From almost Gaussian to Gaussian: Bounding differences of differential entropies
Résumé
We consider lower and upper bounds on the difference of differential entropies of a Gaussian random vector and an almost Gaussian random vector after both are ``smoothed'' by an arbitrarily distributed random vector of finite power. These bounds are important to prove the optimality of corner points in the capacity region of Gaussian interference channels. The upper bound, presented in MaxEnt-2014, follows from the data processing inequality (DPI). For the lower bound we consider a class of almost Gaussian distributions and use the DPI and a symmetry argument. We also show a counterexample that disproves a conjecture we proposed in MaxEnt-2014 regarding a certain difference of integrals.
Origine | Fichiers produits par l'(les) auteur(s) |
---|