Distributed Source Simulation With No Communication
Abstract
We consider the problem of distributed source simulation with no communication, in which Alice and Bob observe sequences and respectively, drawn from a joint distribution , and wish to locally generate sequences and respectively with a joint distribution that is close (in KL divergence) to . We provide a single-letter condition under which such a simulation is asymptotically possible with a vanishing KL divergence. Our condition is nontrivial only in the case where the G\`acs-K\"orner (GK) common information between and is nonzero, and we conjecture that only scalar Markov chains can be simulated otherwise. Motivated by this conjecture, we further examine the case where both and are doubly symmetric binary sources with parameters respectively. While it is trivial that in this case is both necessary and sufficient, we show that when is close to then any successful simulation is close to being scalar in the total variation sense.
Cite
@article{arxiv.1906.06970,
title = {Distributed Source Simulation With No Communication},
author = {Tomer Berg and Ofer Shayevitz and Young-Han Kim and Lele Wang},
journal= {arXiv preprint arXiv:1906.06970},
year = {2019}
}