Evolution of 2D Apoptotic Cellular Automata
Submitted to CEC2015

Jennifer Garner and Daniel Ashlock

Abstract PDF eprint

An apoptotic cellular automata consists of an initial state and an updating rule. These specify an automata that grows for a time and then enters a quiescent state. This study generalizes earlier work on evolving 1D apoptotic automata to evolving 2D automata, producing a type of evolved art. Parameter studies are performed and it is found that the most important factors are algorithm runtime and the symmetry of the initial conditions of the automata. Other parameters such as mutation rate and tournament size are found to be relatively soft, as long as they do not take on extreme values. A collection of examples of renderings of evolved cellular automata are provided and steps for additional work to improve the system are outlined. Examination of automata with asymmetric starting conditions shows that the highest fitness individuals are those that follow a growth pattern that restores symmetry. This strongly suggests that optimizing the size of an apoptotic automata that has a symmetric pattern of states is a substantially easier problem.