Noncompact homogeneous Einstein spaces (Q1266339)

The paper under review is a fundamental work on homogeneous Einstein manifolds of negative Ricci curvature. The author develops a new approach based on two techniques: modified scalar curvature functionals and invariant theory. Using his general theory, he obtains many new results about left-invariant Einstein metrics on solvable Lie groups. It is important to note that all known examples of homogeneous Einstein manifolds \(M\) of negative scalar curvature can be presented in the form \((S,g)\), where \(S\) is a simply connected solvable Lie group and \(g\) a left-invariant Riemannian metric on \(S\). Such a pair \((S,g)\) is called a solvmanifold. It is completely determined by the corresponding Lie algebra \(s= \text{Lie } S\) together with the scalar product \(\langle \cdot , \cdot \rangle\) on \(s\) induced by \(g\). The solvmanifold \((S,g)\) is called standard if the orthogonal complement of \(n= [|s,s]\) is Abelian. This condition is satisfied for all known Einstein solvmanifolds and the author proves that it is an open condition on the Lie algebra's structure constants. Moreover, he proves that for standard Einstein solvmanifolds \(M\) there exists an Iwasawa normal form, i.e., a particularly nice choice of simply transitive group of isometries of \(M\). Using the above normal form, to any standard Einstein solvmanifold he associates certain integral invariants, which he calls the eigenvalue type. He proves that in every dimension only finitely many eigenvalue types can occur. By his main theorem, the moduli space of standard Einstein solvmanifolds of given eigenvalue type is open (with respect to the \(C^\infty\)-topology) in the moduli space of all Einstein solvmanifolds of given dimension. His uniqueness theorem shows that a solvable Lie group admits at most one standard left-invariant Einstein metric up to isometry and scaling. Then he investigates the fine structure of the moduli space of standard Einstein solvmanifolds of given eigenvalue type. Due to a reduction argument, he restricts to solvmanifolds of rank one, i.e., he assumes that the orthogonal complement of \(n= [|s,s]\) is 1-dimensional: \(n^\perp= \mathbb{R} A\). The eigenspace decomposition of \(A\) on \(n\) is determined by the given eigenvalue type. The author introduces a reductive algebraic group \(G \subset GL(n)\) which naturally acts on the variety \(V\subset \wedge^2 n^*\otimes n\) of Lie brackets compatible with the above eigenspace decomposition. The most beautiful of his theorems establishes a natural 1-1-correspondence between closed \(G\)-orbits in \(V\) and standard Einstein solvmanifolds of given eigenvalue type and of rank one. Finally, as an application, he describes explicitly the moduli space of Einstein solvmanifolds near the noncompact rank one symmetric spaces. In particular, he computes its (local) dimension, which is, e.g., \(8m^2-6m-8\) in the case of quaternionic hyperbolic \((m+1)\)-space.