Cosmetic surgeries on knots in \(S^3\) (Q499880)

For a knot in a \(3\)-manifold, take two different slopes on the knot. If the resulting manifolds obtained by Dehn surgery along these slopes are homeomorphic, then the two surgeries are said to be cosmetic. In particular, they are purely cosmetic, if the resulting manifolds are homeomorphic as oriented manifolds, and chirally cosmetic otherwise. Chirally cosmetic surgeries can happen even for knots in the \(3\)-sphere, but purely cosmetic surgeries are expected to be very rare. For example, genus one knots in the \(3\)-sphere do not admit purely cosmetic surgeries. The main result in the paper under review is to give several restrictions on purely cosmetic surgeries for knots in the \(3\)-sphere. More precisely, let \(K\) be a nontrivial knot in the \(3\)-sphere, and let \(r_1\) and \(r_2\) be two distinct slopes. If \(r_1\)-surgery and \(r_2\)-surgery on \(K\) yield homeomorphic manifolds as oriented manifolds, then \(r_1=-r_2\). Moreover, if \(r_1=p/q\), where \(p\) and \(q\) are coprime, then \(q^2\equiv -1\pmod{p}\), and the concordance invariant \(\tau\) defined by Ozsváth-Szabó and Rasmussen vanishes for \(K\). The argument mainly uses correction terms in Heegaard Floer homology.