The principalization sequence: unfolding, the empty blow-up convention, the empty boundary #
The first facts about Hironaka.Sequence.BP
(Hironaka/Resolution/Algebraic/Kol07/Thm35/Principalization.lean): its unfolding (BP_eq); the
empty blow-up convention of [Kol07, 32], "the final outputs of the named blow-up sequence functors …
do not contain empty blow-ups", as a theorem (BP_noEmptyCenters, the definition deletes them); and
the case of an empty boundary (BP_of_isEmpty): for E = ∅ the disjoining sequence is the empty
sequence (disjoinSeq_of_isEmpty), so BP T is BMO_1(X, I, 1, ∅) itself, up to two transports.
First, disjoinSeq T.E = nil holds only propositionally (Fintype.card of an empty index type is
not 0 by rfl), so the disjoined triple (X', π^* I, ∑ E^i) lives on a scheme X' merely equal
to X: concat_pullback_eqToHom moves the BMO_1 part across the eqToHom isomorphism, and the
commutation of BMO_1 with smooth surjections [Kol07, 34.1] at that isomorphism
(dimFreeBMO_commutesWithSmooth) identifies the value. Second, the collapsed family ∑ E^i of an
empty E is not empty: it has the single member E^0 = ∏_{i ∈ ∅} … = 𝒪_X, the empty divisor
(disjoinedFamily_component_bot, card_disjoinedFamily), so the two boundaries differ by one unit
member, which the indifference of BMO_1 to unit members (dimFreeBMO_indifferentToEmptyMembers)
removes along the unique order embedding out of the empty index type. Finally the output of BMO_1
has no empty center (OrderGeSeqAssignment.noEmptyCenters), so the eraseEmpty of the definition
is the identity there (eraseEmpty_eq_self_iff).
The empty-boundary case is what the closed embedding clause of [Kol07, Theorem 35 (5)] reduces
to, since that clause is stated for E = ∅.
The unfolding of the principalization sequence, by definition.
BP T has no empty center: the definition deletes them, as the empty blow-up convention of
[Kol07, 32] requires.
A blow-up sequence that is (propositionally) the empty sequence has composite 𝟙 up to the
identification of its last stage with the base.
For E = ∅ the principalization sequence is BMO_1(X, I, 1, ∅) itself: with k = 0
components there are no disjoining steps [Kol07, 72]. The proof moves the BMO_1 part across the
eqToHom isomorphism X ≅ X' (concat_pullback_eqToHom, then the commutation of BMO_1 with
smooth surjections [Kol07, 34.1] at the isomorphism) after removing the one unit member of the
collapsed family by the indifference of BMO_1 to unit members; the final eraseEmpty is the
identity since the output of BMO_1 has no empty center.