On the structure of pseudo-Riemannian symmetric spaces (Q848900): Difference between revisions

It is known that the classification of simply connected pseudo-Riemannian symmetric spaces reduces to the classification of Lie algebras \(\mathfrak g\) with an \(ad_{\mathfrak g}\)-invariant pseudo-Euclidean metric \(g\) and a \(g\)-skew-symmetric involutive automorphism \(\theta\) (which is defined by an orthogonal symmetric decomposition of \(\mathfrak g\)). The authors propose a scheme for classification of all such triples \((\mathfrak g, g, \theta)\) in terms of some cohomology which they call quadratic cohomology. More precisely, thay associate with a triple \((\mathfrak g, g, \theta)\) the following objects : {\parindent7mm \begin{itemize}\item[i)] a Lie algebra \(\mathfrak h\) with involutive automorphism of dimension which is much smaller then dimension of \(\mathfrak g\); \item[ii)] a semisimple orthogonal \(\mathfrak h\)-module and \item[iii)] a cohomology class of the quadratic cohomology associated with this module. \end{itemize}} As an application, they obtain a classification of symmmetric spaces of signature \((2,m)\), which complete and correct earlier classification results due to Cahen, Parker and to Neukirchner.