Resolution of singularities of pairs preserving semi-simple normal crossings (Q1940353)

This article gives a solution to a problem of [\textit{J. Kollár}, ``Semi log resolutions'', \url{arXiv:0812.3592}, problem 19] on resolution of singularities of pairs except for semi-snc singularities. Let \(X\) be a reduced algebraic variety over a field of characteristic 0 and \(D\) be a \(\mathbb Q\)-Weil-divisor on \(X\). In the sence of Kollar, the pair \((X,D)\) is semi-snc (``semi-simple normal crossings'') at \(a\in X\), if \(X\) is simple normal crossings at \(a\) (i.e. locally a simple normal crossings hypersurface in a smooth variety) and \(D\) is induced by the restriction to \(X\) of a hypersurface that is simple normal crossings with respect to \(X\). Main result of the article is the following theorem on partial resolution of singularities of \((X,D)\): ``Let \(U\subseteq X\) be the largest open subset such that \((U,D|_U)\) is semi-snc. Then there is a morphism \(f:\tilde X \to X\) given by a composite of blowings-up with smooth (admissible) centers, such that [the total transform] \((\tilde X,\tilde D)\) [of \((X,D)\)] is semi-snc and \(f\) is an isomorphism over \(U\).'' The theorem has two special cases previously known already: If \(X\) is smooth, the resolution obtained is the snc-strict log-resolution (i.e. the log resolution of singularities of \(D\) by a morphism that is an isomorphism over the snc locus). Another important case is \(D=0\). Both of these cases are ingredients of the proof of the main theorem, where the reduction to the case that \(X\) has only snc-singularities is used, followed by an induction on the number of components of \(X\). Functoriality of the desingularization morphism holds for algebraic varieties in characteristic 0 with fixed ordering of the components of \(X\) and with respect to etale morphisms that preserve the number of irreducible components of \(X\) and \(D\) passing through every point. Functoriality for general etale morphisms does not hold. The paper is essentially self-contained (apart from resolution of singularities which is used as a ``black box''). It is based on an idea from papers of the first named author and J. Milman, E. Lairez respectively [\textit{E. Bierstone} and \textit{P. D. Milman}, Adv. Math. 231, No. 5, 3022--3053 (2012; Zbl 1257.14002); \textit{E. Bierstone, P. Lairez} and \textit{P. D. Milman}, Adv. Math. 231, No. 5, 3003--3021 (2012; Zbl 1262.14003); \textit{E. Bierstone} and \textit{P. D. Milman}, Invent. Math. 128, No. 2, 207--302 (1997; Zbl 0896.14006)] on characterizing local normal forms of mild singularities. It should be noted, that the mild singularities treated in the previous papers [\textit{E. Bierstone, S. Da Silva, P. M. Milman} and \textit{F. V. Pacheco}, ``Desingularization by blowing up avoiding, simple normal crossings'', \url{arXiv:1206.5316v1}; Zbl 1257.14002] are hypersurface singularities, contrary to the semi-snc case. Therefore, the desingularization invariant is more complicated here: It is the Hilbert-Samuel function and the invariant based on it which is used to characterize semi-snc singularities.