Algebraic degrees of 3-dimensional polytopes (Q2155626)

The authors study realizations of 3-polytopes with edges tangent to the unit sphere. The realization space of a polytope \(P\) is the set of all polytopes which are combinatorially equivalent to \(P\). It is known that in terms of the coordinates of the vertices, the facets or both, the realization space is a semialgebraic set. The authors consider Koebe realizations of a 3-polytope (which are realizations with edges tangent to the sphere) and Springborn realizations (which are Koebe realizations whose edge barycenter is the origin). It is known that Koebe and Springborn realizations admit descriptions as semialgebraic sets defined by rational polynomials. The purpose of this article is to study the minimal degrees of those polynomials. The authors call them Koebe and Springborn degrees, respectively. For greater clarity, we quote one of the main definitions used by the authors: `Given any realization \(R\) of \(P\), we denote with \(\mathbb Q [R]\) the field extension of \(\mathbb Q\) given by the vertex coordinates of \(R\). The Koebe degree \(\kappa (P)\) is the minimal algebraic degree \(| \mathbb Q [Q] : \mathbb Q \)| where \(Q\) varies over the Koebe realizations of \(P\).' The main results of the article under review are given in Theorems 10, 17, 20, and 23. Theorem 10 gives an upper bound of the Koebe degree, which is doubly exponential in the number of vertices or facets. It reads as follows: Let \(P\) be a 3-polytope with \(n\) vertices and a triangular facet. Then its Koebe degree is at most \[ \kappa(P) \leq 8^{2^{3n-9}-1}\cdot 2^{-(3n-9)} < 2^{2^3n}. \] The proof of Theorem 10 uses cylindrical algebraic decomposition. Theorem 17 demonstrates that the Springborn degree is a nontrivial invariant of a 3-polytope. It reads as follows: Let \(k \geq 4\) be an integer, \(B_k\) be the bipyramid over a \(k\)-gon, and \(\phi\) denote Euler's totient function (i.e., \(\phi (n)\) counts the positive integers up to \(n\) that are relatively prime to \(n\)). If \(k\) is odd, the Springborn degree \(\sigma (B_k)\) is either \(\phi(k)/2\) or \(\phi(k)\). If \(k\) is even, the Springborn degree \(\sigma (B_k)\) equals \(\phi(k)\). In particular, the Springborn degree of a 3-polytope is not bounded by any constant and is a nontrivial invariant of a 3-polytope. Theorem 20 is the Koebe degree analog to Theorem 17. It reads as follows: Let \(k \geq 4\) be an integer. Then the Koebe degree \(\kappa (B_k)\) is at least \(\phi (k)/4\). In particular, the Koebe degree of a 3-polytope is not bounded by any constant. Recall that a stacked \(d\)-polytope is a simplicial polytope obtained by starting with a \(d\)-simplex and successively adding vertices beyond a facet. Theorem 23 reads that the Koebe degree of any stacked 3-polytope equals one. The last section of the article contains a few open questions. As an example, we quote Question 25: `Given a polytope \(P\), what is the precise relationship between its Koebe degree \(\kappa (P)\) and its Springborn degree \(\sigma (P)\)? Does \(\kappa (P)\) divide \(\sigma (P)\)? Or are they even equal?'