Poisson polytopes (Q989185)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Poisson polytopes
scientific article

    Statements

    Poisson polytopes (English)
    0 references
    0 references
    0 references
    30 August 2010
    0 references
    Let \(K\subset {\mathbb R}^d\) be a convex set of volume 1. Assume that \(X= X(\eta)\) is a Poisson point process in \({\mathbb R}^d\) of intensity \(\eta\). The intersection of \(K\) with \(X(\eta)\) consists of uniformly distributed random points \(X_1,\dots,X_N\), where \(N\) is a random variable. Define the Poisson polytope \(\Pi_{\eta}\) as the convex hull \([X_1,\dots,X_N]= [K\cap X(\eta)]\). Due to the geometric nature of the available methods, investigations mainly concentrated on the expectation of functionals of random convex hulls such as volume or number of vertices [see \textit{W. Weil} and \textit{J. A. Wieacker}, in: Handbook of convex geometry. Volume B. Amsterdam: North-Holland. 1391--1438 (1993; Zbl 0788.52002)]. The first distributional results were only proven twenty years ago. \textit{P. Groeneboom} [Probab. Theory Relat. Fields 79, No.~3, 327--368 (1988; Zbl 0635.60012)] obtained the central limit theorem for the number of vertices of the Poisson polytope when the convex body \(K\) is the planar disc. In 1994, a central limit theorem for the area of a random polygon in the planar disc was proven by \textit{T. Hsing} [Ann. Appl. Probab. 4, No.~2, 478--493 (1994; Zbl 0806.60004)]. Recently, this result was generalized to arbitrary dimensions by \textit{M. Reitzner} [Probab. Theory Relat. Fields 133, No.~4, 483--507 (2005; Zbl 1081.60008)] where a central limit theorem has been established for \(V(\Pi_{\eta})\) (= the volume of the Poisson polytope) and for \(f_l(\Pi_{\eta})\) (= the number of \(l\)-dimensional faces of the Poisson polytope), when the body \(K\subset {\mathbb R}^d\) has smooth boundary. The situation seems to be much more involved when the underlying convex set is a polytope \(P\). In the planar case, when \(P\) is a convex polygon, a central limit theorem for the number of vertices \(f_o(\Pi_{\eta})\) was proven in [Zbl 0635.60012, loc. cit.], and for the area of \(\Pi_{\eta}\) by \textit{A. J. Cabo} and \textit{P. Groeneboom} [Probab. Theory Relat. Fields 100, No.~1, 31--55 (1994; Zbl 0808.60019)], but it seems that the stated variances are incorrect [see the discussion in: \textit{C. Buchta}, Discrete Comput. Geom. 33, No.~1, 125--142 (2005; Zbl 1065.52003)]. The main result of the present paper is a central limit theorem for the Poisson polytope \(\Pi_{\eta}\) for all dimensions \(d\geq 2\), when the body \(K\) is a polytope in \({\mathbb R}^d\). The Poisson polytope \(\Pi_{\eta}\) is closely related to the random polytope \(P_n\) defined in the following way: fix \(n\in {\mathbb N}\) and choose \(n\) random points \(X_1,\dots,X_n\) independently and uniformly from \(K\). The random polytope \(P_n\) is just the convex hull of these points. Clearly, \(P_n\) equals in distribution the Poisson polytope \(\Pi_{\eta}\), given that the (Poisson-distributed) number of points of \(X(\eta)\cap K\) is precisely \(n\). There have been many results concerning various properties of \(P_n\) as \(n\to\infty\) [see Zbl 0788.52002, loc. cit. for an extensive survey and also \textit{I. Bárány}, in: Stochastic geometry. Lectures given at the C.I.M.E. summer school held in Martina Franca, Italy, September 13--18, 2004. With additional contributions by D. Hug, V. Capasso, E. Villa. Berlin: Springer. Lecture Notes in Mathematics 1892, 77--118 (2007; Zbl 1123.60006) and \textit{I. Bárány}, Bull. Am. Math. Soc., New Ser. 45, No.~3, 339--365 (2008; Zbl 1144.52004) for more recent results].
    0 references
    random polytopes
    0 references
    central limit theorem
    0 references
    approximation of convex bodies
    0 references
    dependency graph.
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references