Computing motivic zeta functions on log smooth models (Q2182422)

The \lq classical\rq\ motivic zeta function \(Z(f;T)\) of Denef-Loeser associated to a polynomial \(f\) over a field of characteristic zero (or, more generally, to a regular function \(f\) on a smooth variety) is a vast singularity invariant associated to the hypersurface given by \(f=0\), specializing to more concrete invariants as Hodge spectrum and hence monodromy eigenvalues. It is defined as a power series in \(T\) with coefficients in a certain Grothendieck ring of varieties, that encode the \(m\)-jets on the hypersurface for varying \(m\). A formula for \(Z(f;T)\) in terms of an embedded resolution of \(f\) shows that \(Z(f;T)\) is in fact a rational function in \(T\). In particular, each irreducible component of the exceptional locus of such an embedded resolution induces a candidate pole of \(Z(f;T)\). It is an important and difficult question to determine the \emph{actual} poles of \(Z(f;T)\). Indeed, typically \lq most\rq\ of those candidate poles cancel out after simplifying the formula. (This is strongly related to the famous {\em monodromy conjecture} relating poles of \(Z(f;T)\) to eigenvalues of local monodromy of \(f\).) In this paper the authors establish a formula for \(Z(f;T)\) in terms of a {\em log smooth model}, in the setting/language of log geometry. This is a kind of partial resolution with toroidal singularities. A crucial advantage of this new formula is that it contains typically much less candidate poles compared with the classical formula. A related formula appeared before in the literature, namely for polynomials \(f\) that are non-degenerate with respect to their Newton polyhedron, see [\textit{G. Guibert}, Comment. Math. Helv. 77, No. 4, 783--820 (2002; Zbl 1046.14008)] and [\textit{B. Bories} and \textit{W. Veys}, Igusa's \(p\)-adic local zeta function and the monodromy conjecture for non-degenerate surface singularities. Providence, RI: American Mathematical Society (AMS) (2016; Zbl 1397.14020)]. This non-degenerate setting can be considered as a special case of the new formula in terms of a log smooth model. An important aspect of the new formula is that it is also valid for a related kind of motivic zeta function, associated to degenerations of Calabi-Yau varieties, which was introduced in [Adv. Math. 227, No. 1, 610--653 (2011; Zbl 1230.11076)] by \textit{L. H. Halle} and the second author. This was for instance used in [\textit{L. H. Halle} and \textit{J. Nicaise}, Math. Ann. 370, No. 3--4, 1277--1320 (2018; Zbl 1400.14045)] to show an analogue of the monodromy conjecture in that setting for a large class of such degenerations.