Threshold for the expected measure of random polytopes

TL;DR

This paper investigates a threshold phenomenon for the expected measure of random polytopes formed by independent points from a log-concave measure, using the Cramer transform to identify conditions for a sharp transition.

Contribution

It establishes a sharp threshold for the expected measure of random polytopes based on the Cramer transform and introduces conditions involving the variance parameter for this transition.

Findings

Threshold occurs near rom N when \, \\ln N \\ll E_{\mu}(\Lambda_{\mu}^{*})

Expected measure approaches 0 or 1 depending on the relation between \\ln N and the Cramer transform expectation

Small variance parameter ollows the sharp threshold behavior for the measure of random polytopes.

Abstract

Let $\mu$ be a log-concave probability measure on ${\mathbb R}^n$ and for any $N>n$ consider the random polytope $K_N={\rm conv}\{X_1,\ldots ,X_N\}$, where $X_1,X_2,\ldots $ are independent random points in ${\mathbb R}^n$ distributed according to $\mu $. We study the question if there exists a threshold for the expected measure of $K_N$. Our approach is based on the Cramer transform $\Lambda_{\mu}^{\ast }$ of $\mu $. We examine the existence of moments of all orders for $\Lambda_{\mu}^{\ast }$ and establish, under some conditions, a sharp threshold for the expectation ${\mathbb E}_{\mu^N}[\mu (K_N)]$ of the measure of $K_N$: it is close to $0$ if $\ln N\ll {\mathbb E}_{\mu }(\Lambda_{\mu}^{\ast })$ and close to $1$ if $\ln N\gg {\mathbb E}_{\mu }(\Lambda_{\mu}^{\ast })$. The main condition is that the parameter $\beta(\mu)={\rm Var}_{\mu }(\Lambda_{\mu}^{\ast })/({\mathbb E}_{\mu }(\Lambda_{\mu }^{\ast }))^2$ should be small.