Pretzel knots with \(L\)-space surgeries (Q274132)

The purpose of the paper under review is to determine all pretzel knots that admit Dehn surgeries creating \(L\)-spaces. An \(L\)-space is a rational homology sphere whose Heegaard Floer homology is the simplest. The result claims that \((2,n)\)-torus knots, and \((-2,3,q)\)-pretzel knots with odd \(q\geq 1\) are the only pretzel knots that admit \(L\)-space surgeries. It was known that these knots admit such surgeries, so the paper is devoted to showing that no other pretzel knot admits an \(L\)-space surgery. The argument goes as follows. Any knot admitting an \(L\)-space surgery is fibered by \textit{Y. Ni} [Invent. Math. 170, No. 3, 577--608 (2007; Zbl 1138.57031); erratum ibid. 177, No. 1, 235--238 (2009)], and moreover, its Alexander polynomial has a strong constraint by \textit{P. Ozsváth} and \textit{Z. Szabó} [Topology 44, No. 6, 1281--1300 (2005; Zbl 1077.57012)]. For example, any coefficient is \(\pm 1\). Finally, fibered pretzel links have been classified by \textit{D. Gabai} [Comment. Math. Helv. 61, 519--555 (1986; Zbl 0621.57003)]. Following Gabai's classification, pretzel knots are classified into three classes. The proof effectively uses this classification, the easy observation that if the determinant \(\det K\) of a knot \(K\) is greater than \(2g(K)+1\), where \(g(K)\) is the genus, then the Alexander polynomial has some coefficient \(a_s\) with \(|a_s|>1\), and the state sum formula of the Alexander polynomial. Interestingly, there exists a unique fibered pretzel knot such that its Alexander polynomial does not obstruct it from admitting an \(L\)-space surgery but it does not admit such surgery. The knot is the \((3,-5,3,-2)\)-pretzel knot. Knot Heegaard Floer homology is used to eliminate this exception.