WASP: A native ASP solver based on constraint learning

Abstract

This paper introduces WASP, an ASP solver handling disjunctive logic programs under the stable model semantics. WASP implements techniques originally introduced for SAT solving that have been extended to cope with ASP programs. Among them are restarts, conflict-driven constraint learning and backjumping. Moreover, WASP combines these SAT-based techniques with optimization methods that have been specifically designed for ASP computation, such as source pointers enhancing unfounded-sets computation, forward and backward inference operators based on atom support, and techniques for stable model checking. Concerning the branching heuristics, WASP adopts the BerkMin criterion hybridized with look-ahead techniques. The paper also reports on the results of experiments, in which WASP has been run on the system track of the third ASP Competition. © 2013 Springer-Verlag.