Uniformly distributed distances -- a geometric application of Janson's inequality (Q1125611): Difference between revisions

Consider a configuration \(P\) of \(n\) points in the plane, of minimal distance~\(1\), and sort the \({n\choose 2}=: m\) pairwise distances as \(1=d_1\leq d_2\leq \ldots\leq d_m=D\), where \(D\) is the diameter of the configuration. Based on a result of \textit{P. Erdős, E. Makai, J. Pach} and \textit{J. Spencer} [Am. Math. Soc., 265-273 (1991; Zbl 0741.52010)] it is rather easy to see that the quantity \(f(P):=\sum^{m-1}_{i=1}(d_{i+1}-d_i)^2\) satisfies \(f(P)=\Omega(n^{-6/7})\), that is, the distances cannot be distributed too evenly. In the paper under review, an intricate probabilistic construction is given for point sets \(P\) that indeed achieve this bound, with \(f(P)=\Theta(n^{-6/7})\). The analysis is based on a continuous version, for Poisson processes that pick points according to a continuous measure, of \textit{S. Janson's} inequality [Random Struct. Algorithms 1, 221-226 (1990; Zbl 0747.05079)].