Controlling and investigating cellular automaton behavior via interactive inversion and visualization of the search space

  • Authors:

  • Fabio Boschetti

  • Affiliations:

  • CSIRO, Exploration and Mining, Bentley, WA, Australia

  • Venue:
  • New Generation Computing
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

An interactive genetic algorithm (IGA) provides a means to optimize the input parameters controlling the behavior of a cellular automaton (CA) The result is one or more combinations of parameters that allow the CA to reproduce geological patterns ol fluid flow and chemical reactions in fractured media.Via the IGA the user can provide subjective feedback on the quality of the CA results, which would otherwise be difficult to express numerically. A simple modification to the IGA ranking process combined with a self-organizing map enables the rapid on-line visualization of the high-dimensional parameter space, and consequent control over the inversion itself. The insights into the topology of the parameter space offer an understanding of which parameters control different CA behaviors.