The Mechanical Verification of a DPLL-Based Satisfiability Solver

Citation

Shankar, N., & Vaucher, M. (2011). The mechanical verification of a DPLL-based satisfiability solver. Electronic Notes in Theoretical Computer Science, 269, 3-17.

Abstract

Recent years have witnessed dramatic improvements in the capabilities of propositional satisfiability procedures or SAT solvers. The speedups are the result of numerous optimizations including conflict-directed backjumping. We use the Prototype Verification System (PVS) to verify a satisfiability procedure based on the Davis–Putnam–Logemann–Loveland (DPLL) scheme that features these optimizations. This exercise is a step toward the verification of an efficient implementation of the satisfiability procedure. Our verification of a SAT solver is part of a larger program of research to provide a secure foundation for inference using a verified reference kernel that contains a verified SAT solver. Our verification exploits predicate subtypes and dependent types in PVS to capture the specification and the key invariants.

Keywords: SAT solver, backlumping, predicate subtype, dependent type, PVS


Read more from SRI

  • Banner and attendees at the IEEE Hard Tech Venture Summit

    Cultivating hard tech startups that scale

    IEEE’s Hard Tech Venture Summit convened innovators at SRI to refine strategies and build new networks.

  • Patient going into a MRI

    Bringing surgical tools inside the MRI

    Drawing on SRI’s unique innovation ecosystem, the startup Medical Devices Corner is seeking to improve cancer surgery by advancing MRI-safe teleoperation.

  • Christopher Mims and Susan Patrick

    PARC Forum: How to AI

    The Wall Street Journal tech columnist Christopher Mims and SRI Education’s Susan Patrick discuss how AI can strengthen human agency.