The Asymptotic Order of the k-SAT Threshold
Statistical Mechanics
2009-09-29 v1 Disordered Systems and Neural Networks
Abstract
Form a random k-SAT formula on n variables by selecting uniformly and independently m=rn clauses out of all 2^k (n choose k) possible k-clauses. The Satisfiability Threshold Conjecture asserts that for each k there exists a constant r_k such that, as n tends to infinity, the probability that the formula is satisfiable tends to 1 if r < r_k and to 0 if r > r_k. It has long been known that 2^k / k < r_k < 2^k. We prove that r_k > 2^{k-1} \ln 2 - d_k, where d_k \to (1+\ln 2)/2. Our proof also allows a blurry glimpse of the ``geometry'' of the set of satisfying truth assignments, and a nearly exact location of the threshold for Not-All-Equal (NAE) k-SAT.
Keywords
Cite
@article{arxiv.cond-mat/0209622,
title = {The Asymptotic Order of the k-SAT Threshold},
author = {Dimitris Achlioptas and Cristopher Moore},
journal= {arXiv preprint arXiv:cond-mat/0209622},
year = {2009}
}
Comments
Conference version to appear in FOCS (Foundations of Computer Science) 2002