Phase transitions in the regular random 3-SAT problem
In this paper we investigate phase transitions in the random 3-SAT problem but we move from the usual setting of classical logic to the more general setting of multiple-valued logics. We deal with regular CNF formulas and use a generalized Davis-Putnam (DP) procedure for testing their satisfiability. We establish the location of the threshold for different cardinalities of the truth value set and show experimentally that the location of the threshold increases logarithmically in the cardinality of the truth value set. We also provide a theoretical explanation of this fact. The DP procedure and the classical random 3-SAT problem appear to be a particular case of our approach.
KeywordsMultiple-valued logics regular CNF formulas satisfiability phase transitions threshold benchmarks
Unable to display preview. Download preview PDF.
- 1.P. Cheeseman, B. Kanefsky, and W. M. Taylor. Where the really hard problems are. In Proceedings of the International Joint Conference on Artificial Intelligence, IJCAI’91, pages 331–337, 1991.Google Scholar
- 3.R. Hähnle. Uniform notation of tableaux rules for multiple-valued logics. In Proc. International Symposium on Multiple-Valued Logic, Victoria, pages 238–245. IEEE Press, 1991.Google Scholar
- 6.R. Hähnle. Personal communication, 1998.Google Scholar
- 7.L. M. Kirousis, E. Kranakis, and D. Krizanc. Aproximating, the unsatisfiability threshold of random formulas. In Proc. 4th Annual European Symposium on Algorithms, ESA’96, pages 27–38, 1996.Google Scholar
- 9.D. Mitchell, B. Selman, and H. Levesque. Hard and easy distributions of SAT problems. In Proceedings of the 10th National Conference on Artificial Intelligence AAAI’92, San Jose, CA, USA, pages 459–465, 1992.Google Scholar