On the hard Lefschetz property of stringy Hodge numbers (Q1012563)

The stringy numbers of a projective variety with Gorenstein canonical singularities are the coefficients of its stringy \(E\)-function supposed it is a polynomial (in general it is a rational function in 2 variables). Introduced by \textit{V. Batyrev} [in: Integrable systems and algebraic: Geometry, World Scientific (1998; Zbl 0963.14015)], they play the role of Hodge numbers in few constructions of string cohomology [\textit{L. Borisov} and \textit{A. Mavlyutov}, Adv. Math. 180, No. 1, 355--390 (2003; Zbl 1055.14044)]. They are also connected to the orbifold cohomology [\textit{W. Chen} and \textit{Y. Ruan}, Commun. Math. Phys. 248, No. 1, 1--31 (2004; Zbl 1063.53091)] and are the Hodge numbers of it for varieties with Gorenstein quotient singularities [\textit{T. Yasuda}, Compos. Math. 140, No. 2, 396--422 (2004; Zbl 1092.14028)]. The main result in the article under review is the analogue of the Hard Lefschetz property for the stringy numbers, i.e., \(h^{p,q}_{st}(Y)\leq h_{st}^{p+1,q+1}(Y)\) for \(p+q\leq d-2\), \(d=\dim (Y)\). It is proved for each of the next classes: i) a three dimensional projective variety with Gorenstein canonical singularities; ii) a projective variety of dimension bigger than 3 with at most isolated Gorenstein singularity, admitting a log resolution with discrepancy bigger than \(\lfloor\frac{d-4}{2} \rfloor\). Its proof is given in sec. 3 after the stringy Hodge numbers are defined briefly in the previous section. Then in the last section is proposed a new example of a six dimensional projective hypersurface with isolated canonical singularity not satisfying the Hard Lefschetz property, and an earlier example [\textit{M. Mustata} and \textit{S. Payne}, Math. Ann. 333, No. 4, 787--795 (2005; Zbl 1082.52007)] is compared with the main theorem.