ProbXiv
Explorersign in
Problem archiveProblem record

Statement

For X,Y,ZX, Y, Z independent vectors in Rd\mathbb{R}^d having densities, we define the standard multivariate entropy power as N(X)≜exp⁡(2dh(X))N(X) \triangleq \exp\left(\frac{2}{d} h(X)\right). The conjecture is that

N(X+Y+Z)N(Z)+N(X)N(Y)≤N(X+Z)N(Y+Z).N(X+Y+Z)N(Z) + N(X) N(Y) \leq N(X+Z)N(Y+Z).

The above is known to be true when ZZ is Gaussian, however we additionally desire a direct proof of this known result.

Note that without the N(X)N(Y)N(X)N(Y) term on the left, this is precisely the submodularity of entropy of sums that follows from standard information inequalities but was first explicitly observed in this paper: M. Madiman, “On the Entropy of Sums”, Proceedings of the 2008 IEEE Information Theory Workshop, Porto, Portugal, May 2008.

The conjecture may be thought of as a way to strengthen this submodularity.

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.