Reproduction

The pairing starts by computing of the fitness of all individuals in a deme (see selection step). Then, the cumulative fitness vector g of their fitnesses is created as

where is a fitness of the individual j in the deme and D represent a number of individuals per deme specified by parameter DEMEsize. Then the D pairs and of numbers in range are generated from a uniform distribution. The individuals k and l are parents of the next offspring i, if

where g{k−1} = 0 in case < g1 and gl−1 = 0 in case < g1 respectively. This procedure is analogous to generating random points in the squared space of all possibilities of parents pairs.