Hybrid SAT Solver Considering Circuit Observability

Abstract

Boolean satisfiability is a NP-hard problem in computer theoretic science. There are two types of SAT solvers, random local solver and DPLL-based complete solver. Some people have proposed hybrid SAT solvers that combined the advantage of them both. They have successfully applied them to solve large or hard random SAT problems and circuit related problems. However, these solvers often find over-satisfying assignments when solving some circuit related problems. In this paper, circuit observability is considered in hybrid SAT solver, which helps reduce the overhead caused by over-satisfying and prune the searching space. The experimental results show that our hybrid SAT solver is more efficient than other solvers for large and hard circuits.