On a problem of Osserman in Lorentzian geometry (Q1360122): Difference between revisions

If \((M,g)\) is a pseudo-Riemannian manifold with curvature tensor \(R\), then for every tangent vector \(v\in T_pM\) the selfadjoint operator \(R_v: (\mathbb{R} v)^\perp \to (\mathbb{R} v)^\perp\), \(x\mapsto R(x,v)v\), is called the associated Jacobi operator. Osserman conjectured that a Riemannian manifold \((M,g)\) is a locally symmetric space of rank one, if for unit vectors \(v\) the eigenvalues of the associated Jacobi operators do not depend on \(v\). The authors investigate the analogous problem for Lorentzian manifolds, which is very different from the Riemannian situation with respect to the methods and results. For instance, one has to treat the cases of the different causal character of \(v\) seperately (timelike, spacelike and null Osserman spaces). Timelike Osserman spaces and spacelike Osserman spaces of dimension 3 and 4 always have constant curvature. In a second part, some situations are described in which a Lorentzian manifold turns out to be locally a warped product, where the product structure corresponds to a splitting \(M=\text{time} \times\) rest space.