Minkowski type problems for convex hypersurfaces in the sphere (Q938578)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Minkowski type problems for convex hypersurfaces in the sphere |
scientific article |
Statements
Minkowski type problems for convex hypersurfaces in the sphere (English)
0 references
26 August 2008
0 references
The classical Minkowski problem is: to find a strictly convex closed hypersurface \(M\subset\mathbb{R}^{n+1}\) such that its Gaussian curvature \(K\) equals a given function \(f\) defined in the normal space of \(M\) (or equivalently defined on \(S^n\)): \(K|_M= f(\nu)\). This problem has been solved in full generality by \textit{S. Y. Cheng} and \textit{S. T. Yau} [Commun. Pure Appl. Math. 29, 495--516 (1976; Zbl 0363.53030)]. Instead of prescribing Gaussian curvature other curvature functions \(F\) can be considered, i.e., one studies the problem \(F|_M= f(\nu)\). If \(F\) is one of the symmetric polynomials \(H_k\), \(1\leq k\leq n\), this problem has recently been solved by \textit{B. Guan} and \textit{P. Guan} [Ann. Math. (2) 156, 655--673 (2002; Zbl 1025.53028)]. In the present paper this problem is considered for strictly convex, closed hypersurfaces \(M\subset S^{n+1}\) and for curvature functions \(F\) the inverses of which are of class \((K)\) which was introduced in the author's paper [Indiana Univ. Math. J. 49, 1125--1153 (2000)]. \(M\) is considered as an embedding in \(\mathbb{R}^{n+2}\) of codimension 2, so that the tangent spaces \(T_x(M)\) and \(T_x(S^{n+1})\) can be identified with subspaces of \(T_x(\mathbb{R}^{n+2})\); the image of the point \(x\) under the Gauss map is exactly the normal vector \(\overline x=\nu\in T_x(S^{n+1})\). The equation of the problem can be written in the form \(F|_M= f(\overline x)\), where \(f\) is given as a function defined in \(S^{n+1}\). There is a dual problem, namely \(\overline F|_{M^*}= f^{-1}(\overline x)\), where \(\overline F\) is the inverse of \(F:\overline F(\kappa_i)= [F(\kappa^{-1}_i)]^{-1}\). Further, \(M_1\) is said to be a lower barrier for the pair \((F,f)\), if \(F|_{M_1}\leq f\), and \(M_2\) is called an upper barrier for \((F,f)\), if \(|_{M_2}\geq f\). The main result is as follows. Let \(F\in C^5(\Gamma_+)\) be a symmetric, positively homogeneous and monotone curvature function such that its inverse \(\overline F\) is of class \((K)\), let \(0< f\in C^5(S^{n+1})\), and assume that there exist upper and lower barriers for \((F,f)\) in the hemisphere, where in addition the barriers \(M_i\) should bound a connected open set \(\Omega\) such that the mean curvature vector of \(M_1\) should point to the exterior of \(\Omega\) and the mean curvature vector of \(M_2\) should point into \(\Omega\). Then the dual problems \(F|_M= f(x)\) and \(\overline F|_{M^*}= f^{-1}(\overline x)\) have strictly convex solutions \(M\) resp. \(M^*\) of class \(C^{6,\alpha}\), \(0<\alpha< 1\), such that their convex bodies are strictly contained in the open hemispheres.
0 references
Minkowski type problems
0 references
Gauss map
0 references
mean curvature vector
0 references