Answer Set Solving with Lazy Nogood Generation

Abstract

Although Answer Set Programming (ASP) systems are highly optimised, their performance is sensitive to the size of the input and the inference it encodes. We address this deficiency by introducing a new extension to ASP solving. The idea is to integrate external propagators to represent parts of the encoding implicitly, rather than generating it a-priori. To match the state-of-the-art in conflict-driven solving, however, external propagators can make their inference explicit on demand. We demonstrate applicability in a novel Constraint Answer Set Programming system that can seamlessly integrate constraint propagation without sacrificing the advantages of conflict-driven techniques. Experiments provide evidence for computational impact. 1998 ACM Subject Classification I.2.3 Deduction and Theorem Proving