An Effective Sharpsat Solution Method Based on Learnt Clause Minimization Approach

Abstract

Abstract The problem of satisfiability of the propositional logic formula (SAT problem) in computer science is an important and difficult problem. Learning clauses often contain redundant literal in CDCL SAT solving, which may have a negative impact on the performance of the solver. To overcome this shortcoming, we show a new sharpSAT solver based on learning clause minimization is proposed. By recoding the CNF formula to reduce the storage space, and applying Boolean constraint propagation to eliminate redundant literals in the learning clause, the algorithm reduces the time cost. The complexity is compared with the existing solver. The experimental results show that the solver has significant application value, reduces the time of instance solution, and increases the number of maximum solvable instances.