Comparing virtual fundamental classes: Gauge theoretical Gromov-Witten invariants for toric varieties (Q2565952)

The Hamiltonian Gromov-Witten invariants have been independently defined in [\textit{K. Cieliebak, A. R. Gaio} and \textit{D. A. Salamon}, Int. Math. Res. Not. 2000, No. 16, 831--882 (2000; Zbl 1083.53084)] and [\textit{I. Mundet i Riera}, Topology 42, 525--553 (2003; Zbl 1032.53079)], and further investigated by these authors in [J. Symplectic Geom. 1, 543--645 (2002)]. In their previous article [Commun. Math. Phys. 227, 551--585 (2002; Zbl 1037.57025)], the authors have adopted a gauge theoretical point of view to define what they called the gauge theoretical Gromov-Witten invariants. In all these approaches one associates numerical invariants to Hamiltonian actions of compact, connected Lie groups on symplectic manifolds. Similarly as in Gromov-Witten theory, whenever the symplectic manifold is quasi-projective, and the symplectic form is taken to be the Kähler form on it, one wishes to use algebraic techniques for defining and computing the gauge theoretical/Hamiltonian Gromov-Witten type invariants. The problem which appears is that typically the relevant moduli spaces in the algebraic setting are oversized, in which case one has to use virtual fundamental classes (see \textit{K.~Behrend} and \textit{B.~Fantechi} [Invent. Math. 128, 45--88 (1997; Zbl 0909.14006)]) in order to define the invariants. Assuming that this problem is solved, one obtains two apparently different sets of invariants: those which are defined using the virtual fundamental class in the sense of \textit{R. Brussee} [New York J. Math. 2, 103--146 (1996; Zbl 0881.53057)], and those defined by means of the virtual fundamental class in the sense of Behrend-Fantechi. In the paper under review the authors state a conjectural, very general principle. It roughly says that whenever one has a Kobayashi-Hitchin correspondence between a moduli space of irreducible solutions of a Hermitian-Einstein type equation of Fredholm type on one hand, and the moduli space of stable complex geometric objects on the other hand, the Kobayashi-Hitchin correspondence maps the gauge theoretical virtual fundamental class (in the sense of Brussee) to the Behrend-Fantechi virtual fundamental class. The main result of the article is that this conjecture holds in the case of linear torus actions on affine spaces, for which the ring of invariants is trivial. Using this fact, one may compute the gauge theoretical/Hamiltonian Gromov-Witten invariants for toric manifolds with purely algebraic methods. A detailed treatment of the toric case, containing explicit computations, has been performed by the reviewer in [Math. Z. 252, 157--208 (2006; Zbl 1087.14037)].