Phlegmatic Mappings for Function Optimization with Genetic Algorithms

The nature of the mapping between the genotype and the phenotype is vital to the performance of a genetic algorithm. We develop the notion of a phlegmatic mapping function, which encapsulates the idea that small changes in the genotype ought to produce small changes in the phenotype. After illustrating a mapping function with this property, we validate the idea using a test suite of function optimisation problems. We show that using a phlegmatic mapping can greatly improve the performance of a genetic algorithm for this problem domain.