From the hyperbolic 24-cell to the cuboctahedron (Q982166)

The authors study, by example, 4-dimensional hyperbolic manifolds and their deformation theory. The examples are intended to copy as closely as possible the theory from 3 dimensions of hyperbolic Dehn filling, where, in dimension 3, a finite volume cusped hyperbolic 3-manifold has many nearby incomplete deformations, many of which can be completed and closed by sewing solid tori into the completed cusps. The authors are handicapped, of course, by the severe restrictions imposed by known rigidity theorems: (1) Mostow-Prasad rigidity and Calabi-Weil local rigidity guarantee that there are no nontrivial families of complete, finite volume structures in dimension at least 3; and (2) Garland-Raghunathan rigidity which implies that a finite covolume discrete subgroup of \(\text{Isorm}(\mathbb{H}^n)\), \(n\geq 4\), is isolated, up to conjugacy, in its representation variety. The authors, therefore, have to relax the conditions of finite covolume or completeness. They begin with the hyperbolic polyhedron known as the hyperbolic 24-cell formed as follows: In 4-dimensional Euclidean space \(\mathbb{R}^4= \{0\}\times \mathbb{R}^4\subset\mathbb{R}^5\) with standard basis vectors \(e_1\), \(e_2\), \(e_3\), \(e_4\), take the convex hull \(H_1\) of the 16 vectors \(\pm 2\cdot e_i\). Then take the 24 midpoints \(v_j\) of the 24 edges of \(H_1\). To each of these vectors fix a fifth coordinate of \(\sqrt{2}\) to obtain a vector \(w_j\) in \(\mathbb{R}^5\) (for example, the vector \((1,1,0,0)\), which is one of the \(v_j\), gives rise to the vector \((\sqrt{2},1,1,0,0)\)). Viewing \(\mathbb{R}^5\) is as Minkowski space, the vector \((\sqrt{2},1,1,0,0)\) has squared norm \(-\sqrt{2}^2+ 1^2+1^2+ 0^2+ 0^2= 0\) and represents an ideal point of \(\mathbb{H}^4\). The convex hull of these ideal points in \(\mathbb{H}^4\) is the hyperbolic 24-cell. The authors give a very nice description of the symmetries of this cell, complete with very nice computer-drawn illustrations. Reflections in the walls of this polyhedron generate a discrete, finite volume Coxeter group that does not admit deformations. But the authors remove two of the nonintersecting walls from this polyhedron to form a geometrically finite, infinite-volume polyhedron (hyperbolic orbifold) with its associated Coxeter group. This new group admits a continuous 1-dimensional family of deformations. As polyhedral angles vary through the deformation and approach the original angles, the authors discover an infinite family of 4-dimensional hyperbolic orbifolds, each providing an infinitesimally rigid, infinite covolume, geometrically finite discrete subgroup of \(\text{Isom}(\mathbb{H}^4)\). For each, there is a natural space at infinity that does admit deformations, but these deformations do not extend to the orbifold. The constructions are very explicit and carefully described and illustrated.