Equifacetted 4-polytopes (Q2519786)

The author introduces the objects of this paper in a strange way: ``The bicyclic polytope \(P_{n,p,q}\) is defined as the \(4\)-polytope which is the convex hull of the \(n\) points \[ \bigl(\cos(2\pi kp/n),\sin(2\pi kp/n),\cos(2\pi kq/n),\sin(2\pi kq/n)\bigr) \] where \(1 \leq p \leq n-1\) and \(1 \leq q \leq n-1\) are fixed integers, and \(k = 1,\ldots,n\). There is no loss of generality in assuming \(p = 1\) and \(\text{gcd}(n,q) = 1\) or \(2\).'' The second sentence is actually a further assumption by the author (clearly not following from the first), to restrict attention to the simplicial case. In this paper, the author makes a conjecture about the combinatorial structure of the facets of the polar of \(P_{n,p,q}\); however, he adduces little evidence to support the conjecture (which is a little complicated to reproduce here), other than a claim that it has been verified for all \(n \leq 50\). Reference to the paper by \textit{Z. Smilansky} [Isr. J.\ Math.\ 70, No.1, 82--92 (1990; Zbl 0713.52006)], which describes the facets and ridges of \(P_{n,p,q}\) in the general case, does lend plausibility to the conjecture. With a slight change of viewpoint from his, associate the vertices with the corresponding points on the grid \(\{0,1,\ldots,n\}^2\) (identified by parallel edges), with the initial vertex at \((0,0)\) (\(= (n,0)\) and so on). Those other points \((kp,kq) \pmod n\) in their convex hull which can be seen from \((0,0)\) give the edges; this picture also illustrates the part played by the quotients that arise in the euclidean algorithm for calculating \(\text{gcd}(n,q)\) when \(p = 1\). (The author's combinatorial description implies a curious symmetry property of these quotients; this is only proved in a special case, though the general case is quite straightforward.)