An evolutionary local search algorithm for the satisfiability problem

Satisfiability problem is an NP-complete problem that finds itself or its variants in many combinatorial problems. There exist many complete algorithms that give successful results on hard problems, but they may be timeconsuming because of their branch and bound structures. In this manner, many successful incomplete algorithms are introduced. In this paper, the improvement of incomplete algorithms is of interest and it is shown that the incomplete algorithms can be more efficient if they are equipped with the problem specific knowledge, goal-oriented operators, and knowledge-based methods. In this aspect, an evolutionary local search algorithm is implemented, tested on a randomly generated benchmark that includes test instances with different sizes, and compared with prominent incomplete algorithms. Also, effects of goaloriented genetic operators and knowledge-based methods used in the evolutionary local search algorithm are examined by making comparisons with blind operators and random methods.