An identity relating moments of functionals of convex hulls (Q1764174)

Denote by \(\mathbb K^d\) the set of all convex bodies (convex compact sets with non-empty interiors) in \(\mathbb R^d\). Fix \(K\in \mathbb K^d\) and choose \(n\) points independently and according to uniform distribution in \(K\). Denote by \(K_n\) the convex hull of the points. \(K_n\) is a random polytope. Denote by \(N_n\) the number of vertices of \(K_n\), and by \(V_n\) and \(D_n\) the volumes of \(K_n\) and \(K\setminus K_n\), respectively. Numerous papers were devoted to the question of how to determine the expected values of these random variables for certain classes of convex bodies. Particular attention was paid to the asymptotic behavior of the expected values as \(n\) tends to infinity. In contrast to the large number of results on expected values, not much is known so far about variances and higher moments. The available results concern throughout very special convex bodies \(K\) and either just \(n=d+1\) random points (in the case of \(n<d+1\) random points trivially \(V_n=0\), \(D_n= \operatorname{vol} K\) (the volume of \(K\)), \(N_n =n\) with probability one). In Sections 2 and 3 the author proves that for every \(K\in \mathbb K^d\), every \(n, k\in\mathbb N\), \(EV_n^k= (\operatorname{vol} K)^k\, E\prod_{i=1}^k(1-N_{n+k}/(n+i))\) and \(\operatorname{Var}(V_n) = \operatorname{Var}(D_n) = (\operatorname{Var}(N_{n+2}) + d_{n+2})(\operatorname{vol} K)^2/[(n+1) (n+2)]\), where \(E\) (Var) stands for expectation (variance) and \(d_{n+2}\) is entirely determined by the expected values \(EN_{n+1}\) and \(EN_{n+2}\). The main result of Section 4 is an explicit formula for \(p_l^{(n)} = \text{Pr}(N_n=l)\), \(l=1,2,\dots, n\), the distribution of \(N_n\) and it yields an answer to a question in [\textit{Yu. Baryshnikov}, Adv. Appl. Math. 17, 101--116 (1996; Zbl 0851.52005)]. Many references are contained in [\textit{C. Buchta} and \textit{M. Reitzner}, J. Reine Angew. Math. 536, 1--29 (2001; Zbl 0973.52004)].