Walksat Stalls Well Below Satisfiability

Improving WalkSAT for Random k-Satisfiability Problem with k > 3

Stochastic local search (SLS) algorithms are well known for their ability to efficiently find models of random instances of the Boolean satisfiablity (SAT) problem. One of the most famous SLS algorithms for SAT is WalkSAT, which is an initial algorithm that has wide influence among modern SLS algorithms. Recently, there has been increasing interest in WalkSAT, due to the discovery of its great ...

Threshold Behaviour of WalkSAT and Focused Metropolis Search on Random 3-Satisfiability

An important heuristic in local search algorithms for Satisfiability is focusing, i.e. restricting the selection of flipped variables to those appearing in presently unsatisfied clauses. We consider the behaviour on large randomly generated 3-SAT instances of two focused solution methods: the well-known WalkSAT algorithm, and the straightforward algorithm obtained by imposing the focusing const...

Improved Noise Tolerance for WalkSat

One noticeable trait of WalkSAT’s (Selman et al. 1994) best performing heuristics, NOVELTY and R_NOVELTY (McAllester et al. 1997), is the narrow band of acceptable noise levels that yield good success rates. We propose a modification to the NOVELTY heuristic, SR_NOVELTY, that adjusts the noise level to be proportional to the difference between the best available move and the second best availab...

Auto-Walksat: A Self-Tuning Implementation of Walksat

Parallel WalkSAT with Clause Learning

We present an extension of WalkSAT, a stochastic local search algorithm for solving SAT problems. Our extension learns new clauses by resolving existing clauses based on the current state of a WalkSAT run. We show that clause learning leads to significant speedup in WalkSAT runs, both in terms of fewer steps and faster runtime. We argue that our WalkSAT implementation is easily parallelizable, ...