ProbXiv
Explorersign in
Problem archiveProblem record

Statement

Let XX and YY be independent random variables with ultra-log-concave (ULC) probability mass functions on Z+\mathbf{Z}_+. Let TαXT_\alpha X denote thinning of XX with parameter α\alpha, let E(t)=H(Poisson⁡(t)){\cal E}(t)=H(\operatorname{Poisson}(t)), and define V(X)=E−1(H(X))V(X)={\cal E}^{-1}(H(X)). Thinned entropy power inequality. For 0<α<10<\alpha<1,

V(TαX+T1−αY)≥αV(X)+(1−α)V(Y).V(T_\alpha X+T_{1-\alpha}Y)\geq\alpha V(X)+(1-\alpha)V(Y).

This is proposed as a discrete analogue of the entropy power inequality, with Poisson distributions playing the role of normal distributions. The paper notes that the corresponding unthinned inequality is false, even for ULC distributions, and uses the proposed inequality to derive concavity of entropy under thinning.

Record

Comments

Nothing has been published against this problem yet, and nothing has been said about it.

Sign in with an institutional address to take part in the discussion. Reading every thread stays open to everyone.

Sign in

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.