Let M be a complete hypersurface in S+' (or CP'+ ). Assume that through each point x of M a (local) ,(x)-dimensional totally geodesic submanifold Sx of 1 (or CP'+ 1) exists in M. A sufficient condition for M itself to be totally geodesic is given in terms of ,u(x). Introduction. In this paper, we give a sufficient condition for a hypersurface of Sn+' (or CPn+I in the complex case) to be totally geodesic. Obviously, if a hypersurface M of Sn+' (or CPn+l) is locally totally geodesic everywhere in M, then M is globally totally geodesic. We ask, therefore, what we can say about M if M contains a lower dimensional totally geodesic submanifold (local) through each point of M. The specific statement of this assumption is given as condition (*) (or (* *) for the complex case) in the following sections. Our main results are stated as Theorems A and A'. These theorems basically tell us that we can often conclude that M is totally geodesic, even if the dimensions of the local totally geodesic submanifolds are considerably lower than that of M itself. In particular, the dimensions of local totally geodesic submanifolds in the complex case may be as low as a half of that of M + 1. Our condition (*) (or (**)) may also be regarded as an extremely simplified version of the so-called axiom of spheres for hypersurfaces of Sn+I1 (or CPn+ 1); see Cartan [5] or Leung-Nomizu [6] for this notion. Hypersurfaces of Sn+ . Let us denote by Sn+I the standard (n + 1)-dimensional sphere of constant curvature c. Let M be a complete Riemannian manifold of dimension n. Furthermore, assume that there is an isometric immersionf of M into S' . Call the pair (M, f) a hypersurface of 1. Our condition (*) is stated as follows. (*) Through each point x of M exists a k(x)-dimensional submanifold S,x (2k(x) > n) of M which is mapped under f isometrically into a k(x)-dimensional totally geodesic sphere of 1. Here k(x) is a positive integer-valued function of M. Let (M, f) be as above. Denote by V and V the Riemannian connections of Sn 1 and M, respectively. Then, for any tangent vector fields X and Y of M, we have Vf*(X)f*(Y) = f*(Vx Y) + a(X, Y), Received by the editors February 18, 1980. 1980 Mathematics Subject Classification. Primary 53C40. ? 1981 American Mathematical Society 0002-9939/81/0000-01 72/$02.00
Read full abstract