OpenMP Genetic Algorithm for Continuous Nonlinear Large-Scale Optimization Problems

Abstract

Genetic algorithms (GAs) are one of the evolutionary algorithms for solving continuous nonlinear large-scale optimization problems. In an optimization problem, when dimension size increases, the size of search space increases exponentially. It is quite difficult to explore and exploit such huge search space. GA is highly parallelizable optimization algorithm; still there is a challenge to use all the cores of multicore (viz. Dual core, Quad core, and Octa cores) systems. The paper analyzes the parallel implementation of SGA (Simple GA) called as OpenMP GA. OpenMP (Open Multi-Processing) GA attempts to explore and exploit the search space on the multiple cores' system. The performance of OpenMP GA is compared with SGA with respect to time required and cores utilized for obtaining optimal solution. The results show that the performance of the OpenMP GA is remarkably superior to that of the SGA in terms of execution time and CPU utilization. In case of OpenMP GA, CPU utilization is almost double for continuous nonlinear large-scale test problems for the given system configuration.