Abstract. In this paper we settle two long-standing questions regarding the combinatorial complexity of Minkowski sums of polytopes: We give a tight upper bound for the number of faces of a Minkowski sum, including a characterization of the case of equality. We similarly give a (tight) upper bound theorem for mixed facets of Minkowski sums. This has a wide range of applications and generalizes the classical Upper Bound Theorems of McMullen and Stanley.Our main observation is that within (relative) Stanley-Reisner theory, it is possible to encode topological as well as combinatorial/geometric restrictions in an algebraic setup. We illustrate the technology by providing several simplicial isoperimetric and reverse isoperimetric inequalities in addition to our treatment of Minkowski sums.The Upper Bound Theorem (UBT) for polytopes is one of the cornerstones of discrete geometry. The UBT gives precise bounds on the 'combinatorial complexity' of a convex polytope P as measured by the number of k-dimensional faces f k (P ) in terms of its dimension and the number of vertices.
Upper Bound Theorem for polytopes. For a d-dimensional polytope P on n vertices and 0where Polytopes attaining the upper bound are called (simplicial) neighborly polytopes and are characterized by the fact that all non-faces are of dimension at least d 2 . Cyclic polytopes are a particularly interesting class of neighborly polytopes whose combinatorial structure allows for an elementary and explicit calculation of f k (Cyc d (n)) in terms of d and n; cf. [Zie95, Section 0]. The UBT was conjectured by Motzkin [Mot57] and proved by McMullen [McM70]. One of the salient features to note is that for given d and n there is a polytope that maximizes f k for all k simultaneously -a priori, this is not to be expected.In this paper we will address more general upper bound problems for polytopes and polytopal complexes. To state the main applications of the theory to be developed, recall that the Minkowski sum of polytopes and game theory. Our methods also apply to the study of mixed faces and we establish strong upper bounds and in particular characterize the case of equality in the most important case.
Upper Bound Theorem for mixed facets. The number of mixed facets of a Minkowski sum is maximized by Minkowski neighborly families.From discrete geometry to combinatorial topology to commutative algebra. An intriguing feature of the UBT is that its validity extends beyond the realm of convex polytopes and into combinatorial topology. Let ∆ be a triangulation of the (d − 1)-sphere and, as before, let us write f k (∆) for the number of k-dimensional faces. For example, boundaries of simplicial d-dimensional polytopes yield simplicial spheres, but these are by far not all. The UBTM too will be the consequence of a statement in the topological domain that we derive using algebra, though we will also briefly comment on a geometric approach to the problem. The appropriate combinatorial/topological setup for the UBPM is that of relative simplicial complexes: A re...