Solving Radial Topology Constrained Problems with Evolutionary Algorithms

  • Authors:
  • P. M. S. Carvalho;L. A. F. M. Ferreira;L. M. F. Barruncho

  • Affiliations:
  • -;-;-

  • Venue:
  • SEAL'98 Selected papers from the Second Asia-Pacific Conference on Simulated Evolution and Learning on Simulated Evolution and Learning
  • Year:
  • 1998

Quantified Score

Hi-index 0.00

Visualization

Abstract

We report key algorithmic specific features involved in the evolutionary radial network problem solution. We focus on the dimensionality problem of large-scale networks and on the singularities of the radial topology search space. We (1) report the difficulties of the canonical genetic algorithm in handling network topology constraints, and (2) present both the genotype information structure and the recombination operator to overcome such difficulties. The proposed recombination operator processes genetic information as meaningful topological structures, and turns radiality and connectivity into genetic transmissible properties. Results are presented to illustrate the difference between the canonical approach and the approach taken.