Intersection cohomology of quotients of nonsingular varieties

Other algebro-geometric (co)homologies (e.g., intersection, equivariant, Lawson, Deligne (co)homologies) (14F43) Vector bundles on curves and their moduli (14H60) Group actions on varieties or schemes (quotients) (14L30)

Abstract: The purpose of this paper is to provide a way to compute the intersection cohomology of the GIT quotient of a nonsingular projective variety. We show that the middle perversity intersection cohomology of the GIT quotient

M / / G

is naturally embedded into the

G

equivariant cohomology of

M^{s s}

under an assumption satisfied by many interesting spaces. This embedding enables us to compute the intersection pairing in terms of the cup product structure of the equivariant cohomology ring. We also show that the image is a subspace defined by truncation. Overall, we can compute the intersection cohomology as a graded vector space with a nondegenerate pairing.

Review text

This article deals with the problem of giving an explicit calculation of the middle intersection cohomology of the singular quotient of a nonsingular projective variety \(M\subset\mathbb P^n\) by an action of a connected complex reductive group \(G=K^{\mathbb C}\). Let \(M^{ss}\) be the set of semi-stable points in \(M\). If the action is locally free then \(X\) is an orbifold and there is a custom way for recovering the usual cohomology \(H_{*}({X})\) in terms of the equivariant cohomology \(H_K^{*}({M^{ss}})\). The above situation becomes richer when the action is not locally free; in that case \(X\) is not anymore an orbifold, but a stratified pseudomanifold; and the equivariant cohomology \(H_K^{*}({M^{ss}})\) is not enough in order to compute directly the intersection cohomology of \(X\). In order to solve this; the author considers the quotient map \(\phi:M^{ss}\rightarrow X\) and constructs a map \[ \widehat\phi:IH_{*}^{\overline{m}}({X})\rightarrow H_K^{*}({M^{ss}}) \] which, under some geometrical conditions, turns out to be a left inverse of the Kirwan map \[ k:H_K^{*}({M^{ss}})\rightarrow IH_{*}^{\overline{m}}({X}) \] where \(\overline{m}\) is the middle perversity in \(X\). The geometrical approach uses two facts: First we have an inequality which establishes a relation between the dimension of the orbits and the number of weights of a maximal torus action near the unstable strata. Broadly speaking, this matter of weight is called the condition of almost balance. Second, there is a way for preserving this balance condition when taking quotients, by considering the orbit-types of non-stable strata. This structure must be locally read transversally to the orbits, up to conjugacy classes. This is the condition of weak balance. The article is organized as follows: Section 1 is an illustrative foreword; in section 2 the author presents the basic geometric and algebraic tools. Section 3 is devoted to the definition of placid maps, a class of stratified maps that preserve prescribed perversities; and to show how these maps preserve the intersection pairing. This is the tool that will allow to relate the middle intersection cohomology of \(X\) with the equivariant cohomology of the non-stable strata. In sections 4 and 5 the author describes the orbit type stratification of the quotient \(X\), and establishes the condition of almost balance. For each stratum \(S\) in \(X\) we have \[ \overline{m}({\text{codim}}(S))\leq\overline{t}({\text{codim}}(\phi^{-1}(S))) \] where \(\overline{t}\) is the top perversity. In other words, \(\phi\) is compatible with respect to the perversities \(\overline{m},\overline{t}\); or \((\overline{t},\overline{m})\)-placid. This will be used in order to construct the arrow \(\hat\phi\) mentioned before. Sections 6 and 7 are devoted to the study of the map \(\hat\phi\). Under the above situation; he computes this morphism, which allows him to present the middle intersection cohomology of \(X\). The article is devoted to the study of singular actions on projective varieties. It is both a work on geometry and algebraic topology. The author deals not only with the geometrical conditions for having intersection cohomology, but also with the calculation of the middle cohomology of the quotient stratified pseudomanifold induced by the action. I consider that this work contributes to enlarge our knowledge of singular actions of Lie groups on stratified spaces. It can be related with the research line that other mathematicians have followed in the last years, such as \textit{M. Goresky, R. Kottwitz} and \textit{R. MacPherson} [Invent. Math. 131, No. 1, 25--83 (1998; Zbl 0897.22009)], \textit{G. Hector} and \textit{M. Saralegi} [Trans. Am. Math. Soc. 338, 263--288 (1993; Zbl 0784.57019)], \textit{F. Kirwan} [J. Am. Math. Soc. 1, 385--400 (1988; Zbl 0645.14021)], \textit{G. Padilla} [Indag. Math., New Ser. 15, 383--412 (2004; Zbl 1072.35219)] and \textit{R. Popper} [Ill. J. Math. 44, 1--19 (2000; Zbl 0949.55004)].

Recommendations

Cites work

Cited in

(19)

This page was built for publication: Intersection cohomology of quotients of nonsingular varieties

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1434285)