Analytic torsion versus Reidemeister torsion on hyperbolic 3-manifolds with cusps (Q2509931): Difference between revisions

On a closed Riemannian manifold with a Euclidean flat bundle one can define the Ray-Singer analytic torsion, a spectral invariant, and the Reidemeister torsion, a topological one. It is a very interesting fact that the two are actually equal: this is the famous Cheeger-Müller Theorem. For closed hyperbolic three-manifolds the collection of Reidemeister torsions associated to the representations of \(\mathrm{SL}_2(\mathbb{C})\) is a strong topological invariant: for example, it determines the volume of the manifold (by a result of \textit{W. Müller} [Prog. Math. 297, 317--352 (2012; Zbl 1264.58026)], and it is also related to the torsion homology. It is somewhat harder to define an analytic torsion for finite-volume, non-compact hyperbolic three--manifolds since the presence of cusps complicates spectral analysis on them. This has been dealt with in a previous paper of the author together with \textit{W. Müller} [J. Funct. Anal. 263, No. 9, 2615--2675 (2012; Zbl 1277.58018)], and also by the reviewer in [\url{arXiv 1212.3161}]. Furthermore, there is as of now no equivalent to the Cheeger-Müller theorem for them. It is not even completely clear what the topological side of such a formula should be; a good candidate was given in recent work of \textit{F. Calegari} and \textit{A. Venkatesh} [\url{arXiv 1212.3847}], and a cruder Reidemeister torsion was introduced in a paper of \textit{P. Menal-Ferrer} and \textit{J. Porti} [J. Topol. 7, No. 1, 69--119 (2014; Zbl 1302.57044)]. There is also related work of the reviewer on this topic, [\url{arXiv 1307.2845}]. In view of such lacks, any result giving a relation between the regularized analytic torsion and a topological invariant is of certain interest. The main theorem in the present paper is such a result: for two different flat bundles coming from holomorphic representations of \(\mathrm{SL}_2(\mathbb{C})\) with dimensions of the same parity it asserts that there is (up to an explicitly computed constant depending on the dimensions) an equality of the quotients of the regularized analytic torsions on one side, and the quotient of Reidemeister torsions of Menal-Ferrer-Porti on the other side. This result has already been used by the author [Ann. Global Anal. Geom. 45, No. 4, 267--285 (2014; Zbl 1323.11031)] to give a crude generalization of a result of N. Bergeron and A. Venkatesh on exponential growth of homological torsion. The proof of the main result uses the relation between the regularized analytic torsion and special values of some Ruelle zeta function (a result which may be of independent interest) and then using results proven by Menal-Ferrer-Porti [loc. cit.]. Another more sophisticated approach was used by the author in a recent work to prove an actual Cheeger-Müller equality [\url{arXiv 1312.6384}]; however the latter formula involves some terms which are not quite explicit.