Compact Kähler quotients of algebraic varieties and geometric invariant theory (Q968810)

This is a continuation of the previous work of the author [Trans. Am. Math. Soc. 362, No. 6, 3243--3271 (2010; Zbl 1216.14045)]. Let \(G\) be a complex reductive group with a linearized action on a smooth polarized complex manifold \(X\). Let \(K\) be the maximal compact subgroup of \(G\). The action of \(K\) is Hamiltonian. If one considers the induced moment map \(\mu: X\rightarrow {Lie}(K)^*\), then it is well known from geometric invariant theory (GIT) that the two quotients \(X/\!\!/G\) and \(\mu^{-1}(0)/K\) are symplectomorphic. In the paper under review, the author studies the more general case of a complex algebraic irreducible \(G\)-variety \(X\) equipped with a symplectic form (with respect to which the action of \(K\) is still Hamiltonian) which is no longer necessarily the curvature form of a connection on an ample line bundle. Let us give some details. Recall that \(X\) is a \(G\)-variety if the action map \(G\times X\rightarrow X\) is regular. Furthermore, it is irreducible if \(G\) acts transitively on the set of irreducible components of \(X\). In this setting, one can still define the set \(X(\mu)\) of semistable points. They are the points whose closure of the \(G\)-orbit meets the \(0\)-level set of the moment map. Now, one says that \(X\) has only 1-rational singularities if for any resolution of singularities \(f:\tilde{X}\rightarrow X\), the sheaf \(R^1 f_* \mathcal{O}_{\tilde{X}}\) vanishes. The main result of this paper says that if the level set of the moment map is compact and non empty, and \(X\) has at worst 1-rational singularities, then one has the following nice consequences, the first two of which were proved in the previous work of the author [loc. cit.]: The set \(X(\mu)\subset X\) is Zariski open, the analytic Hilbert quotient \(X(\mu)/\!\!/G\) is a projective algebraic variety, the map \(X(\mu)\rightarrow X(\mu)/\!\!/G\) is a good quotient, and finally, \(X(\mu)\) coincides with the GIT-semistable set for the linearization of the \(G\)-action with respect to a certain Weil divisor. The generalization of the geometric invariant theory for linearizations induced by Weil divisor has been studied by \textit{J. Hausen} [Compos. Math. 140, No. 6, 1518--1536 (2004; Zbl 1072.14057)]. The finiteness of the momentum map quotients for a given \(G\)-variety is also discussed. The author shows that there exist only finitely many subsets of \(X\) that can be realized as the set of semistable points with respect to some \(K\)-invariant Kähler structure and some momentum map \(\mu\) with compact zero fibre \(\mu^{-1}(0)\). A refinement of the main result is given in the case the group \(G\) is semi-simple. Finally, to sum up, this nice work shows that momentum map quotients of algebraic Hamiltonian \(G\)-varieties have very strong algebraic properties.