Intrinsic volumes and Gaussian polytopes: the missing piece of the jigsaw (Q683762): Difference between revisions

Denote by \(\gamma_d\) the standard Gaussian measure on \({\mathbb R}^d\) with density \(\varphi_d=(2\pi)^{-d/2}\exp\left(-\|x\|^2/2\right)\), \(x\in {\mathbb R}^d\). Given \(n\geq d+1\) let \(X_1\),\dots, \(X_n\) be independent random points that are distributed on \({\mathbb R}^d\) according to the probability measure \(\gamma_d\). The random convex hull \(K_n=[X_1,\dots, X_n]\) of these points is a Gaussian polytope. For further background information and references see the survey paper [\textit{M. Reitzner}, in: New perspectives in stochastic geometry. Oxford: Oxford University Press. 45--76 (2010; Zbl 1202.60025)]. For \(l\in\{0,\dots, d\}\) denote by \(V_l(K_n)\) the \(l\)-th intrinsic volume of \(K_n\). Using the classical Efron-Stein jackknife inequality, \textit{D. Hug} and \textit{M. Reitzner} [Adv. Appl. Probab. 37, No. 2, 297--320 (2005; Zbl 1089.52003)] obtained a first upper bound of the form \(\operatorname{Var}[V_l(K_n)]\leq c_d (\log n)^{(l-3)/2}\) with a constant \(c_d\in (0, +\infty)\) only depending on space dimension \(d\) (but not on \(l\)). In a remarkable paper [\textit{P. Calka} and \textit{J. E. Yukich}, Probab. Theory Relat. Fields 163, No. 1--2, 259--301 (2015; Zbl 1360.60030)], the precise variance asymptotic was derived, showing thereby that the upper bound from [Hug and Reitzner, loc. cit.] does not have the right order of magnitude. In fact, Theorem 1.5 from [loc. cit.] says that \(\lim_{n\to \infty} (\log n)^{{{d+3}\over{2}}-l} \operatorname{Var}[V_l(K_n)] =c_{d,l}\) with constant \(c_{d,l}\in [0, +\infty)\) only depending on \(d\) and \(l\). However, using the methods of [Calka and Yukich, loc. cit.] one is not able to exclude the possibility that \(c_{d, l}=0\). The aim of the present paper is to fill this gap and to show that, in fact, \(c_{d, l}>0\). The rest of this paper is structured as follows. In Section 2, the authors recall the essential steps of a geometric construction from [the first author and \textit{V. Vu}, Ann. Probab. 35, No. 4, 1593--1621 (2007; Zbl 1124.60014)] and prove some auxiliary results that are needed in the proof of the main result of the paper. The latter is the content of the final Section 3.