English

Harnessing the Bethe free energy

Probability 2017-11-17 v3 Discrete Mathematics

Abstract

A wide class of problems in combinatorics, computer science and physics can be described along the following lines. There are a large number of variables ranging over a finite domain that interact through constraints that each bind a few variables and either encourage or discourage certain value combinations. Examples include the kk-SAT problem or the Ising model. Such models naturally induce a Gibbs measure on the set of assignments, which is characterised by its partition function. The present paper deals with the partition function of problems where the interactions between variables and constraints are induced by a sparse random (hyper)graph. According to physics predictions, a generic recipe called the "replica symmetric cavity method" yields the correct value of the partition function if the underlying model enjoys certain properties [Krzkala et al., PNAS 2007]. Guided by this conjecture, we prove general sufficient conditions for the success of the cavity method. The proofs are based on a "regularity lemma" for probability measures on sets of the form Ωn\Omega^n for a finite Ω\Omega and a large nn that may be of independent interest.

Keywords

Cite

@article{arxiv.1504.03975,
  title  = {Harnessing the Bethe free energy},
  author = {Victor Bapst and Amin Coja-Oghlan},
  journal= {arXiv preprint arXiv:1504.03975},
  year   = {2017}
}

Comments

This version replaces version 1 and the RANDOM 2015 version of the paper, which contained critical errors affecting the main results

R2 v1 2026-06-22T09:16:39.751Z