Projective deformations of weakly orderable hyperbolic Coxeter orbifolds (Q2517166)

The paper under review computes the deformation space of projective structures in a class of compact hyperbolic Coxeter orbifolds of dimension \(n\geq 3\). A hyperbolic orbifold admits a natural projective structure, induced for instance from the projective model of hyperbolic space, and in general it is difficult to compute the moduli space of its projective structures for \(n\geq 3\). This paper describes the deformation space (a neighborhood in the moduli space) of the projective structure induced from the hyperbolic one for this family of Coxeter orbifolds, which is a relevant contribution. Those Coxeter groups are realized by an \(n\)-dimensional, compact, and hyperbolic polyhedron. The group is generated by reflections on its facets (codimension 1 sides), in particular the orbifold is a polyhedron with mirrors in its facets. Such an polyhedron is simple (every vertex meets precisely \(n\) facets, where \(n\) is the dimension). To each face of codimension 2 (called ridge) one associates a natural number, the order of the rotation that is the product of the adjacent facets. Weak orderability is a way to control the number of those ridges that have order 2. More precisely, it means that one can order the facets so that each one has at most \(n\) other adjacent facets that come later for this order and that their common ridge has order 2. In addition, it is also required that those adjacent facets are in generic position. This genericity always holds true for \(n=3\) [\textit{S. Choi} et al., Geom. Dedicata 159, 125--167 (2012; Zbl 1266.57012)]. Another condition is added for the number of facets \(f\) and ridges \(e\): \(e-n f+n(n+1)/2=0\), which is always satisfied for three dimensional polytopes. Under the above conditions, it is shown that the deformation space of projective structures is a ball of dimension the number of ridges of order at least 3 minus \(n\). There are two particular cases. In dimension three, just weak orderability is required. Another remarkable case is for a polyhedron obtained by iterated truncation on a simplex, where the conditions are always satisfied. For the proof, the deformation space is written as the solution space of the so called Vinberg equations. The authors prove that those equations define a submersion, hence the solution space is smooth, and they compute its dimension (in dimension 3, those have already been described by \textit{L. Marquis} in [Geom. Dedicata 147, 47--86 (2010; Zbl 1243.52015)]). Finally the authors provide examples that show the necessity of these conditions, in particular an example where the deformation space is singular.