Equivariant deformations of the affine multicone over a flag variety (Q2480307): Difference between revisions

This paper determines a large class of examples of a certain invariant Hilbert scheme introduced in [\textit{V.~Alexeev} and \textit{M.~Brion}, J. Algebr. Geom. 14, 83--117 (2005; Zbl 1081.14005)] by relating them to (strict) wonderful varieties, a class of algebraic varieties, equipped with an action of a linear semisimple group and satisfying axioms inspired by the well known compactifications of symmetric spaces of \textit{C. De Concini} and \textit{C. Procesi} [in: Invariant theory, Proc. 1st 1982 Sess. C.I.M.E., Montecatini/Italy, Lect. Notes Math. 996, 1--44 (1983; Zbl 0581.14041)]. Early in the paper, the authors nicely illustrate their main results with a few explicit examples. We set up some notation. Let \(G\) be a complex connected semisimple algebraic group and let \(\{\lambda_1, \ldots, \lambda_s\}\) be linearly independent dominant weights (after the choice of a Borel subgroup and a maximal torus in it), such that the monoid \(\Gamma\) they generate in the full monoid of dominant weights \(\Lambda^+\) satisfies the following saturation condition due to \textit{D.~Panyushev} [Ann. Inst. Fourier 47, No. 4, 985--1011 (1997; Zbl 0878.14008)]: \(\mathbb{Z}\Gamma \cap \Lambda^+ = \Gamma\). Here \(\mathbb{Z}\Gamma\) stands for the subgroup generated by \(\Gamma\) in the character group of the maximal torus. The authors show that the invariant Hilbert scheme \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) is isomorphic to an affine space (with a specific linear action by the adjoint torus \(T_{\mathrm{ad}}\) of \(G\)). By definition, this Hilbert scheme parametrizes the \textit{nondegenerate} closed \(G\)-subvarieties \(Y\) of \(V:=V(\lambda_1) \oplus \ldots \oplus V(\lambda_s)\) with the property that as \(G\)-modules \[ \mathbb{C}[Y] \cong \bigoplus_{\lambda \in \Gamma} V(\lambda)^*. \] Here \(V(\lambda)^*\) is the dual of the \(G\)-module associated to the dominant weight \(\lambda\) and a subvariety of \(V\) is called nondegenerate if its projection to every isotypical component of \(V\) is nontrivial. The proof, which uses wonderful varieties, roughly runs as follows. First the authors determine the tangent space \(T_{X_0}\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) to \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) at its `most degenerate point' \(X_0\). This \(X_0\) is the affine multicone of the title. To prove that \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) is smooth they find another closed point \(X_1\) on \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) and, using a result from [\textit{V.~Alexeev} and \textit{M.~Brion}, loc. cit.], they determine the dimension of the (open) \(T_{\mathrm{ad}}\)-orbit on \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) through \(X_1\). As this dimension agrees with that of \(T_{X_0}\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) it follows that \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) is smooth and, again by [\textit{V.~Alexeev} and \textit{M.~Brion}, loc. cit.], that it is isomorphic to an affine space. To find the closed point \(X_1\), the authors proceed as follows, generalizing an idea exploited by \textit{S. Jansou} [J. Algebra 306, No. 2, 461--493 (2006; Zbl 1117.14005)]. They first associate a certain combinatorial datum \((S^p(\Gamma),\Sigma(\Gamma))\), introduced by \textit{D. Luna} [Hautes Études Sci. Publ. Math. 91, 161--226 (2001, Zbl 1085.14039)] and called \textit{spherical system}, to \(\Gamma\). (Actually, Luna's spherical systems are more general. Because \(\Gamma\) satisfies Panyushev's saturation condition the spherical systems considered here are \textit{strict}; cf. the appendix of [\textit{D.~Luna}, J. Algebra 313, No. 1, 292--319 (2007; Zbl 1116.22006)]). Next, they argue that there exists a wonderful variety \(X\) \textit{with} spherical system \((S^p(\Gamma),\Sigma(\Gamma))\), thus confirming a conjecture of Luna [loc.~cit.] for the case of strict spherical systems and strict wonderful varieties. The main part of the proof of the existence of \(X\) is not given in the paper under review; instead the reader is referred to a preprint version [\url{arXiv: math.AG/0603690}]. More recently, the authors expanded this existence proof in [Classification of strict wonderful varieties, \url{arxiv:0806.2263}]. Continuing with the sketch of the proof, the authors also show that \(X\) embeds \(G\)-equivariantly into \(\mathbb{P}(V(\lambda_1)) \times \ldots \times \mathbb{P}(V(\lambda_s))\). The desired subvariety \(X_1\) of \(V\), finally, is the \(G\)-orbit closure of a vector \(v \in V\) which lies in the affine multicone \(\widetilde{X}\) over \(X\) and whose image \([v]\) in \(X\) belongs to the open \(G\)-orbit of \(X\) (wonderful varieties have a unique open orbit). This finishes the overview of the proof. Furthermore, the authors establish that the universal family over \(\mathrm{Hilb}^G_{\underline{\lambda}}(V)\) is given by the quotient morphism \(\mathrm{Spec}(\mathbb{C}[\pi^{-1}(X)]) \to \pi^{-1}(X)/G\), where \(\pi: \widetilde{X} \dashrightarrow X\) is the natural (rational) map.