A large-dimensional independent and identically distributed property for nearest neighbor counts in Poisson processes (Q1814751)

For an arbitrary point \(Q\) of a homogeneous Poisson point process \(\Pi\) with constant intensity rate \(\lambda_d\) in \(d\)-dimensional Euclidean space \(\mathbb{R}^d\), consider the number \(N_{d,r}\) \((r=1,2,\dots)\) of Poisson points that have that given point \(Q\) as their \(r\)th nearest neighbor (with respect to \(L_p\) distance, \(1\leq p<\infty\)). The main objective of this paper is the following limit theorem: As \(d\to\infty\), \(N_{d,1},N_{d,2},\dots\) are i.i.d. asymptotically Poisson random variables with mean 1: \((N_{d,1},N_{d,2},\dots)\Rightarrow(N_1,N_2,\dots)\) as \(d\to\infty\), where the convergence is as ``convergence in distribution'' and \(N_1,N_2,\dots\) are i.i.d. Poisson random variables with mean 1. The proof relies on Rényi's characterization of Poisson processes and a representation in the limit as \(d\to\infty\) of each nearest neighbor count \(N_{d,r}\) as a sum of countably many dependent Bernoulli random variables. The proof of the theorem follows as a consequence of the dependent structure of these Bernoulli random variables.