On locally homogeneous pseudo-Riemannian compact Einstein manifolds (Q778998)

The authors prove that any standard compact Clifford-Klein form of a simple non-compact Lie group admits at least one Einstein metric. The result emerges and is related to a conjecture by \textit{T. Kobayashi} and \textit{T. Yoshino} [Pure Appl. Math. Q. 1, No. 3, 591--663 (2005; Zbl 1145.22011)] stating that if a homogeneous space \(G/H\) of a connected semisimple non-compact Lie group has a compact Clifford-Klein form, then there is a structure of a Clifford-Klein form on \(G/H\), which is standard. This means that \(G/H\) is obtained by an Onishchik triple \((G,H, L)\) which is standard. On the other hand, using Kobayashi's conjecture as a motivation, there is a general investigation among geometers about the possible Einstein Clifford-Klein forms. So the paper under review makes a useful contribution in this direction. The approach used by the authors is to reduce the problem to finding Einstein metrics on compact homogeneous spaces \(G_u/H_u\), which are dual to non-compact homogeneous spaces \(G/H\). Such duality is provided by a result of \textit{I. Kath} [Transform. Groups 5, No. 2, 157--179 (2000; Zbl 0965.53037)]. Furthermore, by taking into account Onishchik's classification of Onishchik triples which are standard, as well as a classification of Kobayashi-Onishchik spaces, the authors reduce the problem into a short list of homogeneous spaces, for several of which the existence of invariant Einstein metrics is known by older results. Finally, they reduce the problem to the task of existence on invariant Einstein metrics on the homogeneous space \(G/H=\mathrm{SO}(4n)/\mathrm{Sp}(n)\). For this particular space, looking for Einstein metrics in the set of all \(\mathrm{SO}(4n)\)-invariant metrics is quite difficult, so the authors employ a method used by \textit{A. Arvanitoyeorgos} et al. [Can. J. Math. 61, No. 6, 1201--1213 (2009; Zbl 1183.53037)], under which they search for invariant Einstein metrics in a subset determined by a subgroup \(H\subset K\subset G\), where \(K=\mathrm{Sp}(n)\mathrm{Sp}(1)\). They call such metrics ``of centralizer type''. In this subset, the Einstein equation becomes quadratic, hence they prove existence, so the original theorem follows.