Analytic torsion, dynamical zeta function, and the Fried conjecture for admissible twists (Q2231671)

Much of the work in the author's paper is extending David Fried's paper in 1986. We review the following historical facts: \begin{itemize} \item In 1935, Reidemeister, Franz and de Rham introduced the concept of ``torsion'' for certain finite simplicial complexes. (Milnor's account) \item Reidemeister torsion was formally proposed in [\textit{K. Reidemeister}, Bull. Am. Math. Soc. 56, 297--307 (1950; Zbl 0037.39602)] \item Selberg proved Selberg trace formula, an extension of Poisson summation formula in [\textit{A. Selberg}, J. Indian Math. Soc., New Ser. 20, 47--87 (1956; Zbl 0072.08201)] \item Smale proved a h-cobordism theorem in [\textit{S. Smale}, Am. J. Math. 84, 387--399 (1962; Zbl 0109.41103)] \item The Whitehead torsion and Reidemeister torsion were discussed in extension of s-cobordism theorem in 1966 by Minor. \item R. Seeley defined complex power of an elliptic operator in 1967. \item Ray and Singer discussed the differential counter-part of R-torsion, which is usually called as analytic torsion in 1971; a follow up paper in 1973 discussed the case for complex manifolds using \(\overline{\partial}\) operator. Arnold Shapiro suggested R-torsion may have a formula in terms of differential invariants, which motivated the term ''analytic torsion''. It should be pointed out that in their 1973 paper, Ray and Singer computed the case for Riemann surfaces and found the formula can be expressed using Selberg zeta functions. \item Cheeger, Muller proved R-torsion and analytic torsion is equal (1977, 1978). \item Quillen proposed incorporating analytic torsion for the metric on the determinant line bundle (1982--1985) to overcome the difficulty of estimating self-intersection in Arakelov theory. \item Partly motivated by Smale's work on dynamical systems and Ray-Singer's 1973 paper, Fried introduced the Ruelle zeta function: \[ R_{\rho}(s)=\det(I-\rho(\gamma)e^{-s l(\gamma)}) \] where \(s\) is a complex number and \(\rho(\gamma)\) is a presentation of the fundamental group in the hyperbolic manifold \(X\). The \(\gamma\) is varying over prime geodesics on \(X\) and \(l(\gamma)\) denotes its length. Fried's result suggested one can compare the Ruelle zeta function with analytic torsion for \(SX\) (unit tangent bundle, usually considered as sphere bundle on \(X\)). The conjecture was proved for the case when \(\rho: \pi_1(SX)\rightarrow O(m)\) is acylic and \(X\) a closed, oriented hyperbolic manifold (meaning \(X\cong \Gamma /\ H^{d}\) and \(\Gamma\) is a discrete torsion free subgroup of the hyperbolic orthogonal group) as well as the case when it is orthogonal. This answers a problem posed by Singer at the Vancouver ICM Congress in 1974. \item (Source: Muller): Fried conjectured that a similar relationship holds for all compact locally symmetric manifolds X and acyclic orthogonal bundles over \(SX\). This conjecure was recently proved by the author of the paper. (2018) \item Muller proved that a similar relationship holds for all representations, not just acyclic or orthogonal ones. (2020) \end{itemize} The article under review extends the result to the case on a closed odd dimensional locally symmetric space of reductive type. The extension is not entirely trivial, as the definition of Ruelle zeta function (3.15) cannot be directly extended from Fried's paper. In particular the author makes heavy use of Selberg's zeta function associated to a representation (3.19, for example). Armed with the definitions, the author proves the result (Theorem 4.4) by utilizing various step-wise simplifications on the structure of the Lie group. One crucial element used in the authors's paper, which may be worth mentioning is that he uses Dirac-operator methodology in the setting of Lie-theory, which was pioneered by Atiyah, Bott, Kostant during 1960--1980s. However, the paper is rather hard to read with small notational issues: \begin{itemize} \item The author suggests in page 13 ``If \(\gamma \in G\), we denote by \(Z(\gamma)\subset G\) the centraliser of \(\gamma\) in \(G\)\ldots''. But the usual mathematical notation for centralizer is \(C(g)\), not \(Z(g)\). \item In page 15, the author claims ``\ldots the restriction induces an isomorphism of rings\ldots''. But in fact \(R(T)\) is a free-module over the representation ring \(R(K)\). \end{itemize} Due to time restraint I am not able to proof read for every detail in the paper. I suspect a ``condensed'' version with a shorter introduction and a clean over-view of the proof strategy may be helpful. For now the paper does not contain any clear examples on ``why'' the theorem is true.