A Generalized Parallel Genetic Algorithm in Erlang
Amanda Bienz, Kossi Fokle, Zachary Keller, Ed Zulkoski, Scott M. Thede · 2012
The focus of this paper is to implement a genetic algorithm using parallel programming. Genetic algorithms are well-suited to “parallelization,” since they model many individuals. Three implementations of a genetic algorithm were created for this paper - a standard sequential programming algorithm, a parallel algorithm using a master process to control the algorithm’s operations, and a parallel algorithm using a grid structure for the individuals. These implementations were tested on a single workstation as well as a server with many processors. The parallel algorithms outperformed the sequential algorithm, and their performance improved when run on the server with more processors.