The principalization sequence commutes with closed embeddings when E = ∅ #
Clause (5) of [Kol07, Theorem 35], the functoriality for closed embeddings of [Kol07, 34.3]: for a
closed embedding j : Y ↪ X of smooth k-schemes with 𝒪_X/I_X = j_*(𝒪_Y/I_Y) and E = ∅,
BP(X, I_X, ∅) = j_* BP(Y, I_Y, ∅). With E empty the disjoining part of BP is empty
(BP_of_isEmpty: BP T = BMO_1(X, I, 1, ∅)), so the clause is exactly [Kol07, Claim 71.2] for
BMO_1, "the functoriality properties required in (35) follow … from (71.2)" [Kol07, 72]. Claim
71.2 is proved in Hironaka/Resolution/Algebraic/Stage/ClosedEmbeddingGeneral.lean for the functor
at a fixed dimension (Hironaka.Stage.tower_bmo_one_eq_pushforward_closedEmbedding_aux); it is read
here at the dimension of
X through dimFreeBMO_seq_eq_tower (BMO_1 is the dimension-free functor, and the closed
subscheme Y has dimension ≤ dim X, bmoClass_of_closedEmbedding). BP_eq_of_isEmpty
says that BP does not see how an empty boundary is indexed, through the indifference of BMO_1
to unit members; with it the clause holds for any empty boundary on Y.
Clause (5) of [Kol07, Theorem 35]: for a closed embedding j : Y ↪ X carrying
𝒪_X/I_X = j_*(𝒪_Y/I_Y) and E = ∅, BP(X, I_X, ∅) = j_* BP(Y, I_Y, j^{-1} ∅). The disjoining
part is empty (BP_of_isEmpty) and the BMO_1 part is [Kol07, Claim 71.2] at the dimension of
X.
BP does not see how an empty boundary is indexed: two triples on the same (X, I) whose
boundaries both have empty index type have the same principalization sequence (the indifference of
BMO_1 to unit members, at the dimension of X).