Twisted Reidemeister torsion and the Thurston norm: graph manifolds and finite representations (Q327406)

The main result of the paper is the following theorem: Let \(N\) be a 3-manifold. Let \(\theta\in H_{1}(N;\mathbb{Z})\simeq H^{2}(N,\partial N;\mathbb{Z})\). For every representation \(V\) of \(\pi_{1}(N)\) the Thurston norm \(\|\theta\|\) satisfies \((\dim V)\cdot\|\theta\| \geq\mathrm{width~}\tau(N;V_{\theta})\). Here \(\|\theta\|\) is the Thurston norm of \(\theta\); \(V_{\theta}\) is a vector space over \(K\) with a linear action by \(\pi_{1}(N)\) induced by \(\theta\in\mathrm{Hom}(\pi_{1}(N),\mathbb{Z})\); \(\tau(N;V_{\theta})\in K(t)\) is the twisted Reidemeister torsion associated with \(V_{\theta}\); and the \(\mathrm{width}\) is computed from the difference between the highest and the lowest exponents in a polynomial. Furthermore, if \(N\) is an irreducible 3-manifold which is not \(D^{2}\times S^{1}\) there exists a representation \(V_{\theta}\) such that equality occurs. In this case the Thurston norm is said to be detected. A non-trivial consequence is that the Thurston norm is determined by the fundamental group. The proof is divided into two cases: graph manifolds and other manifolds. For graph manifolds the proof is based on recent results of the second author on circle-bundles over 3-manifolds. For other manifolds the proof modifies arguments in an earlier paper [\textit{S. Friedl} and \textit{S. Vidussi}, J. Reine Angew. Math. 707, 87--102 (2015; Zbl 1331.57017)], which in turn uses recent advances in ``cubulated'' manifolds.