Sufficient conditions for constant mean curvature surfaces to be round (Q697376)

This work begins with H. Hopf's two proofs that a simply connected immersed constant mean curvature surface is a round sphere, and then extends it in two directions: (1) The first main result is that a compact immersed \(C^{2,\alpha}\) constant mean curvature disk with piecewise \(C^{2,\alpha}\) boundary must be a part of a round sphere if the boundary consists of lines of curvature and has at most three vertices of angle less than \(\pi\). The proof uses a notion of rotation index at the umbilic points of the surface, extended from the corresponding notion in Hopf's second proof. The author provides a counterexample with four boundary vertices to show that the condition ``at most three vertices'' is sharp. This result has an application to capillary surfaces, which are constant mean curvature surfaces that meet all smooth parts of the boundary of some three-dimensional domain at constant angles. Assume that the three-dimensional domain is bounded by planes and spheres. Then the Terquem-Joachimsthal theorem implies that the boundary of the capillary surface consists of lines of curvature, so if it has at most three boundary vertices, it must be part of a round sphere (by the first main result of the paper). The author also gives a result showing that, under certain conditions, unbalanced capillary hypersurfaces cannot exist (``unbalanced'' is defined in the paper). He further gives a motivation for conjecturing that the singular curves of stationary compound soap bubbles are lines of curvature. An affirmative answer to this conjecture would imply that simply connected symmetric double and triple and quadruple bubbles consist only of parts of round spheres. (2) The author also generalizes Hopf's result to hypersurfaces in higher dimensional Euclidean spaces. If a compact immersed constant mean curvature hypersurface in \(n\)-dimesional Euclidean space has constant \(k\)th order mean curvature for some \(k\) between 2 and \(n-1\), then it is a round hypersphere. Although the extra condition on the \(k\)th order mean curvature is added here, it is not assumed that the hypersurface is simply connected, as Hopf had. It is also not assumed that the hypersurface is embedded (as A. Ros did in a similar earlier result).