Unimodular covers of multiples of polytopes (Q1856377): Difference between revisions

Let \(P\subset \mathbb R^n\) be an \(n\)-dimensional lattice polytope, i.e., a polytope whose vertices belong to the standard lattice. \textit{G. Kempf}, \textit{F. Knudsen}, \textit{D. Mumford} and \textit{B. Saint-Donat} [Toroidal embeddings. I. (1973; Zbl 0271.14017)] have shown that there exists a constant \(c_P\) (depending on \(P\)) such that \(c_P\,P\) admits a regular triangulation into unimodular simplices. However, it is still an open (and challenging) problem to decide whether there exists a constant \(c_{ut}\) such that for all \(c \geq c_{ut}\) the polytope \(c\,P\) admits such a triangulation and whether there exists a uniform bound (independent of \(P\)) for such a \(c_{ut}\). In the paper under review the authors study a closely related problem, namely they look for unimodular coverings instead of triangulations and obtain the following beautiful and remarkable result. There exists a constant \(c_{uc}(n)\) only depending on the dimension such that for all \(c\geq c_{uc}(n)\) the polytope \(c\,P\) can be covered by unimodular simplices. The constructive proof gives an upper bound on \(c_{uc}(n)\) of size \(O(n^5\,(3/2)^{n^{3/2}})\). A priori it was not clear at all that there really exists such a constant and it was only known that \(c_{uc}(2)=1\) and \(c_{uc}(3)=2\). They prove their result by passage to rational cones, for which they establish a similar bound. It is also worth mentioning that only one argument in their proof leads to the exponential term in the bound. Namely, they give an exponential upper bound on the height of vectors which are needed for a unimodular triangulation of a simplical cone.