Topological and Hodge L-classes of singular covering spaces and varieties with trivial canonical class (Q1736080)

For closed, oriented topological manifolds, the signature is multiplicative under finite covers. This property is no longer true for general Poincaré duality spaces, as shown by examples of \textit{C. T. C. Wall} [Ann. Math. (2) 86, 213--245 (1967; Zbl 0153.25401)]. In order to take into account the existence of singularities, one replaces the usual homology by the intersection homology of \textit{M. Goresky} and \textit{R. MacPherson} [Topology 19, 135--165 (1980; Zbl 0448.55004)]. The latter depends on a parameter, called perversity, and satisfies a Poincaré duality for pseudomanifolds, involving different perversities. A first possibility for an autoduality in the presence of singularities is represented by Witt spaces for which the intersection homologies of the two middle perversities coincide. For them, \textit{P. H. Siegel} has defined a bordism-invariant signature [Am. J. Math. 105, 1067--1105 (1983; Zbl 0547.57019)]. \textit{M. Banagl} [Extending intersection homology type invariants to non-Witt spaces. Providence, RI: American Mathematical Society (AMS) (2002; Zbl 1030.55005)] introduced a more general situation than Witt conditions. Under some hypotheses involving the existence of Lagrangian sheaves, he constructs a self-dual sheaf ``between'' the two middle perversities and satisfying Poincaré duality. Pseudomanifolds having such a structure are called L-pseudomanifolds; they include Witt spaces. \textit{M. Banagl} [Ann. Math. (2) 163, No. 3, 743--766 (2006; Zbl 1101.57013)] proved that the signature of the self-dual sheaf of an L-pseudomanifold is a bordism-invariant, independent of the choice of the self dual sheaf. He also introduced homological characteristic L-classes, that are also independent of this choice. Recall that these classes agree with the Goresky-MacPherson-Siegel L-classes in the case of a Witt space. The first main result of the paper under review is the multiplicativity of the signature and, more generally, the topological L-class of closed, oriented L-pseudomanifolds. The second theorem is the proof of the Brasselet-Schürmann-Yokura conjecture for normal complex projective 3-folds with at most canonical singularities, trivial canonical class and positive irregularity. This conjecture is the equality of topological and Hodge L-class for compact complex algebraic rational homology manifolds.