AbstractsBiology & Animal Science

On the evolution of genotype-phenotype mapping: exploring viability in the Avida articial life system

by Tomonori Hasegawa




Institution: Dublin City University
Department: School of Electronic Engineering; Dublin City University. Research Institute for Networks and Communications Engineering (RINCE)
Year: 2015
Keywords: Bioinformatics; Mathematical models; Artificial life; Computer simulation; Avida; Machine self reproduction
Record ID: 1180830
Full text PDF: http://doras.dcu.ie/20431/


Abstract

The seminal architecture of machine self-reproduction originally formulated by John von Neumann underpins the mechanism of self-reproduction equipped with genotype and phenotype. In this thesis, initially, a hand-designed prototype von Neumann style selfreproducer as an ancestor is described within the context of the artificial life system Avida. The behaviour of the prototype self-reproducer is studied in search of evolvable genotype-phenotype mapping that may potentially give rise to evolvable complexity. A finding of immediate degeneration of the prototype into a self-copying mode of reproduction requires further systematic analysis of mutational pathways. Through demarcating a feasible and plausible characterisation and classification of strains, the notion of viability is revisited, which ends up being defined as quantitative potential for exponential population growth. Based on this, a framework of analysis of mutants' evolutionary potential is proposed, and, subsequently, the implementation of an enhanced version of the standard Avida analysis tool for viability analysis as well as the application of it to the prototype self-reproducer strain are demonstrated. Initial results from a one-step single-point-mutation space of the prototype, and further, from a multi-step mutation space, are presented. In the particular case of the analysis of the prototype, the majority of mutants unsurprisingly turn out to be simply infertile, without viability; whereas mutants that prove to be viable are a minority. Nevertheless, by and large, it is pointed out that distinguishing reproduction modes algorithmically is still an open question, much less finer-grained distinction of von Neumann style self-reproducers. Including this issue, speciifc limitations of the enhanced analysis are discussed for future investigation in this direction.