The GIT aspect of generalized Kähler reduction. I (Q2416862)

Generalized geometry offers generalized versions of some classical structures. These versions contain the classical structure as a particular case but also go beyond it, sometimes encompassing several structures. For instance, symplectic and complex structures both become particular cases of generalized complex structures. A generalized Kähler structure consists then of a pair of commuting generalized complex structures whose composition is equal to (minus) a generalized metric. Theorems also have generalized versions. For instance, Marsden-Weinstein's symplectic reduction is a particular case of generalized complex reduction. This paper aims at describing an aspect of generalized Kähler reduction: namely, an analogue of the Kempf-Ness theorem, which, for `good' Hamiltonian equivariant Kähler manifolds, identifies symplectic and complex quotients. The complex quotient is in the sense of GIT, and this is why the paper bears its title. The main result of this paper is a generalized analogue of the Kempf-Ness Theorem (Theorems 6.5 and 6.6) under certain hypotheses. These hypotheses are that the manifold and the group are compact, the action is locally free on the zero locus of the moment map, and that the action is ``strong Hamiltonian'', which is defined in Section 5 in terms of the Poisson structure associated to one of the generalized complex structures. Under the same hypotheses, it is proved that the stable locus is a generalized holomorphic bundle over the reduction/quotient. As for the rest of the paper, the first sections of the paper are a quick and precise account of the facts needed, including some technical important results in Section 4, whereas the last one (Section 7) shows that a (stable) orbit of the complexified group becomes a Hamiltonian equivariant Kähler manifold. It must be said that the author uses and refers to the generalized moment map as introduced in [\textit{Y. Lin} and \textit{S. Tolman}, Commun. Math. Phys. 268, No. 1, 199--222 (2006; Zbl 1120.53049)] instead of the more established reference [\textit{H. Bursztyn} et al., Adv. Math. 211, No. 2, 726--765 (2007; Zbl 1115.53056)]. This is a relatively short and straightforward paper which, despite all the hypotheses, offers results in a very interesting direction. The author leaves for future work the case when the group does not necessarily act (locally) freely on the zero locus of the moment map.