Rigidityofinvariantconvexsetsinsymmetricspaces.PDFVIP

RIGIDITY OF INVARIANT CONVEX SETS IN SYMMETRIC SPACES BRUCE KLEINER AND BERNHARD LEEB Abstract. The main result implies that a proper convex subset of an irreducible higher rank symmetric space cannot have Zariski dense stabilizer. 1. Introduction In this paper we study convex subsets of symmetric spaces, and their stabilizers. The main results show that in the higher rank case convex sets are strongly restricted, and under mild assumptions can only arise from rank 1 constructions. This rigidity phenomenon for convex subsets is yet another example of a rigidity property enjoyed by higher rank symmetric spaces that has no analog for rank 1 symmetric spaces. One can generate a supply of convex subsets of any Hadamard space by starting with geodesic segments, geodesic rays, complete geodesics, and horoballs, and then taking tubular neighborhoods and intersec- tions. When X is a Hadamard manifold with pinched negative curva- ture convex subsets are abundant: by a theorem of Anderson [And83], any closed subset A of the geometric boundary ∂∞X is the limit set of a closed convex subset Y ⊂ X . On the other hand, for general Hadamard spaces (or manifolds) it can be difficult to control the convex hull of even “small” subsets, like the union of three rays. A group Γ of isometries of a Hadamard space X is convex cocompact if there is a Γ-invariant convex subset C ⊂ X with compact quotient C/Γ. Discrete convex cocompact subgroups of the isometry group of hyperbolic 3-space are an important class in the theory of Kleinian groups; basic examples are uniform lattices, Schottky groups and quasi- Fuchsian groups. Analogous examples exist in Isom(Hn ), as well as the isometry groups of other rank 1 symmetric spaces. In a higher ran