Sumset and inverse sumset theorems for Shannon entropy
Abstract
Let be an additive group. The sumset theory of Pl\"unnecke and Ruzsa gives several relations between the size of sumsets of finite sets , and related objects such as iterated sumsets and difference sets , while the inverse sumset theory of Freiman, Ruzsa, and others characterises those finite sets for which is small. In this paper we establish analogous results in which the finite set is replaced by a discrete random variable taking values in , and the cardinality is replaced by the Shannon entropy . In particular, we classify the random variable which have small doubling in the sense that when are independent copies of , by showing that they factorise as where is uniformly distributed on a coset progression of bounded rank, and . When is torsion-free, we also establish the sharp lower bound , where goes to zero as .
Cite
@article{arxiv.0906.4387,
title = {Sumset and inverse sumset theorems for Shannon entropy},
author = {Terence Tao},
journal= {arXiv preprint arXiv:0906.4387},
year = {2020}
}
Comments
32 pages, no figures. An error in the quantitative bounds in Lemma 6.1 (pointed out by Lampros Gavalakis and Ioannis Kontoyiannis) has been fixed (at the cost of making these bounds exponentially worse)