Integer decomposition property of dilated polytopes (Q470953)

Let \(P \subset {\mathbb R}^m\) be a \textit{lattice polytope}, i.e., the convex hull of finitely many points in \({\mathbb Z}^m\). We say that \(P\) possesses the \textit{integer decomposition property} if for any integer \(n > 0\) and any \(x \in nP \cap {\mathbb Z}^m\), there exist \(x_1, x_2, \dots, x_n \in P \cap {\mathbb Z}^m\) such that \(x = x_1 + x_2 + \cdots + x_n\). The literature calls such a polytope also \textit{integrally closed} and \textit{normal}, though these definitions are sometimes conflicting with other (closely related) definitions. It is well known that for any lattice polytope \(P\) there exists an integer \(k > 0\) such that \(kP\) has the integer decomposition property. Thus the smallest such \(k\) is an invariant of \(P\), which turns out to be an interesting object of study. There are various other similar invariants of a lattice polytope: {\parindent=0.7cm\begin{itemize}\item[{\(\bullet\)}] the smallest integer \(k>0\) such that \(jP\) has the integer decomposition property for all \(j \geq k\); \item[{\(\bullet\)}] the smallest integer \(k>0\) such that \(kP\) is very ample; \item[{\(\bullet\)}] the maximal degree among elements of the minimal Hilbert basis of the \textit{homogenization} of \(P\), i.e., the cone \(C(P) := {\mathbb R}_{ \geq 0 } (P \times \{ 1 \})\); \item[{\(\bullet\)}] the degree of \(\delta_P\); \item[{\(\bullet\)}] the maximal degree among elements in Box(\(P\)). \end{itemize}} Here we call a lattice polytope \(P\) \textit{very ample} if for every sufficiently large integer \(n > 0\) and any \(x \in nP \cap {\mathbb Z}^m\), there exist \(x_1, x_2, \dots, x_n \in P \cap {\mathbb Z}^m\) such that \(x = x_1 + x_2 + \cdots + x_n\); the \textit{degree} of an integer point in \(C(P)\) is its last coordinate; a \textit{Hilbert basis} of a polyhedral cone \(C \subseteq {\mathbb R}^{ m+1 }\) is a set of integer points in \(C\) that generates the semigroup \(C \cap {\mathbb Z}^{ m+1 }\); the polynomial \(\delta_P\) is the numerator of the \textit{Ehrhart series} of \(P\), defined through \[ 1 + \sum_{ n \geq 1 } \# \left( nP \cap {\mathbb Z}^m \right) z^n = {{ \delta_P(z) } \over { (1-z)^{ \dim(P)+1 } }} \, ; \] and Box(\(P\)) is a set of integer points in \(C(P)\) that are not in the semigroup generated by \((P \times \{ 1 \}) \cap {\mathbb Z}^{ m+1 }\) but that come, in a prescribed sense, from empty simplices (whose only integer points are vertices) contained in~\(P\). These six invariants of a lattice polytope have motivation and applications in various areas, e.g., commutative algebra, toric geometry, combinatorial number theory, and geometric combinatorics. The paper under review gives an exhaustive study of inequalities among these invariants, including examples for strict inequalities (between quantities that are ordered) and possible permutations (between quantities that have no relation). In paper form (i.e., aside from the book [\textit{W. Bruns} and \textit{J. Gubeladze}, Polytopes, rings, and K-theory. New York, NY: Springer (2009; Zbl 1168.13001)]) this is as good as an introduction to these topics get.