A New Local Search Strategy for SAT

Recently, a new search strategy called configuration checking (CC) was proposed, for handling the cycling problem of local search. The CC strategy was used to improve the EWLS algorithm, a state-of-the-art local search for Minimum Vertex Cover (MVC). In this paper, we use this strategy to develop a local search algorithm for SAT called CWcc and a local search algorithm for weighted MAX-2-SAT called ANGScc. The CC strategy takes into account the circumstances of the variables when selecting a variable to flip. Experimental results show that the configuration checking strategy is more efficient than previous strategies for handling the cycling problem. We further improve CWcc; the resulting algorithm SWcc outperforms a state-of-the-art local search SAT solver TNM. ANGScc is also competitive with a state-of-the-art weighted MAX-2-SAT local search algorithm. Finally, we conduct some further analysis and experiments to compare the CC strategy with two other methods for handling the cycling problem: the tabu mechanism and the promising decreasing variable exploitation strategy.