Figure 5.

Coupled Genetic Algorithm Schematic. Top: Schematic of a standard genetic algorithm. An initial population (purple) of individuals or chromosomes (orange), consisting of individual bits or genes (black), reproduce with crossover of random sets of genes and mutations. Fitness according to a fixed fitness function is calculated for each chromosome, after which individuals with highest fitness are selected and cloned to form a new population. After multiple iterations of this process, the algorithm stops when no higher fitness can be generated in any new individual. Bottom: Schematic of a coupled genetic algorithm. Coevolving populations undergo reproduction as usual but are then coupled in the computation of each other’s fitness function, which in turn are variable and can incorporate integration of past fitness functions as a form of memory, at which point learning methods can be implemented.