Evolving an Harmonic Number Generator with ReNCoDe

Evolutionary Algorithms (EA) are loosely inspired in the ideas of natural selection and genetics. Over the years some researchers have advocated the need of incorporating more ideas from biology into EAs, in particular with respect to the individuals’ representation and the mapping from the genotype to the phenotype. One of the first successful proposals in that direction was the Artificial Regulatory Network (ARN) model. Soon after some variants of the ARN with increased capabilities were proposed, namely the Regulatory Network Computational Device (ReNCoDe). In this paper we further study ReNCoDe, testing the implications of some design choices of the underlying ARN model. A Genetic Programming community-approved symbolic regression benchmark (the harmonic number) is used to compare the performance of the different alternatives.