InterJournal Complex Systems, 217
Status: Accepted
Manuscript Number: [217]
Submission Date: 981117
Revised On: 990817
Evolutionary Algorithms on a Self-Organized Dynamic Lattice
Author(s): Paul Halpern

Subject(s): CX.07

Category: Brief Article

Abstract:

We investigate a new optimization model, one which places an Evolutionary Algorithm with some, but not all, of the features of the Evolution Strategy approach of Rechenberg and Schwefel on a changing lattice of site connections -- namely a structurally dynamic cellular automaton. After seeding 1000-site lattices with real values, one per site, we observe these site values' behavior over time, subject to evolutionary survival of the fittest. In addition to the standard Darwinian optimization processes of selection, breeding, mutation and replacement, we add the feature of a self-selective neighborhood structure. Each generation, the link structure of the lattice is organized by the sites themselves, based upon fitness criteria. In our model, we have found, for various parameters, an overall tendency to greater average fitness. We have discovered, as well, for particular mutation rates, evidence of phase transitions in lattice link structure.

Retrieve Manuscript
Retrieve Previous Revision's Abstract
Submit referee report/comment


Public Comments: