About Selecting the Number of Processors for Parallel Multipopulation Genetic Algorithm

Introduction. The paper considers some features of the parallel implementation a multipopulation genetic algorithm, as well approaches to its optimization. results experiments with use different number processors and methods generating initial populations are presented in order optimize algorithm according several criteria (assessment computational time resources). On example specific test problem, estimates given for choosing optimal obtain desired result. purpose this work is conduct problem alternative population evaluate effectiveness algorithm. Results. For result 90–94 % optimum, most efficient terms computing resources 4 an uniform scanning space factor values. To achieve exceeding 94 by K1 (computational resources), 8 values showed best If we also take into account criterion K2, then 90–98 it necessary processors, 99–100 12 or 16 processes, depending on C1 C2 (cost respectively). Conclusions. Performed show that more than random method population. Experiments maximum efficiency PMGA, must be chosen precision. Keywords: generation, choice (populations),