Computer Software
Print ISSN : 0289-6540
A SAT Encoding of Pseudo-Boolean Constraints via Boolean Cardinality Constraints
Yushi MINAMITakehide SOHMutsunori BANBARANaoyuki TAMURA
Author information
JOURNAL FREE ACCESS

2018 Volume 35 Issue 3 Pages 3_65-3_78

Details
Abstract

In this paper, we propose a new encoding method for a set of PB (Pseudo-Boolean) constraints into propositional satisfiability (SAT) problems, in which Boolean Cardinality (BC) constraints are used as an intermediate form. The proposed method has the following three features. First, it can maintain general arc consistency by unit propagation of a SAT solver. Second, it can encode equivalent PB constraints with the same solutions—even their coefficients and right hand side value are different—into the same intermediate form and SAT instance. Third, for PB constraints whose number of kinds of coefficients is relatively small compared with the number of terms, the intermediate form becomes simpler and they can be encoded with a small number of clauses. Such PB constraints often appear in international PB solver competition benchmarks. In experiments, we compared the proposed encoding method with existing methods, BDD and Sorter. The former maintains general arc consistency by unit propagation, while the later maintains consistency checking that is weaker than general arc consistency. As the result, for PB constraints in which the number of different coefficients is not more than 10%, we confirmed that the proposed method is better than those two methods in terms of the number of encoded clauses and the efficiency in solver performance.

Content from these authors
© 2018 Japan Society for Software Science and Technology
Previous article Next article
feedback
Top