Parallelization of Spider Monkey Optimization (SMO) algorithm

Abstract

The parallelization is the adopted programming approach to reduce the execution time and increase the performance of programs in many elds. This approach makes the processors treat many processes/tasks in one moment of time, in contrast to the classical method, which is \the sequential approach" that treats a single problem at a moment of time. For this reason, we have implemented an application called ParSMO and it contains the im- plementation of a swarm intelligence based algorithm (SI), which is called: Spider Monkey Optimization (SMO) algorithm for numerical optimization. This thesis is composed of two main parts, the rst part is the implementation of the SMO algorithm sequentially, whereas the second part is the implementation of the rst proposed parallel SMO algorithm in the lit- erature, using the Multiprocessing package in Python 3.7. The aim of this work is comparing the sequential SMO with the parallel one, in terms of the execution time, the near-optimum solution and the objective space density of the last generation. The experimental study that was realized on ParSMO application using the two test problems \Dekkers and Aarts" (DA) and \Six-hump Camelback" (ShC) illustrates that the parallel SMO algorithm outperforms the sequential SMO in terms of gaining a shorter execution time and a better objective space density in the last generation.