Hypersurfaces in statistical manifolds (Q1023416)

A statistical structure on a manifold \(M\) consists of a Riemannian metric \(g\) and a torsion-free affine connection \(\nabla\) that satisfies \( (\nabla_Xg)(Y,Z) = (\nabla_Yg)(X,Z)\) for all vector fields \(X\), \(Y\), and \(Z\) on \(M\). The statistical manifold \((M, \nabla, g)\) has constant curvature \(k\) if \[ R^\nabla(X,Y)Z = k(g(Y,Z)X-g(X,Z)Y) \] where \(R^\nabla\) is the curvature tensor of \(\nabla\) and \(X\), \(Y\), and \(Z\) are arbitrary vector fields on \(M\). If \(M\) is flat, the statistical structure is said to be Hessian, and one can define a Hessian curvature tensor. The author introduces a holomorphic version of statistical structures and their curvature. An immersion \(f: M \rightarrow \widetilde M\) between two statistical manifolds \((M,\nabla,g)\) and \((\widetilde M,\widetilde \nabla, \tilde g)\) is called a statistical immersion if \( g = f^\ast \tilde g\) and \(g(\nabla_XY,Z) = \tilde g(\widetilde \nabla_Xf_\ast Y,f_\ast Z)\) for all vector fields \(X\), \(Y\), and \(Z\) on \(M\). In this formula \(\widetilde \nabla\) denotes the connection in the pull--back bundle \(f^\ast T\widetilde M\) induced from the connection \(\widetilde \nabla\) on \(\widetilde M\). The author investigates the consequences of the existence of a co--dimension one statistical immersion under certain assumptions on \(M\) and \(\widetilde M\). Assuming that \(M\) has constant holomorphic curvature and \(M\) has constant curvature, the existence of a co--dimension one statistical immersion \(f :M \rightarrow \widetilde M\) implies that \(M\) is holomorphically flat. With further assumptions concerning the shape operator of \(f\), \(\widetilde M\) must be flat as well. For the second result, one assumes that \(\widetilde M\) is Hessian with constant Hessian curvature \(\tilde c\) and that \(M\) is Hessian and \(\nabla\) is the Levi--Civita connection of \(g\) and concludes that \(\tilde c \geq 0\).