Geometric configuration of Riemannian submanifolds of arbitrary codimension (Q1685522): Difference between revisions

If \(E\) is an \(n\)-dimensional real vector space and \(\text{End}(E)\) is the vector space of endomorphisms of \(E\), \(u=(u_1,\dots,u_q)\in\mathbb N^q\) is a multi-index with length \(|u|=u_1+\dots+u_q\), \(A=(A_1,\dots,A_q)\in\text{End}^q(E)\), \(t=(t_1,\dots,t_q)\in\mathbb R^q\), \(u\in\mathbb N^q\), \(tA=t_1A_1+\dots+t_qA_q\) with \(t^u=t_1^{u_1}\dots t_q^{u_q}\), then the musical functions \(\alpha_\flat, \alpha^\sharp:\mathbb N^q\to\mathbb N^q\) are defined as \(\alpha_\flat(i_1,\dots,i_q)=(i_1,\dots,i_\alpha-1,\dots,i_q)\) if \(i_\alpha\geq 1\) or \(0\) if \(i_\alpha=0\), and \(\alpha^\sharp(i_1,\dots,i_q)=(i_1,\dots,i_\alpha+1,\dots,i_q)\) for \(\alpha\in\{1,\dots,q\}\). The generalized Newton transformation is a system of endomorphisms \(T_u=T_u(A)\) satisfying \(T_0=I\), \(T_u=\sigma_uI-\sum\limits_\alpha A_\alpha T_{\alpha_\flat(u)}=\sigma_uI-\sum\limits_\alpha T_{\alpha_\sharp(u)}A_\alpha\) for \(u\in\mathbb N^q\), where \(\sigma_u\) are the coefficients of the Newton polynomial \(P_A:\mathbb R^q\to\mathbb R\) given by \(P_A(t)=\det(I+tA)=\sum\limits_{|u|\leq n}\sigma_ut^u\). If \(\overline{M}^{n+q}\) is a \((n+q)\)-dimensional connected and orientable Riemannian manifold with metric \(\langle\;,\;\rangle\) and Levi-Civita connection \(\nabla\), \(P^n\) is an oriented connected \(n\)-dimensional submanifold of \(\overline{M}^{n+q}\), \(\Sigma^{n-1}\) an \((n-1)\)-dimensional compact hypersurface of \(P^n\), and \(\Psi:M^n\to\overline{M}^{n+q}\) an oriented connected and compact submanifold of \(\overline{M}^{n+q}\) with boundary \(\partial M^n\), then \(M^n\) is said to be a submanifold of \(\overline{M}^{n+q}\) with boundary \(\Sigma^{n-1}\) if the restriction of \(\Psi\) to \(\partial M^n\) is a diffeomorphism onto \(\Sigma^{n-1}\). In this paper, the authors consider the problem of describing the geometry of \(M^n\) along its boundary \(\partial M^n\) with respect to the geometry of the inclusion \(\Sigma^{n-1}\subset P^n\). They give a partial solution of this problem. If the submanifold \(P^n\) is totally umbilical in \(\overline{M}^{n+q}\), then the following formula holds along the boundary \(\partial M^n\): \[ \langle T_u\nu,\nu\rangle=\frac1{n-1-|u|}\sum\limits_{l\leq u}\binom{n-1-|l|}{|u|-l}\rho^l\mu^{u-l}\sigma_{|l|}(A_\Sigma), \] for any multi-index \(u=(u_1,\dots,u_q)\) with length \(|u|\leq n-1\), where \(T_u\) stands for the family of the generalized Newton transformations associated to the matrix \(A=(A_1,\dots,A_q)\), and \(A_\alpha\) is a system of matrices of the shape operators corresponding to a local orthonormal frame of the manifold \(M^n\) and \(A_\Sigma\) is the shape operator of the inclusion \(\Sigma^{n-1}\subset P^n\), \(\sigma_r(A_\Sigma)\) is the elementary symmetric function of the eigenvalues of \(A_\Sigma\) and \(\rho=(\rho_1,\dots,\rho_q)\), \(\mu=(\mu_1,\dots,\mu_q)\) stand for vectors depending on the geometric configuration and the umbilicity factor of the submanifold \(P^n\). Also, the authors show that the ellipticity of the generalized Newton transformation \(T_u\) implies the transversality of the \(q\)-codimension submanifolds \(M^n\) and \(P^n\) along the boundary \(\Sigma^{n-1}\) when \(P^n\) is totally geodesic submanifold of \(\overline{M}^{n+q}\). This paper generalizes a similar result by \textit{L.~J.~Alías} and \textit{J.~M.~Malacarne} in [Rev. Mat. Iberoam. 18, No. 2, 431--442 (2002; Zbl 1038.53060)] for codimension one.