Generalizing DPLL to richer logics

Kenneth L. McMillan, Andreas Kuehlmann, Mooly Sagiv

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The DPLL approach to the Boolean satisfiability problem (SAT) is a combination of search for a satisfying assignment and logical deduction, in which each process guides the other. We show that this approach can be generalized to a richer class of theories. In particular, we present an alternative to lazy SMT solvers, in which DPLL is used only to find propositionally satisfying assignments, whose feasibility is checked by a separate theory solver. Here, DPLL is applied directly to the theory. We search in the space of theory structures (for example, numerical assignments) rather than propositional assignments. This makes it possible to use conflict in model search to guide deduction in the theory, much in the way that it guides propositional resolution in DPLL. Some experiments using linear rational arithmetic demonstrate the potential advantages of the approach.

Original languageEnglish
Title of host publicationComputer Aided Verification - 21st International Conference, CAV 2009, Proceedings
Pages462-476
Number of pages15
DOIs
StatePublished - 2009
Event21st International Conference on Computer Aided Verification, CAV 2009 - Grenoble, France
Duration: 26 Jun 20092 Jul 2009

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume5643 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference21st International Conference on Computer Aided Verification, CAV 2009
Country/TerritoryFrance
CityGrenoble
Period26/06/092/07/09

Fingerprint

Dive into the research topics of 'Generalizing DPLL to richer logics'. Together they form a unique fingerprint.

Cite this