high performance computing on graphics processing units: hgpu.org

This work is focussed on the development and analysis of a new class of algorithms, called cellular memetic algorithms (cMAs), which will be evaluated here on the satisfiability problem (SAT). For describing a cMA, we study the effects of adding specific knowledge of the problem to the fitness function, the crossover and mutation operators, and to the local search step in a canonical cellular genetic algorithm (cGA). Hence, the proposed cMAs are the result of including these hybridization techniques in different structural ways into a canonical cGA. We conclude that the performance of the cGA is largely improved by these hybrid extensions. The accuracy and efficiency of the resulting cMAs are even better than those of the best existing heuristics for SAT in many cases.