An analysis of synchronous and asynchronous parallel distributed genetic algorithms with structured and panmictic Islands

Abstract

In a parallel genetic algorithm (PGA) several communicating nodal GAs evolve in parallel to solve the same problem. PGAs have been traditionally used to extend the power of serial GAs since they often can be tailored to provide a larger efficiency on complex search tasks. This has led to a considerable number of different models and implementations that preclude direct comparisons and knowledge exchange. To fill this gap we begin by providing a common framework for studying PGAs. This allows us to analyze the importance of the synchronism in the migration step of parallel distributed GAs. We will show how this implementation issue affects the evaluation effort as well as the search time and the speedup. In addition, we consider popular evolution schemes of panmictic (steady-state) and structured-population (cellular) GAs for the islands. The evaluated PGAs demonstrate linear and even super-linear speedup when run in a cluster of workstations. They also show important numerical benefits when compared with their sequential counterparts. In addition, we always report lower search times for the asynchronous versions.