I am using MultiPopCoevolutionaryEvaluator for implementing cooperative coevolution.
I need to evaluate an individual from one subpopulation by collaborating with the current best individual from other subpopulations (i.e., the best collaborators in the previous generation). So, I need to write codes in
"protected Individual producePrevious(int subpopulation, EvolutionState state, int thread)" of the MultiPopCoevolutionaryEvalutor (if I am not wrong!).
From the Fitness.java, I can see that the collaborators can be tracked through "public void setContext(Individual cont)" and with context[i].
So, what is the way to manage or copy these collaborators in "protected Individual producePrevious(int subpopulation, EvolutionState state, int thread)" of the MultiPopCoevolutionaryEvalutor?
My problem is to collaborate an individual with the current best collaborators from other subpopulation except the generation 0 where collaborators are selected randomly because fitness is evaluated yet.