Home // ADVCOMP 2015, The Ninth International Conference on Advanced Engineering Computing and Applications in Sciences // View article


An Automatic Code Generator for Parallel Evolutionary Algorithms: Achieving Speedup and Reducing the Programming Efforts

Authors:
Omar Andres Carmona Cortes
Jackson Amaral da Silva
Eveline de Jesus Viana Sa
Andrew Rau-Chaplin

Keywords: Parallel Computing; Evolutionary Computation; Programming Effort; Code Generator

Abstract:
Building parallel applications is not a trivial task, especially when these applications involve different kinds of evolutionary computation because two different fields have to be mastered. In order to overcome this problem, we propose an automatic code generator that automatically creates Java code for parallel evolutionary algorithms considering four models of parallelism: master-slave, island, cellular, and hierarchical. Furthermore, two evolutionary algorithms can be created: genetic algorithms and evolution strategies. A speedup experiment on a parallel genetic algorithm showed that good performance can be achieved using our generator. Moreover, we applied COCOMO and COCOMO II models in order to demonstrate that the cost for programming this kind of application can be considerably reduced in terms of effort, time and people when the generator is used. According to COCOMO II model, the generator may save about 6 months using 2 people.

Pages: 36 to 41

Copyright: Copyright (c) IARIA, 2015

Publication date: July 19, 2015

Published in: conference

ISSN: 2308-4499

ISBN: 978-1-61208-419-0

Location: Nice,France

Dates: from July 19, 2015 to July 24, 2015