Strongly self-dual polytopes
Ákos G. Horváth, István Prok
TL;DR
This work broadens Lovász's theory of strongly self-dual polytopes by examining three-dimensional instances beyond the classic $L$-type family and by constructing a non-$L$-type ssd-polyhedron with $23$ vertices. It derives fundamental geometric and duality properties for ssd-polytopes, including a uniform diagonal length $\alpha=\sqrt{2+2r}$ and a duality map $\sigma$ that pairs vertices with orthogonal facets, and leverages these to constrain possible combinatorics. The paper further provides a constructive dual-edge framework via $\Phi_r$, enabling numerical exploration of new ssd-polyhedra from non-regular faces, and presents a concrete non-$L$-type example, together with an eight-vertex classification that confirms the finite nature of small SSD families. The results reveal a richer structural landscape for ssd-polytopes than previously recognized and offer algorithmic tools for discovering new instances with potential implications for self-polar and diameter-constrained configurations on the sphere.
Abstract
This article aims to study the class of strongly self-dual polytopes (ssd-polytopes for short), defined in a paper by Lovász \cite{lovasz}. He described a series of such polytopes (called $L$-type polytopes), which he used to solve a combinatorial problem. From a geometrical point of view, there are interesting questions: what additional elements of this class exist, and are there any with a different structure from the $L$-type ones? We show that in dimension three, one of their faces defines $L$-type polyhedra. Illustrating the algorithm of the proof, we present an ssd-polytope of 23 vertices whose combinatorial structure differ from those of $L$-type ones. Finally, with an elementary discussion, we prove that for fewer than nine vertices, there are only fifth one ssd-polyhedra, four of them can be constructed by Lovász's method, and we can find the fifth one with "the diameter gradient flow algorithm" of Katz, Memoli and Wang \cite{katz-memoli-wang}.
