Abstract

AbstractThe Conflict Driven Clause Learning (CDCL) Boolean Satisfiability (SAT) solvers are very effective in solving large and numerous crafted and industrial instances. Paradoxically, we do not know much about the reasons for their effectiveness and their running is hard to trace. This paper participates in the quest to understand the CDCL solvers. Specifically, we empirically study the behavior of one of their essential components which is the conflict analysis module. We show that this module returns generally a relevant backjump level whatever the analyzed clause. We also classify the falsified clauses according to their capacity to produce pertinent learned clauses. We use this classification to induce the apparition of specific clauses in the implication graph by ordering the list of clauses watched by the propagated literals. Finally, we advance some explanations on the effectiveness of CDCL solvers.KeywordsCDCL SolversConflict AnalysisEmpirical Study

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call