Bounding relative entropy by the relative entropy of local specifications in product spaces
Katalin Marton

TL;DR
This paper establishes an inequality linking the relative entropy of joint distributions to local specifications, leading to explicit bounds on logarithmic Sobolev constants for certain density functions.
Contribution
It introduces a novel inequality connecting global relative entropy to local conditional entropies, enabling explicit bounds on logarithmic Sobolev constants for product space densities.
Findings
Derived an inequality relating relative entropy to local specifications.
Provided explicit lower bounds for the logarithmic Sobolev constant.
Extended previous results by relaxing conditions on mixed partial derivatives.
Abstract
For a class of density functions on we prove an inequality between relative entropy and the sum of average conditional relative entropies of the following form: For any density function on , where and denote the local specifications for resp. , i.e., the conditional density functions of the 'th coordinate, given the other coordinates. The constant depends on the properties of the local specifications of . The above inequality implies a logarithmic Sobolev inequality for . We get an explicit lower bound for the logarithmic Sobolev constant of under the assumptions that: (i) the local…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGeometric Analysis and Curvature Flows · Markov Chains and Monte Carlo Methods · Nonlinear Partial Differential Equations
