Étale fundamental groups of Kawamata log terminal spaces, flat sheaves, and quotients of abelian varieties (Q739072)

Removing a codimension \(\geq 2\) subset from a smooth complex algebraic variety \(X\) does not change the fundamental group. The commutative-algebraic version of this fact is the well-known Zariski-Nagata theorem on purity of the branch locus [\textit{O. Zariski}, Proc. Natl. Acad. Sci. USA 44, 791--796 (1958; Zbl 0087.35703)]. In the paper under review, the authors deal with the case where \(X\) is allowed to have klt singularities, or more precisely \(X\) admits a divisor \(\Delta\) such that the pair \((X, \Delta)\) is klt. They prove that given a sequence of finite surjective quasi-étale (i.e.,~étale in codimension one) morphisms \[ \cdots \to Y_2 \to Y_1 \to Y_0 = X \] such that all the induced maps \(Y_i \to X\) are Galois, then all but finitely many of the maps \(Y_{i+1} \to Y_i\) are étale (Theorem~1.1). Actually, they prove a somewhat more general result (Theorem~2.1) which is important for the applications, but too complicated to reproduce here. As an immediate application, any such \(X\) admits a finite surjective quasi-étale Galois cover \(\tilde X \to X\) such that any quasi-étale cover of \(\tilde X\) is in fact étale (Theorem~1.5). Equivalently, the étale fundamental group of \(\tilde X\) agrees with that of its smooth locus. The proof of Theorem~2.1 uses the theory of Whitney stratifications and a recent result by Xu (Zbl 1291.14057). The paper is very well-written and the authors make some effort to show by examples that their statements are sharp. Several interesting corollaries are given. First of all, using a local version of the main result (Theorem~1.9), it follows that there is a uniform bound \(r \in \mathbb N\) such that \(rD\) is Cartier for any \(\mathbb Q\)-Cartier divisor \(D\) on \(X\). The most interesting application is Theorem~1.17, a characterization of quotients of abelian varieties by finite groups acting freely in codimension two in terms of a vanishing condition on the first and second Chern class. This is a generalization of an old result in dimension three by \textit{N. I. Shepherd-Barron} and \textit{P. M. H. Wilson} [J. Algebr. Geom. 3, No. 2, 265--281 (1994; Zbl 0807.14031)]. It has since been generalized further by \textit{S. Lu} and \textit{B. Taji} [``A characterization of finite quotients of abelian varieties'', Preprint, \url{arXiv:1410.0063}] to the case of actions free in codimension one.