Reconstructing Polytopes and Pseudomanifolds
Joshua Hinman
TL;DR
The paper resolves Grünbaum’s edge–ridge reconstruction question by proving that every $4$-polytope is determined by edge–ridge incidences, and shows that for general $d\ge 3$ not all $d$-polytopes are determined by the $(d-3)$-skeleton together with its dual $(d-3)$-skeleton. It then extends reconstruction phenomena to the simplicial setting: for $d\ge 4$ and $\lceil d/2\rceil \le k \le d-2$, any homology $(d-1)$-manifold is determined by the incidences of its $k$- and $(k-1)$-faces, and this extends to certain normal pseudomanifolds whose $(2d-2k-1)$-dimensional links are homology manifolds. However, not every normal $(d-1)$-pseudomanifold is determined by its $(d-2)$-skeleton, establishing both positive reconstruction results under specific conditions and negative counterexamples in general. The work blends combinatorial polytope theory with topological tools like Alexander duality and Dancis-type reconstruction, providing a unified view of when partial skeleta suffice and when they do not for both polytopes and simplicial complexes.
Abstract
We prove that every 4-polytope is determined by its edge-polygon incidences, solving an open problem of Grünbaum. For each $d \geq 3$, we show that not every $d$-polytope is determined by its $(d-3)$-skeleton and dual $(d-3)$-skeleton together, answering a question of Samper. In the simplicial realm, we prove that for $d \geq 4$ and $\lceil \frac{d}{2} \rceil \leq k \leq d-2$, every homology $(d-1)$-manifold is determined by the incidences of its $k$- and $(k-1)$-faces. For $d \geq 5$ and $\lceil \frac{d+1}{2} \rceil \leq k \leq d-2$, we extend our proof to normal $(d-1)$-pseudomanifolds whose $(2d-2k-1)$-dimensional links are homology manifolds. Finally, we prove that not every normal $(d-1)$-pseudomanifold is determined by its $(d-2)$-skeleton.
