Binary clause reasoning in QBF

  • Authors:
  • Horst Samulowitz;Fahiem Bacchus

  • Affiliations:
  • Department of Computer Science, University of Of Toronto, Toronto, Ontario, Canada;Department of Computer Science, University of Of Toronto, Toronto, Ontario, Canada

  • Venue:
  • SAT'06 Proceedings of the 9th international conference on Theory and Applications of Satisfiability Testing
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Binary clause reasoning has found some successful applications in SAT, and it is natural to investigate its use in various extensions of SAT. In this paper we investigate the use of binary clause reasoning in the context of solving Quantified Boolean Formulas (QBF). We develop a DPLL based QBF solver that employs extended binary clause reasoning (hyper-binary resolution) to infer new binary clauses both before and during search. These binary clauses are used to discover additional forced literals, as well as to perform equality reduction. Both of these transformations simplify the theory by removing one of its variables. When applied during DPLL search this stronger inference can offer significant decreases in the size of the search tree, but it can also be costly to apply. We are able to show empirically that despite the extra costs, binary clause reasoning can improve our ability to solve QBF.