Computing stable models in parallel

Answer-set programming (ASP) solvers must handle difficult computational problems that are NP-hard. These solvers are in the worst case exponential and their scope of applicability, despite recent impressive gains in performance, remains limited. One way to deal with limitations of answer-set programming is to exploit parallelism. In this paper, we design and implement a parallel algorithm, parstab, that computes stable models of logic programs. We describe preliminary experimental studies of parstab, running it on seven machines and comparing its performance to a serial execution. Our results are encouraging. For some problems, significant speedups are obtained by running parstab on multiple machines.

• Type: Conference
• Year: 2001
• Where: ASP
• Authors: Raphael A. Finkel, Victor W. Marek, Neil Moore, Miroslaw Truszczynski
• Keywords: Answer-Set Programming, ASP 2001, Automated Reasoning, Difficult Computational Problems, Worst Case