Performance comparison of three topologies of the island model of a parallel genetic algorithm implementation on a cluster platform

Parallel genetic algorithms (PGAs) have been used to improve the potential of genetic algorithms, which are efficient search techniques that have been employed in producing satisfactory solutions to optimization problems in which the application of standard techniques is not feasible or recommended. We implemented three communication topologies of the island model of a PGA, with the aim of analyzing the performance of the migration models applied. The topologies implemented were the star, the unidirectional ring and the bidirectional ring topologies. The implementation was developed on a cluster platform through the use of the standard Message Passing Interface (MPI) library. In order to test the performance of the algorithm with its variants, well-known benchmark functions were used.