Entropy power inequalities
May 1 to May 5, 2017
American Institute of Mathematics,
San Jose, California
and Mokshay Madiman
This workshop will be devoted to bringing
researchers from different communities (including probability, functional
analysis, information and estimation theory), who study and use forms of
Shannon's Entropy Power Inequality (EPI). This result and its many extensions
and generalizations have been proved using a number of different approaches,
including the de Bruijn identity, optimal transport and minimum mean square
estimation (MMSE). Although the EPI is interesting in its own right, it has
further significant impact by connecting several important and active research
areas, including functional inequalities on Riemannian manifolds, inequalities
in convex geometry, entropic versions of results in additive combinatorics and
bounding the capacity of communication channels.
The main topics for the workshop are to:
- Understand and help unify the existing approaches to the EPI, and related
results including reverse EPIs, forms of log-Sobolev inequalities and sumset
bounds in additive combinatorics.
- Develop new EPIs in infinitely divisible settings, to include stable laws and
Establish new connections arising from these methods.
Material from the workshop
A list of participants.
The workshop schedule.
A report on the workshop activities.
A list of open problems.