Optimal Matching and Empirical Measures

From MaRDI portal





Let \(P_ n\) be the n th empirical measure associated to the uniform measure P on the unit square. The exact order of convergence for the dual bounded Lipschitz distance \(\beta (P_ n,P)\) and the Prokhorov distance \(\rho (P_ n,P)\) is found: \[ \lim_{n\to \infty}\Pr (c\leq (n/\log n)^{1/2}\beta (P_ n,P)\leq C)=1, \] \[ \lim_{n\to \infty}\Pr (c\leq (n^{1/2}/\log^{3/4}n)\rho (P_ n,P)\leq C)=1, \] with some constants c and C. The results solve a long-open problem in empirical measures. They are obtained in a relatively simple fashion, using known results for optimal matching problems and their connections with Glivenko-Cantelli convergence of empirical measures.











This page was built for publication: Optimal Matching and Empirical Measures

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4206152)