On embeddings of homogeneous spaces with small boundary (Q855343): Difference between revisions

Let \(G\) be a connected linear algebraic group over an algebraically closed field \(k\) of characteristic zero and \(H\subset G\) be an algebraic subgroup. A small \(G/H\)-embedding is an irreducible \(G\)-variety \(X\) containing \(G/H\) as an open orbit and such that \(\text{codim}X\setminus(G/H)\geq2\). The paper develops a theory of small embedding under some technical restrictions: \(G\) is supposed to have trivial Picard group and trivial character lattice (e.g., be semicircle and simply connected) and \(H\) be a connected subgroup such that \(G/H_1\) is quasiaffine and \(k[G/H_1]\) is finitely generated, where \(H_1\) is the common kernel of all characters \(H\to k^{*}\). Unlike the general theory of equivalent embedding of homogeneous spaces [see \textit{D. Luna} and \textit{Th. Vust}, Comment. Math. Helv. 58, 186--245 (1983; Zbl 0545.14010)], the theory of small embedding leads to explicit classification and concrete constructions. The basic tool is the total coordinate ring \(R\) of \(G/H\), which is the direct sum of the spaces of global sections of reflexive sheaves corresponding to the divisor classes on \(G/H\). In the above setup, \(R=k[G/H_1]\). Now ``maximal'' (in a sense; e.g., complete) small embeddings are realized as good quotients of certain open subsets in \(Z=\text{Spec}R\) by the torus \(H/H_1\). The latter are described in terms of the GIT fan of \(Z\) [cf. the authors' preprint \url{arXiv:math.AG/053107}]. In particular, this description implies that there are only finitely many complete small embedding of \(G/H\). As an application, there are constructed several examples of small embedding: those admitting a structure of a toric variety, projective small embedding with at most \(\mathbb{Q}\)-factorial singularities, homogeneous spaces admitting small equivalent completions but no smooth ones, etc. The paper is written in a comprehensible manner, with a number of clarifying examples.