Knot Floer homology and the four-ball genus (Q1426933)

Using their previous work on Floer homology, the authors define a concordance invariant \(\tau(K)\) for knots \(K\), which behaves similarly to the classical (Murasugi) signature \(\sigma(K)\). In particular, \(\tau\) is additive under connected sum, changes sign under mirroring, changes by at most \(\pm 1\) under a crossing change, and estimates (from below) the four-ball genus of knots. In fact, \(\tau(K)\) is a multiple of \(\sigma(K)\) for alternating knots \(K\). In general, \(\tau\) can estimate the four-ball genus better than \(\sigma\), as shown by some examples. The four-ball genus of these examples was previously known (see, e.g., the cited papers of \textit{L. Rudolph} [Geom. Topol. Monogr. 2, 555--562 (1999; Zbl 0962.57004)], \textit{H. Kawamura} and \textit{K. Tanaka} and also by myself [Ann. Scuola Norm. Sup. Pisa Cl. Sci. 2(2), 237--285 (2003)]) by the Rudolph-Bennequin inequality. Later \textit{C. Livingston} [arXiv: math. GT/0311036 (2003)] in fact deduced the Rudolph-Bennequin inequality (in the original form of the estimate, without the number of negative Seifert circles; see Rudolph's op.cit.\ article) from \(\tau\). Similarly, \textit{J. Rasmussen} [arXiv: math. GT/0402131 (2004)] defined a signature type invariant \(s(K)\) from Khovanov's homology, and proved the Rudolph-Bennequin inequality from it. He conjectures that his and the invariants of the authors of the paper under review are equivalent.