- Implement a two-population cooperative GA where the solutions in each population code for half of a 50-bit OneMax problem, and
the populations take it in turns to evolve for a generation.
If you have already implemented a single population binary-coded EA for this problem use the same parameters within each population. Otherwise use 100 in each population, binary tournament selection, 1-point crossover with probability 0.7 and bit-wise mutation with probability 0.01.
Investigate the effect of random versus best pairing strategies.
- Now repeat this experiment, but this time use a competitive
model. Let the fitness that one population gets be the OneMax
score and the fitness that the other gets be (50 – OneMax) i.e.,
ZeroMax. What happens to the two populations?