Reconstructing a simple polytope from its graph

From MaRDI portal





A theorem of \textit{R. Blind} and \textit{P. Mani-Levitska} [Aequationes Math. 34, 287-297 (1987; Zbl 0634.52005)] says that the combinatorial type of a simple polytope is already determined by its graph. \textit{G. Kalai} gave a constructive proof [J. Comb. Theory, Ser. A 49, No. 2, 381-383 (1988; Zbl 0673.05087)], which has a horrendous time-complexity though. In the paper under review the authors suggest a reconstruction algorithm which is based on Kalai's ideas combined with techniques from combinatorial optimization. Empirically, this yields a faster method, although the algorithm still requires exponential running time. The paper also reports on the results of experiments on various classes of simple polytopes in dimensions up to~\(8\). A key concept in the algorithm is the notion of a quasi-facet of a simple \(d\)-polytope, that is, a \((d-1)\)-regular, connected, and non-separating subgraph. Perles conjectured that the quasi-facets of a simple polytope are precisely the induced subgraphs of the facets. The reconstruction algorithm discussed in the paper under review would greatly benefit if this conjecture were true. And, in fact, it turned out to be true for all classes of polytopes studied in the aforementioned runtime experiments. However, recently \textit{C. Haase} and \textit{G. M. Ziegler} [arXiv: math.CO/0011170] found a counter-example to Perles' conjecture. For the entire collection see [Zbl 0944.00089].











This page was built for publication: Reconstructing a simple polytope from its graph

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4518986)