Fivebranes and knots (Q658979)

Continuing the line of research for which its author was awarded the 1990 Fields Medal, this paper develops a gauge theoretic approach to Khovanov homology for knots. The reviewer expects this approach to substantially impact future research in the field. The background to the paper is summarized as follows. In 1984, Vaughan Jones determined an invariant of knots \(\mathcal{J}(q,K)\) which is a Laurent polynomial with integer coefficients. This knot invariant is mysterious in that, in Jones' original approach, it seems to have nothing at all to do with topology beyond an incidental-looking appeal to Markov's Theorem; it seems rather to have to do with the representation theory of operator algebras. The mystery was cleared up for physicists by the author's 1989 paper [\textit{E. Witten}, ``Quantum field theory and the Jones polynomial'', Math. Phys. 121, No. 3, 351--399 (1989; Zbl 0667.57005)] in which the Jones polynomial was interpreted as an observable in Chern-Simons theory. In 2000, \textit{M. Khovanov} [``A categorification of the Jones polynomial'', Duke Math. J. 101, No. 3, 359--426 (2000; Zbl 0960.57005)] categorified the Jones polynomial, reinterpreting it as an Euler characteristic of a chain complex, thus explaining why the Jones polynomial is a Laurent polynomial with integer coefficients. The challenge became to interpret Khovanov's homologies as observables in 4-dimensional TQFT's, combining Khovanov's and Witten's approaches, and providing a unified explanation for the Jones polynomial being both a Laurent polynomial with integer coefficients, and also being a topological invariant of a knot. One proposal, due to \textit{S. Gukov}, \textit{A. Schwarz} and \textit{C. Vafa} [``Khovanov-Rozansky homology and topological strings'', Lett. Math. Phys. 74, No. 1, 53--74 (2005; Zbl 1105.57011)], was to reinterpret Khovanov homology as the space of BPS states of M-theory membranes. The paper under review reinterprets the above proposal in the language of gauge theory. The basic idea underlying the paper is to apply electric-magnetic dualities to 3-dimensional Chern-Simons theory, but there is no obvious way to do this directly. Instead, following the author's earlier paper [``A new look at the path integral of quantum mechanics'', in: T. Mrowka (ed.) et al., Perspectives in mathematics and physics: Essays dedicated to Isadore Singer's 85th birthday. Somerville, MA: International Press. Surveys in Differential Geometry Vol. 15, 345--419 (2011; Zbl 1250.81060)], the path integral of Chern-Simons theory is reinterpreted as a path integral of 4-dimensional Yang-Mills theory. The boundary condition of the 4-dimensional TQFT is that of the D3-NS5 system of Type IIB superstring theory with a non-zero theta angle. We would like to apply T-duality to obtain a theory on some \(V\times S^1\) from which the path integral could be interpreted as the trace, but the most naive way to do this fails instructively. So instead, Witten first applies S-duality to convert the D3-NS5 system to a D3-D5 system, and then applies T-duality to obtain the requisite theory, in which the path integral appears as the partition function (Witten index) of a 5-dimensional theory. The Jones polynomial can then be computed by counting solutions of certain elliptic partial differential equations in 4-dimensions, and Khovanov homology can be computed by counting solutions of related equations in \((4+1)\)-dimensions. This computation was subsequently carried out by \textit{D. Gaiotto} and \textit{E. Witten} [``Knot Invariants from Four-Dimensional Gauge Theory'', preprint, \url{arXiv:1106.4789}], rigorizing the above result. A conceptually satisfying alternative to the brane constructions is given in Section 5, beginning from a 6-dimensional \((0,2)\) superconformal field theory and recovering the 5-dimensional theory, then the 4-dimensional theory, then the 3-dimensional theory as topological twists of certain of its reductions.