This paper proposes new randomized fitness functions for a genetic algorithm used to solve the satisfiability problem. The fitness functions follow the general idea of probability amplification. The first function is inspired by the Lovász Local Lemma, while the second one is based on a randomized 2-SAT approximation. The genetic algorithm uses some specific components derived from unit propagation. The crossover operator and the restart strategy are designed to benefit from the application of unit propagation. A local search algorithm is applied on the best solution at each step of the algorithm in order to improve it. Competitive results were obtained for different benchmarks when compared with state-of-the-art algorithms.
This article is authored also by Synbrain data scientists and collaborators. READ THE FULL ARTICLE