A structure theorem for the Gromov-Witten invariants of Kähler surfaces (Q2463361)

This excellent paper goes a long way towards merging the Gromov-Witten theory and the minimal model program in dimension two. In particular, it is shown that for compact Kähler surfaces of positive geometric genus two major hopes are realized: Gromov-Witten invariants can be expressed as sums of local contributions from curves, and they transform straightforwardly under birational transformations (blow-ups). The last circumstance reduces their computation to minimal models: Abelian varieties, \(K3\) surfaces, properly elliptic or general type surfaces, where it can be carried out explicitly. There is a mild technical assumption that the canonical divisor has a representative with smooth components but it can probably be removed by further analysis. The main result is a structure theorem that expresses the Gromov-Witten potential as a sum of a universal degree zero contribution (computed by Kontsevich and Manin), minimal model dependent contribution and local contributions from exceptional curves in the canonical divisor. The latter correspond to Gromov-Witten potentials of some standard surfaces. They are given by universal power series that are computed in the paper. This structure theorem explains the famous result of Taubes on the topological nature of Gromov-Witten invariants in dimension two. For the surfaces under consideration the required topological information is the cohomology pullback induced by the inclusion of the canonical divisor and the parity of the holomorphic Euler characteristic. The techniques used are essentially symplectic involving non-integrable perturbations of complex structures and geometric analysis of non-linear Cauchy-Riemann equations. For example, a key lemma asserts that the complex structure can be deformed to force images of (pseudo)holomorphic curves to lie in a support of the canonical divisor. This directly implies strong vanishing results for Gromov-Witten invariants, in particular vanishing of all non-trivial invariants for Abelian varieties and \(K3\) surfaces. The Kähler condition is however essential as shown by symplectic counterexamples to localization due to Taubes and McMullen. The paper is very well written and is a pleasure to read.