The \(\kappa\)-nullity of Riemannian manifolds and their splitting tensors (Q6046831)

Recall that for a Riemannian manifold \((M,g)\) the Riemann curvature tensor \(R\) is given by \(R(X,Y)Z = \nabla_X \nabla_Y Z - \nabla_Y \nabla_X Z - \nabla_{[X,Y]} Z\) for smooth vector fields \(X\), \(Y\), \(Z\), where \(\nabla\) is the Levi-Civita connection of \(g\). According to \textit{S.-s. Chern} and \textit{N. H. Kuiper} [Ann. Math. (2) 56, 422--430 (1952; Zbl 0052.17601)] the nullity space at \(p \in M\) is defined by \begin{align*} \mathcal{N}|_p = \{z \in T_p M | R_p(x,y)z = 0 \text{ for all }x,y \in T_p M\}, \end{align*} and the index of nullity at \(p\) is given by \(\nu(p) = \mathrm{dim}\,\mathcal{N}|_p\). Let \(A_p(x,y)z \equiv g_p(y,z)x-g_p(x,z)y\) and \(B_p(x,y)z \equiv R_p(x,y)z-\kappa A_p(x,y)z\). For \(\kappa \in \mathbb{R}\), following \textit{T. Ôtsuki} [J. Math. Soc. Japan 6, 221--234 (1954; Zbl 0058.37702)], we similarly define the \(\kappa\)-nullity distribution at \(p\) by \begin{align*} \mathcal{N}_\kappa|_p = \{z \in T_p M | B_p(x,y)z = 0 \text{ for all }x,y \in T_p M\}, \end{align*} and the index of \(\kappa\)-nullity at \(p\) by \(\nu_\kappa(p) = \mathrm{dim}\,\mathcal{N}_\kappa|_p\) to yield a distribution \(\mathcal{N}_\kappa\) and an upper semi-continuous function \(\nu_\kappa\) on \(M\) respectively. The orthogonal distribution \(\mathcal{N}^\perp_\kappa\) is said to be the \(\kappa\)-conullity distribution of \(M\) and its dimension at \(p\) defines a lower semi-continuous function on \(M\). Since the Levi-Civita connections for \(g\) and \(|\kappa|g\) coincide, the \(\kappa\)-nullity distribution at \(p\) for \(g\) coincides with the \(\kappa/|\kappa|\)-nullity distribution at \(p\) for \(|\kappa|g\) when \(\kappa \neq 0\). So, with no loss of generality, it suffices to restrict our attention to the cases where \(\kappa=-1\), \(\kappa=0\), and \(\kappa=1\). In this article, the authors consider manifolds of low \(\kappa\)-conullity or \(\kappa\)-nullity where \(\kappa=-1\), \(\kappa=0\), and \(\kappa=1\). They obtain new results for manifolds with low \(\kappa\)-conullity. In particular, the authors show that a Riemannian \(n\)-manifold with \(n\geq3\) with maximal \((-1)\)-conullity equal to 2, (i) Has that the rigid motions of the Minkowski plane \(E(1,1) \equiv \mathrm{SO}_0(1,1) \ltimes \mathbb{R}^2\) is locally isometric to an open set \(U\) of points with \((-1)\)-conullity 2 if \(M\) is of constant scalar curvature and the \((-1)\)-conullity distribution is integrable over \(U\); (ii) Has that the universal cover of \(M\) is homogeneous if \(M\) is complete with finite volume, 3-dimensional or has scalar curvature bounded away from \(-n(n-1)\); (iii) Is isometric to \(E(1,1)\) or the universal cover of \(\mathrm{SL}(2,\mathbb{R})\) with a left-invariant metric if \(M\) is homogeneous and simply connected. The authors prove that a simply connected complete \(n\)-Riemannian manifold \(M\) with maximal 0-conullity equal to 2 and positive scalar curvature bounded away from 0 splits as a Riemannian product \(\mathbb{R}^{n-1} \times \Sigma\), where \(\Sigma\) is diffeomorphic to a 2-sphere. Moreover, the authors show that the only simply connected complete Riemannian manifolds with constant 1-conullity equal to 2 and constant scalar curvature are the 3-dimensional Sasakian space forms, namely isometric to 3-dimensional Lie groups with certain left-invariant metrics: the Berger sphere \(\mathrm{SU}(2)\), the universal cover of \(\mathrm{SL}(2,\mathbb{R})\), or the Heisenburg group, \(\mathrm{Nil}^3\). The proofs concerning Riemannian manifolds with maximal conullity 2 are laid out in Section 3 of the article and all heavily rely on \textit{A. Rosenthal}'s idea of the splitting tensor \(C_T\) [Mich. Math. J. 15, 433--440 (1968; Zbl 0177.24501)]. In the context of scalar curvature that is constant (or bounded away from \(\kappa\)) the authors show in Section 2 that the splitting tensor \(C_T\) for \(T \in \mathcal{N}_\kappa\) and \(||T||=1\) has that \(\mathrm{tr}\, C_T=0\) and \(\det C_T = \kappa\) along the geodesic \(\gamma\) where \(\dot{\gamma}=T\) for cases \(\kappa = -1\), \(\kappa = 0\), and \(\kappa = 1\). The results follow from the heavy restrictions placed on the matrix representation of \(C_T\).