2 The Effects of Population Diversity

Genetic programming works by selecting individuals and recombining their genetic material to produce new offspring. It is common for genetic programming to use only the subtree crossover operator. When mutation operators are used, their effects are not unlike crossover and they are often used at low rates. The type of individuals in the population will obviously affect the type of individuals selected and the type of offspring produced. Recombining similar or dissimilar individuals will effect offspring size and fitness. Diverse and non-diverse populations can make the search more difficult depending on the landscape defined by the representation, operator and fitness function. The possibility of context and content conflicts in the representation can further complicate the effects of population diversity. Also, population diversity can effect selection pressure, making the search more random or more focused.