ, the marked triple(E^j, J, 1, (E − E^j)|_{E^j})of clause (3), forJnonzero on every component ofE^j. Its family is the members ofE − E^jrestricted toE^j, followed by an EMPTY member (the unit ideal): the exceptional member that the restricted marked triple ofRestriction.leancarries atZ_{-1} = ∅, so that the latter is exactly its pull-back along the trivial blow-up; Kollár's(E − E^j)|_{E^j}has no such member (the total-transform convention of this library, as forE_0ofBD.piMinusOne). The obligations:E^jis smooth of relative dimension one less (isSmoothDivisor_component), the family is snc (isSnc_restrictedFamilyandisSnc_append_top).BD.maxOrd_le_one_of_eq_map,BD.bdClass_one_of_eq_map: the hypothesis of clause (3),τ_*(O_{E^j}/J) = O_X/I, i.e.I = J.map τ, puts(X, I, E)in the class ofBD_{n,1,j}: "every local equation ofE^jis an order 1 element inI" (the proof of [Kol07, Lemma 102]), somax-ord I ≤ 1.
Used by AssemblyZero.lean, and outside this directory by
Hironaka/Resolution/Algebraic/OrderReduction/Step24ClosedEmbedding.lean
(Hironaka/Resolution/Algebraic/OrderReduction/Step21BoundaryClearing.lean imports it and uses only
BDData).
BD_{n,0,j}, the functor of [Kol07, Lemma 102] at the mark 0: on BDClass n 0 j (where
max-ord I ≤ 0 forces I to be the unit ideal), the trivial blow-up π_{-1} of Z_{-1} = E^j
with its empty blow-up deleted ([Kol07, 32]); a smooth blow-up sequence of order 0 starting with
(X, I, E) by isOrderSeq_piMinusOne (Center.lean).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The reduction to the tuned ideal on BDClass n m j (the first paragraph of the proof of
[Kol07, Lemma 102]; the convention of [Kol07, Theorem 68] below the mark): from a smooth blow-up
sequence functor F of order s(m) on a class containing the tuned triples of BDClass n m j, the
functor of order m on BDClass n m j whose value on (X, I, E) is F(X, W_{s(m)}(I), E) when
max-ord I = m and the empty sequence when max-ord I < m. The instance of
OrderSeqAssignment.ofTunedClass at C := BDClass n m j (the mark bound is the global hm).
Equations
- Hironaka.BD.tunedFunctor hm F hDom' = F.ofTunedClass ⋯ hDom'
Instances For
The tuned triple of a triple of BDClass n m j with max-ord I = m lies in the standing class
Domain n (s(m)) j of Restriction.lean (the first paragraph of the proof of [Kol07, Lemma 102]):
W_{s(m)}(I) is D-balanced of maximal order s(m) (isDBalanced_tuned, maxOrd_tuned), the
dimension and the members of E are unchanged.
The functor BD_{n,m,j} over one field ([Kol07, Lemma 102] with the first paragraph of its
proof), for a mark m ≥ 1 and a marked functor B of the shape of [Kol07, Theorem 69] at mark
s(m) in dimensions ≤ n − 1 (the inductive hypothesis): the functor of Output.lean at order
s(m) on the tuned triple, and the empty sequence below the mark. Lemma 102's data BDData n m j
over every field is assembled from it in AssemblyTuned.lean (bdData).
Equations
- Hironaka.BD.dataFunctor n m j hm B hDom = Hironaka.BD.tunedFunctor hm (Hironaka.BD.functor n (IsLocalRing.tuningParam m) j B hDom) ⋯
Instances For
The functor BD_{n,m,j} over one field for every mark ([Kol07, Lemma 102], "for every
m, j"), by a case split on the mark: zeroFunctor at m = 0 and dataFunctor at m ≥ 1. The
inductive input B, a marked functor at mark s(m) in dimensions ≤ n − 1, is an argument at
every mark so that the functor has one signature, Kollár's BD_{n,m,j} for every m; the
definition does not need it at m = 0, because there I is the unit ideal and the value is the
trivial blow-up π_{-1} alone, without any inductive input (zeroFunctor).
Equations
- Hironaka.BD.totalFunctor n 0 x x_2 x_3 = Hironaka.BD.zeroFunctor n x
- Hironaka.BD.totalFunctor n m.succ x x_2 x_3 = Hironaka.BD.dataFunctor n (m + 1) x ⋯ x_2 x_3
Instances For
The marked triple (E^j, J, 1, (E − E^j)|_{E^j}) of clause (3) of [Kol07, Lemma 102], for
an ideal J ⊂ O_{E^j} nonzero on every irreducible component of E^j; its family is the members of
E − E^j restricted to E^j, followed by an empty member (see the module docstring). E^j is
smooth over k of relative dimension one less than X (isSmoothDivisor_component,
Restriction.lean), J is nonzero on every component by hypothesis, and the family is snc
(isSnc_restrictedFamily, isSnc_append_top).
Equations
- One or more equations did not get rendered due to their size.
Instances For
The marked triple of clause (3) has dimension ≤ n − 1 when (X, I, E) has dimension ≤ n
(Kollár's BMO_{n−1,1}, [Kol07, Lemma 102]).
If O_X/I = τ_*(O_{E^j}/J), i.e. I = J.map τ, then max-ord I ≤ 1 ("Every local equation of
E^j is an order 1 element in I", the proof of [Kol07, Lemma 102]): I contains the ideal of
E^j (ker τ = ⊥.map τ ≤ J.map τ), which has order 1 along E^j (a smooth divisor), and I is
the unit ideal off E^j (its support lies in the closed image of τ).
Under the hypothesis of clause (3) of [Kol07, Lemma 102], (X, I, E) lies in the class of
BD_{n,1,j} when dim X ≤ n.