Analytic torsion of arithmetic quotients of the symmetric space \(\mathrm{SL} (n,\mathbb{R})/ \mathrm{SO}(n)\) (Q1688814)

The spectral zeta functions of a compact Riemannian manifold are a classical object of study in differential geometry, as is the analytic torsion they are used to define. Recent work of \textit{N. Bergeron} and \textit{A. Venkatesh} [J. Inst. Math. Jussieu 12, No. 2, 391--447 (2013; Zbl 1266.22013)] studies the analytic torsion in certain sequences of arithmetically defined locally symmetric spaces and uses this to deduce results on the torsion subgroup of the integral cohomology of arithmetic groups. The zeta functions and analytic torsion are not defined for noncompact manifolds but from the arithmetic viewpoint it is very interesting to study the homology of such spaces, for example those associated to \(\mathrm{SL}(n, \mathbb Z)\). In the compact setting analytic torsion can be defined using the trace of the heat operator on differential forms and it is natural to attempt to use the trace formula to give a similar definition of analytic torsion, and study it, in general. In rank one (for hyperbolic manifolds) part of this program was dealt with, in work of the second author with J. Pfaff and of the reviewer. The goal of this paper is to extend the definition of analytic torsion to the algebraically simplest example of an arithmetic lattice, that in a split form of \(\mathrm{SL}(n)\) (such as \(\mathrm{SL}(n, \mathbb Z)\)). The asymptotics in congruence covers is the object of a followup work by the authors [``Approximation of \(L^2\)-analytic torsion for arithmetic quotients of the symmetric space \(\mathrm{SL}(n,\mathbb R) / \mathrm{SO}(n)\)'', Preprint, \url{arXiv:1709.07764}]. The main result of the paper is thus that if \(M\) is a manifold locally isometric to the symmetric space \(\mathrm{SL}(n, \mathbb R)\) associated to a congruence subgroup of \(\mathrm{SL}(n, \mathbb Z)\) then one can use the trace formula to define zeta-functions for the Laplacians on forms on \(M\) with coefficients in local systems satisfying a certain condition. Moreover these zeta-functions extend to meromorphic functions on \(\mathbb C\) (this is a corollary of Theorems 1.1 and 1.2). While these functions might not be regular at \(0\), in the case \(n=3\) the authors prove that their linear combination appearing in the definition of analytic torsion is regular at 0, so one can apply this usual definition also in this case. To prove these results the authors have to delve into, and make more explicit, Arthur's computation of the unipotent terms on the geometric side (both to establish the general result and for the more precise statement in the \(n=3\) case) and then examine their behaviour when the trace formula is applied to the heat kernel at \(t \to 0\). They also have to check that the spectral gap of the coefficient systems they consider carries over to the large-time behaviour of the terms on the spectral side coming from Eisenstein series, which uses work of the second author, and of his with T. Finis and E. Lapid. In fine, the application of the trace formula to the heat kernel also depends on recent work of Finis-Lapid on the geometric side.