LandscapeEC: Comparing 2D to 3D Cellular Evolutionary Algorithms

摘要

Evolutionary Computation (EC) is a field of Computer Science that utilizes the basic principles of biological evolution to create a system that can be used to find an acceptable solutions to problems, which are called individuals. A specific kind of EC system, called a cellular Evolutionary Computation (cEA), models space and distance to limit interaction interaction within the system in an attempt to maintain diversity in the population and increase the chances of discovering an optimal solution. In this paper, we compare two kinds of cEA systems, one that models space in two dimensions, and one that models space in three dimensions. We compare the performance of 2D and 3D worlds on several 3SAT problems and the ONESMAX problem, and find that at least in our experiments 2D worlds tended to outperform 3D worlds and that the optimal parameter settings for 3D worlds were sometimes significantly different than those for 2D worlds.