Improving the performance of evolutionary algorithms for the satisfiability problem by refining functions

Abstract

The performance of evolutionary algorithms (EAs) for the satisfiability problem (SAT) can be improved by an adaptive change of the traditional fitness landscape. We present two adaptive refining functions containing additional heuristic information about solution candidates: One of them is applicable to any constraint satisfaction problem with bit string representation, while the other is tailored to SAT. The influence of the refining functions is controlled by adaptive mechanisms. A comparison of the resulting EA with other approaches from literature indicates the suitability of our approach for SAT.