Finite Dehn surgeries on knots in \(S^3\) (Q679792)

For a knot in the \(3\)-sphere, a Dehn surgery is called a finite surgery if the resulting manifold has finite fundamental group. For non-hyperbolic knots, finite surgeries have been classified. Hence, the remaining challenge is to classify them for hyperbolic knots. There are a lot of results concerning this topic. Let \(K\) be a hyperbolic knot. Any nontrivial cyclic surgery is integral, \(K\) has at most two such surgeries, and if two, they correspond to consecutive integers. Also, any finite surgery slope is either an integer or a half-integer. The distance between two finite surgery slopes is at most \(3\), and \(K\) has at most \(4\) nontrivial finite surgery slopes, and at most one half-integral finite surgery slope. These are obtained by using mainly \(\mathrm{PSL}_2(\mathbb{C})\)-representations of \(3\)-manifold groups, hyperbolic geometry and geometric and combinatorial topology. On the other hand, Heegaard Floer homology theory brings new progress. If \(K\) admits a nontrivial finite surgery, then it is fibered, and the nonzero coefficients of the Alexander polynomial of \(K\) are alternating \(\pm 1\). If \(p/q\) is a nontrivial finite surgery slope, then \(|p/q|\geq 2g(K)-1\), where \(g(K)\) is the genus of \(K\). If \(K\) admits a cyclic surgery slope \(p\), then there exists a Berge knot \(K_0\) such that \(K\) and \(K_0\) yield the same lens space by \(p\)-surgery, and have the same knot Floer homology. The present paper under review makes amazing progress on finite surgery for hyperbolic knots. The main result claims the following. (1) The distance between any two finite surgery slopes is at most \(2\). Hence there are at most \(3\) nontrivial finite surgery surgeries on \(K\). (2) If \(K\) admits \(3\) nontrivial finite surgeries, then \(K\) is the \((-2,3,7)\)-pretzel knot. (3) If \(K\) admits two noncyclic finite surgeries, then such pairs of slopes are listed in the paper. (4) If \(K\) admits two finite surgery slopes at distance \(2\), then such pairs of slopes are also listed. As a byproduct, a new constraint for a prism manifold obtained by Dehn surgery on a knot is given. In addition, \(4m\) and \(4m+4\) are shown to be characterizing slopes for a torus knot of type \((2,2m+1)\). The arguments mainly use \(\mathrm{PSL}_2(\mathbb{C})\)-representations, the correction terms of Heegaard Floer homology, and the Casson-Walker invariant.