A general Hilbert-Mumford criterion. (Q1395353)

The author studies the problem of constructing good quotients. If \(G\) is a complex reductive group and \(X\) is a normal algebraic \(G\)-variety, a \(G\)-invariant open subset \(V\) admits a good quotient if there exists an affine \(G\)-invariant morphism \(V\to V//G\) where \(V//G\) is the spectrum of the invariant functions. The main problem is to construct all open subsets which admit a good quotient. These are called good \(G\)-sets. \textit{A. Białynicki-Birula} and \textit{J. Swiecicka} [see e.g. Am. J. Math. 113, 189-201 (1991; Zbl 0741.14031) and Transform. Groups 1, 153--185 (1996; Zbl 0912.14016)] have studied several aspects of this question in a series of articles cited in the references. In particular, they show that if \(T\) is a maximal torus in \(G\), then a good \(G\)-set is a \(G\)-invariant good \(T\)-set. Thus one studies the question of for which good \(T\)-sets \(U\) is the intersection of all the \(G\)-translations of \(U\) a good \(G\)-set? In the present article, the case of \((T,2)\)-maximal subsets is treated. These are good \(T\)-sets \(U\) such that \(U//T\) can be embedded into a toric variety, and it is maximal. That is, \(U\) is not a \(T\)-saturated proper subset of a larger set having the same properties.