Variance asymptotics and scaling limits for random polytopes (Q329472): Difference between revisions

Scaling limits for statistics of random polytopes is a long-studied problem in stochastic geometry with a rich history. A natural recipe to generate a random polytope is to throw points at random in a box, and then take their convex hull. Some statistics of interest include the number of \(k\)-faces \(f_k\) and the volume \(\mathrm{Vol}\) of this random polytope. The scaling limit problem is to establish precise convergence theorems for these statistics as the intensity of the points thrown go to infinity. A more formal description is as follows: let \(P_\lambda\) be a homogeneous Poisson point process on \(\mathbb R^d\) with rate \(\lambda\). Fix an observation window \(K \subset \mathbb R^d\), take the convex hull of the points of \(P_\lambda\) that fall in \(K\). Call this random polytope \(K_\lambda\). It is generally believed that \(f_k(K_\lambda)\) and \(\mathrm{Vol}(K_\lambda)\) satisfy central limit theorems as \(\lambda \to \infty\) for most \(K\). This is first established for \(K\) with smooth boundary by \textit{M. Reitzner} [Probab. Theory Relat. Fields 133, No. 4, 483--507 (2005; Zbl 1081.60008)], and for \(K\) a convex polytope by \textit{I. Bárány} and \textit{M. Reitzner} [Adv. Math. 225, No. 4, 1986--2001 (2010; Zbl 1204.52007); Ann. Probab. 38, No. 4, 1507--1531 (2010; Zbl 1204.60018)]. However, for the latter case, the authors could not obtain precise asymptotics. This paper gives the explicit asymptotics when \(K\) is a simple polytope. The proof technique builds on the scaling transform \(T^{(\lambda)}\), put forward by \textit{Yu. Baryshnikov} [Probab. Theory Relat. Fields 117, No. 2, 163--182 (2000; Zbl 0961.60017)].