A Genetic Algorithm (GA) is a form of complex system in which various structures interact via sufficiently complicated operators and rules that it is difficult to characterize the behavior of the system exactly. However, competition at this level is in the control of upper-level structure(such as genetic operators and corresponding parameter setting, etc.). Thus, how to optimize this structure to make individuals(chromosomes) evolve better is essential to GA application. In this paper, we first present a model or formalism for a multi-level GA (denotednGA) which describes evolution occurring on several levels and scales simultaneously. Then we present an instance of an nGA (DAGA2) with a particular desired set of distributed and adaptive characteristics. One of the main features of DAGA2 is “on-line” competition among multiple GAs during problem solution. Performance of this system, realized using the PVM (Parallel Virtual Machine) tools, was determined for several well-known test functions. The results demonstrate that this model, in contrast to some earlier “GA within GA” approaches, shows promise not only as a theoretical framework, but also as a practical tool for GA search.
[1]
L. Darrell Whitley,et al.
GENITOR II: a distributed genetic algorithm
,
1990,
J. Exp. Theor. Artif. Intell..
[2]
L. Darrell Whitley,et al.
Serial and Parallel Genetic Algorithms as Function Optimizers
,
1993,
ICGA.
[3]
Erik D. Goodman,et al.
Coarse-grain parallel genetic algorithms: categorization and new approach
,
1994,
Proceedings of 1994 6th IEEE Symposium on Parallel and Distributed Processing.
[4]
Bernard Manderick,et al.
A Massively Parallel Genetic Algorithm: Implementation and First Analysis
,
1991,
ICGA.
[5]
Lalit M. Patnaik,et al.
Adaptive probabilities of crossover and mutation in genetic algorithms
,
1994,
IEEE Trans. Syst. Man Cybern..
[6]
Kalyanmoy Deb,et al.
Genetic Algorithms, Noise, and the Sizing of Populations
,
1992,
Complex Syst..