On closures of cycle spaces of flag domains (Q863445)

First some notation: Let \(G_0\) be a noncompact semisimple Lie group, \(K_0\) a maximal compact subgroup, \(G\) and \(K\) are complexifications, \(Z=G/Q\) is a flag manifold (\(Q\) a complex parabolic subgroup of \(G\)), and \(\Omega =G/K\) an affine symmetric space. Let \(D\) be an open \(G_0\)-orbit in \(Z\). We may associate to \(D\) a cycle space \(M_D=G (C_0)\) (where \(C_0\) is the unique \(K_0\)-orbit in \(D\) that is a complex submanifold) and a total space of cycles \(C_q (D)\) (\(q\) is the dimension of the base cycle \(C_0\)). These cycle spaces have been objects of intense study in representation theory and geometry. The main result of the paper is that the cycle space \(M_D\) is closed in \(C_q(D)\). The authors hope that this result will help pave the way to an explicit description (highly desired) of \(C_q(D)\) by expressing \(C_q(D)\) as a Luna slice bundle over \(M_D\). Also, as a corollary to the main result, \(M_D=C_q(D)\) in the case that \(M_D\) is open in \(C_q(D)\). This observation can be useful in cases when the cycles have complex-geometric meaning (e.g., period domains). The authors prove the main theorem by passing to the setting of the Akhiezer-Gindikin universal domain \(U\subset \Omega\). Specifically the main result is a consequence of the following technical theorem proved in the paper: \textit{If \(X\) is an arbitrary \(G\)-equivariant compactification of \(\Omega\), and \(U\) the Akhiezer-Gindikin domain, then the interior of \(Cl_X (U)\) is \(U\).} Schubert incidence geometry and its previous consequences are key ingredients in this technical result, including the important fact that \(M_D\) is biholomorphic to \(U\) in case that \(D\) is not of Hermitian holomorphic type. [For background details, see \textit{G. Fels, A. Huckleberry} and \textit{J. A. Wolf}, Cycle spaces on flag domains, Boston, MA: Birkhäuser (2006; Zbl 1084.22011).]