Coevolution

Sample. Choose the samples based on the information detained by the other processes   ®   coevolution.

Count. If n is the population size, we still have n sample whole individuals for each partial individual.

Choice. Use the best results from all the other processes since we aim to find the best whole individual, and the same completion for all the partial individuals in the population to minimize the communication.

Size. Experimentally, 1 or 2 samples are enough for a performance comparable to the sequential GA.

Communication. We need to update the samples from time to time   ®   every 50 - 100 generations is enough.

Biological examples. The World Cup (soccer), the Oplympic Games, virus against immune system, etc.