Sharp affine isoperimetric inequalities for the volume decomposition functionals of polytopes (Q2048617)

The authors introduce a new functional for polytopes, the so-called volume decomposition functional, as follows: given a polytope \(P\subset{\mathbb R}^n\) having the origin in its interior, if \(u_1,\dots,u_N\) are the unit outer normal vectors to its facets, then the \(k\)th volume decomposition functional \(X_k(P)\), \(k=1,\dots,n\) is defined as \[ X_k(P)^n=\sum_{\dim(\mathrm{span}\{u_{i_1},\dots,u_{i_n}\})=k} V_P(\{u_{i_1}\})\cdots V_P(\{u_{i_n}\}). \] Here, if \(F\) is a facet of \(P\) with outer normal \(u\), \(V_P(\{u\})=V_n(\mathrm{conv}\{0,F\})\) represents the cone volume measure of \(P\) with respect to \(u\). When \(k=n\), \(X_n(P)\) coincides with the centro-affine functional \(U\) defined by \textit{E. Lutwak} et al. [Trans. Am. Math. Soc. 353, No. 5, 1767--1779 (2001; Zbl 0971.52011)]. The authors investigate basic properties of these new functionals, such as the nice identity \(V_n(P)^n=X_1(P)^n+\dots+X_n(P)^n\). Then, as a natural extension of some issues addressed in [\textit{E. Lutwak} et al., Trans. Am. Math. Soc. 353, No. 5, 1767--1779 (2001; Zbl 0971.52011)], they pose the appealing question whether, given a polytope \(P\subset{\mathbb R}^n\) having its centroid at the origin, there exists a constant \(c(n,k)\), \(k=1,\dots,n-1\), such that \[ \dfrac{X_k(P)}{V_n(P)}\leq c(n,k). \] This problem is completely solved in dimension 3, namely, they prove that \(c(3,1)=(1/3)^{2/3}\), \(c(3,2)=(2/3)^{1/3}\) and \(c(3,3)=(2/9)^{1/3}\), and that each of these bounds is attained if and only if \(P\) is a parallelepiped. In arbitrary dimension, the cases \(k=1\) and \(k=2\) (here when \(P\) is a polytope whose any three outer normal vectors -- up to their antipodal normal vectors -- are linearly independent) are also solved. The authors conclude the paper extending, in a natural way, the functionals \(X_k\), \(k=1,\dots,n-1\), to general convex bodies \(K\subset{\mathbb R}^n\), and posing then the analogous interesting question of determining the best upper bound \(c(n,k)\) for the ratio \(X_k(K)/V_n(K)\).