Clause Size Reduction with all-UIP Learning

0Citations
Citations of this article
4Readers
Mendeley users who have this article in their library.

This artice is free to access.

Abstract

Almost all CDCL SAT solvers use the 1-UIP clause learning scheme for learning new clauses from conflicts, and our current understanding of SAT solving provides good reasons for using that scheme. In particular, the 1-UIP scheme yields asserting clauses, and these asserting clauses have minimum LBD among all possible asserting clauses. As a result of these advantages, other clause learning schemes, like i-UIP and all-UIP, that were proposed in early work are not used in modern solvers. In this paper, we propose a new technique for exploiting the all-UIP clause learning scheme. Our technique is to employ all-UIP learning under the constraint that the learnt clause’s LBD does not increase (over the minimum established by the 1-UIP clause). Our method can learn clauses that are significantly smaller than the 1-UIP clause while preserving the minimum LBD. Unlike previous clause minimization methods, our technique is not limited to learning a sub-clause of the 1-UIP clause. We show empirically that our method can improve the performance of state of the art solvers.

Cite

CITATION STYLE

APA

Feng, N., & Bacchus, F. (2020). Clause Size Reduction with all-UIP Learning. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 12178 LNCS, pp. 28–45). Springer. https://doi.org/10.1007/978-3-030-51825-7_3

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free