OBSERVING ARTIFICIAL LIFE EVOLUTION: A FORMAL ALGEBRAIC FRAMEWORK

Abstract

Establishing the presence of evolutionary behavior is a major step in Artificial-life (ALife) studies. We present a general abstract algebraic formal framework for this aim. The framework is based upon the notion of high-level observations made upon the ALife model at hand. Starting with defining observed entities, the framework prescribes a series of definitions, followed by axioms that must be met in order to establish evolutionary processes in the model. The framework is sufficiently generic to be applicable to a wide variety of ALife studies and does not depend upon the low-level dynamics and structure of the underlying chemistry. The framework is illustrated on four different kind of ALife models including Cellular Automata based Langton Loops and Evoloops, Lambda calculus based Algorithmic Chemistry, and two new experimental models - the Reduced Instruction Set Artificial Chemistry and the Artificial Graph Chemistry. Generic design principles for ALife research are drawn based upon the framework design and the case study analysis.