The monomial split on a triple #
The modules Hironaka/Resolution/Algebraic/MarkedOrderReduction/Split.lean, SplitOrder.lean,
SplitMain.lean, SplitSupport.lean and SplitFunctorial.lean prove [Kol07, Definition–Lemma 110]
for a normal-crossings family on a scheme smooth over k (namespace Hironaka.BMO.Snc); this
module instantiates them on a triple T = (X, I, E) ([Kol07, Notation 64]: X smooth of finite
type over k, hence Noetherian; I nonzero on every component; E with normal crossings), in the
forms used by the marked order reduction. Kollár's standing characteristic-zero hypothesis is
carried as the statements carry it.
Every monomial ideal of the triple's family is invertible (the invertibility of M(I) in
[Kol07, Definition–Lemma 110]).
The monomial part is invertible.
I = M(I) · N(I) ([Kol07, Definition–Lemma 110]).
ord_D N(I) = 0 at every component ([Kol07, Definition–Lemma 110]: "cosupp N(I) does not
contain any of the E^i", in the fine form).
N(I) is nonzero on every component.
The split is unique ([Kol07, Definition–Lemma 110]: "uniquely").
The monomial part absorbs monomials ([Kol07, 111, Step 1]).
The nonmonomial part ignores monomials ([Kol07, 111, Step 1]).
For E = ∅, N(I) = I (the last paragraph of [Kol07, 111]).
An invertible ideal sheaf supported on E is its own monomial part ([BM08, (5.2)]: "N(I) is
divisible by no such prime ideal").
M(h^* I) = h^* M(I) for a smooth h: the monomial part commutes with smooth pull-back.
N(h^* I) = h^* N(I) for a smooth h: the nonmonomial part commutes with smooth pull-back.
The monomial part commutes with change of fields ([Kol07, 34.2]).
The nonmonomial part commutes with change of fields ([Kol07, 34.2]).