Seifert surgery on knots via Reidemeister torsion and Casson-Walker-Lescop invariant (Q2343059)

Let \(\Sigma\) be a homology sphere and \(K \subset \Sigma\) a knot. In this paper the authors discuss (under various hypotheses on \(\Sigma, K\)) the question of which Dehn surgeries on \(K\) yield a manifold that is Seifert-fibered. Seifert-fibered manifolds are well-understood topologically, and are heuristically ``rare'' among three-manifolds so that it is expected that in general few surgeries will yield such a manifold. The main result in the paper is that when \(\Sigma\) has vanishing Lescop invariant \(\lambda(\Sigma)\) (a review of this invariant is given in [\textit{C. Lescop}, Global surgery formula for the Casson-Walker invariant. Annals of Mathematics Studies. 140. Princeton, NJ: Princeton Univ. Press (1996; Zbl 0949.57008)]) and the Alexander polynomial of \(K\) is equal to that of the figure-eight (namely \(t^2 - 3t + 1\)), assuming in addition that \(q\not=\pm 1\) and a further condition on a relation between the Alexander polynomial and the Lescop invariant of the \(2/q\) surgery on \(\Sigma \setminus K\), then the latter is not a Seifert fibered manifold. The authors prove another similar result with the assumptions that \(q\not=\pm 1\) and \(\lambda(\Sigma) \not= 0\) replaced by an additional hypothesis on the Lescop invariant of the result of the \(2/q\)-surgery. The proofs are based on earlier results of the first author [Osaka J. Math. 43, No. 4, 823--837 (2006; Zbl 1141.57004); Algebr. Geom. Topol. 7, 1509--1529 (2007; Zbl 1145.57004)] which allow to give a surgery presentation (in \(\mathbb S^3\)) of the \(2/q\)-surgery on \(\Sigma\) from the Seifert structure. Calculations then give a contradiction with the hypotheses.