Reformulating propositional satisfiability as constraint satisfaction

Abstract

We study how propositional satisfiability (SAT)p roblems can be reformulated as constraint satisfaction problems (CSPs).We analyse four different mappings of SAT problems into CSPs. For each mapping, we compare theoretically the performance of systematic algorithms like FC and MAC applied to the encoding against the Davis-Putnam procedure applied to the original SAT problem. We also compare local search methods like GSAT and WalkSAT on a SAT problem against the Min-Conflicts procedure applied to its encoding. Finally, we look at the special case of local search methods applied to 2-SAT problems and encodings of 2-SAT problems. Our results provide insight into the relationship between propositional satisfiability and constraint satisfaction, as well as some of the potential benefits of reformulating problems as constraint satisfaction problems.