The \(L^2\)-Alexander torsions of 3-manifolds (Q2255254): Difference between revisions

In this note, the authors introduce a new invariant of a \(3\)-manifold \(M\), the \(L^2\)-Alexander torsion. It depends on a (factored) representation \(\phi: \pi_1(M)\to G\to\mathbb{Z}\) and is a function \(\tau: (0,\infty)\to [0,\infty)\) (well defined modulo multiplication by \(t\mapsto t\)). It generalizes a similar invariant defined previously for knots by Li and Zhang. It also generalizes the ``classical'' \(L^2\)-torsion (if \(G=\pi_1(M)\) and the representation is trivial) and therefore gives the volume of the hyperbolic pieces in the \(JSJ\)-decomposition. The note contains announcements of further structural results of this invariant and of computations. In particular, it is shown how for a knot complement with \(G\) the maximal abelian quotient, the new invariant is given in terms of the Alexander polynomial. The value of the invariant is then given (at least the value near \(0\) and near \(\infty\)) for a graph manifold \(M\) or for a fibered manifold \(M\) (and certain ``non-degenerate'' representations of \(\pi_1(M)\)); it is explicitly given in terms of the Thurston norm \(\chi_\phi(N)\) and looks like \(t\mapsto t^{\chi_{\phi}(N)}\) for large \(t\). The same holds for general \(3\)-manifolds, but only for suitable \(G\). Finally, the symmetry relation \(\tau(t^{-1})/\tau(t)=t^n\) for an integer \(n\) is announced. Proofs are not given.