Coxeter group in Hilbert geometry (Q2406842): Difference between revisions

Consider the projective sphere \(S(V)=S^d\) consisting of half-lines of \(V=\mathbb{R}^{d+1}\). A proper open set \(\Omega\) in \(S(V)\) is convex if there is an affine chart containing \(\Omega\); it is properly convex if an affine chart contains the closure of \(\Omega\). The author makes a detailed study of actions of discrete subgroups of \(\mathrm{SL}_{d+1}^{\pm}(\mathbb{R})\) on a properly convex \(\Omega\). Such \(\Omega\)'s admit a nice metric and a measure that is invariant by the group \(Aut(\Omega)\). The author considers actions of \(\Gamma\) on \(\Omega\) such that \(\Omega/\Gamma\) has finite volume and such that the actions of all finite index subgroups on \(\mathbb{R}^{d+1}\) are irreducible. In fact, the discrete groups \(\Gamma\) considered are generated by certain sets of reflections across the facets of a special type of projective polytopes \(P\) called Coxeter polytopes. Then, the set \(C = \bigcup_{\gamma \in \Gamma} \gamma(P)\) is properly convex in \(S^d\). The author addresses a number of question like the following ones: (i) What are the possible Zariski closures of such \(\Gamma\)? (ii) Is \(\Omega\) the largest properly convex open set that is preserved? (iii) When is \(\Omega\) the smallest properly convex open set that is preserved? (iv) When does \(\Gamma\) preserve a unique properly convex subset? The author considers special polytopes called perfect Coxeter polytopes defined in terms of a concept known as the link of a polytope at a vertex. The actions being of finite covolume or cocompact etc. are characterized in terms of the links at all vertices. Further, the Zariski closure of the \(\Gamma\)'s considered are shown to be either equal to \(\mathrm{SO}_{d,1}^0(\mathbb{R})\) or to \(\mathrm{SL}_{d+1}(\mathbb{R})\). Several other technical results are obtained that we do not state here. One of the tools used is a procedure of truncating a polytope by cutting a simple vertex and constructing a new polytope -- this was used in the context of hyperbolic geometry by Vinberg. The subject is a very rich one and several questions are still to be answered. The author defers the study of the moduli space of Coxeter polytope for later. The author provides a clear plan and description of the various sections and makes it easier to read this technical paper.