The Analysis and Research of Parallel Genetic Algorithm

Abstract

With the application of the genetic algorithm (GA) deeply developed, the research of parallel genetic algorithm (PGA) and its realization become very important. Because of PGA inner parallel mechanism, its parallel process becomes a very naturally resolvable method. In this paper, four kinds of parallel models of parallel genetic algorithms, such as masterslave model, coarse-grained model, fine-grained model and mixed model, are simply generalized and evaluated. For every model, its characteristics are displayed. As for the existing problem to each model, the concerning parameters are illustrated in order to improve them. Then some main evaluation models of parallel genetic algorithms are presented. At the end, it is shown that parallel genetic algorithms should go on further study in the future.

The Adaptive Algorithm of Parallel Genetic Optimization Based on Orthogonal Wavelet of Space Diversity

“…Among them, coarse-grained parallel genetic algorithm distributes several sub populations to their corresponding processor, each processor is not only independent of calculation fitness, but also independent of selection, restructure crossover and mutation operation, and regularly send each other the best individual fitness, thus speeding up to meet the requirements of the termination conditions. It is a parallel genetic algorithm of the most widely used currently [7].…”

A review of Genetic Algorithms and Parallel Genetic Algorithms on Graphics Processing Unit (GPU)

“…It helps generate useful solutions to search problems. Various problem models in various applications were solved such as Travelling Salesman Problem (TSP), production scheduling, function optimization, machine language and so forth [3]. Parallel Genetic Algorithm (PGA) is referred as an algorithm that works by dividing large problem into smaller tasks.…”