English

From Super Poincar\'e to Weighted Log-Sobolev and Entropy-Cost Inequalities

Probability 2007-12-20 v1 Differential Geometry

Abstract

We derive weighted log-Sobolev inequalities from a class of super Poincar\'e inequalities. As an application, the Talagrand inequality with larger distances are obtained. In particular, on a complete connected Riemannian manifold, we prove that the log\dd\log^\dd-Sobolev inequality with \dd(1,2)\dd\in (1,2) implies the L2/(2\dd)L^{2/(2-\dd)}-transportation cost inequality W2/(2\dd)\rr(fμ,μ)2/(2\dd)Cμ(flogf),μ(f)=1,f0W^\rr_{2/(2-\dd)}(f\mu,\mu)^{2/(2-\dd)}\le C\mu(f\log f), \mu(f)=1, f\ge 0 for some constant C>0C>0, and they are equivalent if the curvature of the corresponding generator is bounded below. Weighted log-Sobolev and entropy-cost inequalities are also derived for a large class of probability measures on Rd\R^d.

Keywords

Cite

@article{arxiv.0712.3142,
  title  = {From Super Poincar\'e to Weighted Log-Sobolev and Entropy-Cost Inequalities},
  author = {Feng-Yu Wang},
  journal= {arXiv preprint arXiv:0712.3142},
  year   = {2007}
}