On the satisfiability and maximum satisfiability of random 3-CNF formulas
SODA '93 Proceedings of the fourth annual ACM-SIAM Symposium on Discrete algorithms
Annals of Mathematics and Artificial Intelligence
Lemma and cut strategies for propositional model elimination
Annals of Mathematics and Artificial Intelligence
Evolutionary algorithms for the satisfiability problem
Evolutionary Computation
Random 3-SAT: The Plot Thickens
CP '02 Proceedings of the 6th International Conference on Principles and Practice of Constraint Programming
Random 3-SAT: The Plot Thickens
Constraints
Persistent and Quasi-Persistent Lemmas in Propositional Model Elimination
Annals of Mathematics and Artificial Intelligence
An empirical analysis of search in GSAT
Journal of Artificial Intelligence Research
Backbone fragility and the local search cost peak
Journal of Artificial Intelligence Research
Towards an understanding of hill-climbing procedures for SAT
AAAI'93 Proceedings of the eleventh national conference on Artificial intelligence
Hi-index | 0.00 |
This paper presents empirical evidence of a satisfiability threshold in random 3CNF formulas. The paper also expands on and supports the conjecture of Mitchell, Selman, and Levesque that *hard* randomly generated CNF formulas will be hard for any reasonable satisfiability algorithm. We report statistics for a much larger set of variables than have been previously reported; in particular, we show that for each clause to variable ratio less than 4.2, the percentage of satisfiable formulas increases, and for each clause to variable ratio greater than 4.2, the percentage of satisfiable formulas decreases as a function of number of variables. We found that several algorithms behaved qualitatively in the same fashion. We report on the relative performance of each algorithm.