A Thread Building Blocks Based Parallel Genetic Algorithm

Genetic Algorithms are biologically-inspired computational methods commonly used for many different optimization problems in various domains. They are also known to require significant computational time to produce optimal solutions. This paper presents a genetic algorithm library that encompasses the main genetic operations such as selection, recombination and mutation. The library can be run in serial or parallel form using Intel's Thread Building Blocks which is a cross-platform multi-threading library. An evaluation with different benchmark functions has shown that the parallel implementation can achieve up to 3:1 speed-up over the serial implementation on a dual core computer with ordinary specifications.