Two-symmetric Lorentzian manifolds (Q644217)

A \(k\)-symmetric pseudo-Riemannian space (\(k\geq 1\)) is a pseudo-Riemannian manifold \((M,g)\) for which \(\nabla^kR=0\), but \(\nabla^{k-1}R\neq 0\), where \(R\) is the curvature tensor of \((M,g)\). In case of a positive definite metric, this condition already implies that \(\nabla R=0\), i.e., that the Riemannian manifold at hand is a locally symmetric space, see \textit{K. Nomizu} and \textit{H. Ozeki} [Proc. Natl. Acad. Sci. USA 48, 206--207 (1962; Zbl 0139.39201)] and \textit{S. Tanno} [Ann. Mat. Pura Appl., IV. Ser. 96, 233--241 (1973; Zbl 0277.53013)]. In this paper, the authors determine the local structure of two-symmetric Lorentzian manifolds. More precisely, they show that a locally indecomposable Lorentzian manifold is two-symmetric if and only if locally there exist coordinates such that the metric is of a certain type (that of a special pp-wave). This generalizes the corresponding result in the four-dimensional case, which was obtained by \textit{O. F. Blanco, M. Sánchez} and \textit{J. M. Senovilla} [J. Phys. Conf. Ser. 229, 5 pp. (2010)]. The main step in the proof is to calculate the holonomy group of a two-symmetric Lorentzian manifold \((M,g)\). This implies that \((M,g)\) is a pp-wave, see e.g. Section 5.4 of \textit{A. Galaev} and \textit{T. Leistner} [ESI Lectures in Mathematics and Physics, 53--96 (2008; Zbl 1152.53036)], which already restricts the local form of the metric. Finally, explicit calculations of \(\nabla R\) and \(\nabla^2R\) give the result.