Optimization

Note. The evaluation is often the most time demanding part of a GA.

And. We still need to evaluate n or 2n whole individuals for each process   ®   once or twice the effort of a sequential algorithm.

So. The model is not efficient unless we can optimize the evaluation.

How. Notice that the samples are the same for the whole population and for many generations.

®   Store as much information about the evaluation of the samples as possible, and use it to reduce the evaluation time.

®   Reevaluate this information after the process communication phase.