High Locality Representations for Automated Programming

Abstract

We study the locality of the genotype-phenotype mapping used in grammatical evolution (GE). GE is a variant of genetic programming that can evolve complete programs in an arbitrary language using a variable-length binary string. In contrast to standard GP, which applies search operators directly to phenotypes, GE uses an additional mapping and applies search operators to binary genotypes. Therefore, there is a large semantic gap between genotypes (binary strings) and phenotypes (programs or expressions). The case study shows that the mapping used in GE has low locality leading to low performance of standard mutation operators. The study at hand is an example of how basic design principles of modern heuristics can be applied to explain performance differences between different GP approaches and demonstrates current challenges in the design of GE.