No label defined (Q4263289)

Quantum invariants of 3-manifolds and the associated topological quantum field theories give rise to some finite-dimensional representations of mapping class groups of surfaces. The author states the question: whether these representations have finite image. It was shown, that this is the case in genus 1, but for \(g>1\) \textit{L. Funar} (to appear in Pacific J. Math.) has shown that the image of the mapping class group is an infinite group, except for the representations corresponding to a few small levels. The author of the paper under review completes Funar's result and simplifies his proof by exhibiting such a mapping class explicitly. Although the principal idea is the same as Funar's, the proof in the reviewed paper is based on a simple Kauffman bracket computation. The main result of the paper is the following: Let \(\Sigma\) be a surface of genus \(g>1\), and \(\alpha\) and \(\beta\) be the Dehn twists along simple closed curves \(a\) and \(b\) meeting transversely in exactly two points but with opposite signs. A regular neighborhood of \(a\cup b\) in \(\Sigma\) is a four-holed sphere \(S\). It is assumed that none of the four boundary circles of \(S\) is null-homologous in \(\Sigma \smallsetminus S\). For \(g>1\) such a pair always exists. Proposition. The image of the mapping class \(\alpha^{-1} \beta\) in the TQFT-representation corresponding to \(SU(2)\) at level \(k\) has infinite order except if \(k=1,2,4,\) or \(8\). This means that no power of the mapping class \(\alpha^{-1} \beta\) acts as scalar matrix and that a product of at least two Dehn twists is needed to get a matrix of infinite order. For the entire collection see [Zbl 0919.00057].