Dehn surgeries and rational homology balls (Q507061)

The main question here is which Dehn surgeries on a knot in the \(3\)-sphere bound rational homology \(4\)-balls. For the unknot, this problem has been settled by \textit{P.~Lisca} [Geom. Topol. 11, 429--472 (2007; Zbl 1185.57006)]. The first result in the present paper is to show that for any knot and any positive integer \(q\), there exist only finitely many positive integers \(p\) such that \(p/q\)-surgery on the knot bounds a rational homology \(4\)-ball. This is proved by using correction terms of lens spaces in Heegaard Floer homology theory. Moreover, it is shown that for any knot, there are at most two positive integral surgeries such that the resulting manifold bounds a rational homology \(4\)-ball, and if there are two, they are consecutive squares. In particular, no positive integral surgery on the \((p,q)\)-torus knot bounds a rational homology \(4\)-ball if \(p,q>1\) and \(p>9q\). Finally, for the family of torus knots of type \((kq\pm 1,q)\) with \(q>1\), all integral surgeries on the knot for which the resulting manifolds bound rational homology \(4\)-balls are determined.