Derivatives commute with completion #
Kollár's D(Î) = \widehat{D(I)}, where ^ denotes completion [Kol07, Lemma 74(5)]. With the
extended coordinate derivations ∂̂ᵢ of Hironaka/Algebra/Local/CompletionCoords.lean (∂̂ᵢ ∘ ι = ι ∘ ∂ᵢ, so that R̂ carries the coordinate structure c.adicCompletion, given its regularity) and
the identification I R̂ = Î, this is the case e = id of the base-change formula
D(φ(I)) = φ(D(I)) of Hironaka/Algebra/Local/Derivative.lean (D_map): D(I R̂) = D(I) R̂.
Iterating, Dʳ(I R̂) = Dʳ(I) R̂, in particular MC(Î) = \widehat{MC(I)} (Kollár's "since taking
derivatives commutes with completion (74.5), we see that \widehat{MC(I)} = MC(Î)" in the
discussion preceding [Kol07, Theorem 92]), and, since I R̂ ∩ R = I, I is MC-invariant iff Î is
— the transfer used in the proof of [Kol07, Theorem 92].
Extension of ideals to R̂ reflects inclusions (from I R̂ ∩ R = I).
The extended coordinate derivations restrict to the originals, read through
c.adicCompletion: ∂̂ᵢ (ι f) = ι (∂ᵢ f).
Derivatives commute with completion [Kol07, Lemma 74(5)]: D(I R̂) = D(I) R̂, the case
e = id of D_map applied to ι : R → R̂ and the extended derivations.
Iterated: Dʳ(I R̂) = Dʳ(I) R̂.
Kollár's notation D(Î) = \widehat{D(I)} [Kol07, Lemma 74(5)], with Î the image in R̂ of
the completion of the module I (range_adicCompletion_map_subtype).
MC(Î) = \widehat{MC(I)}, i.e. D^{m-1}(I R̂) = D^{m-1}(I) R̂ (the discussion preceding
[Kol07, Theorem 92]).
D(I R̂) ∩ R = D(I).
MC(Î) ∩ R = MC(I).
I is MC-invariant with respect to m iff I R̂ is: the transfer of MC-invariance to the
completion in the discussion preceding [Kol07, Theorem 92], with the definition (53.1) of
[Kol07, 53].