GitLab

Say we have a problem where a large number of solutions do not return proper objective values. In the first generation opt-pilot dispatches n jobs, where n = initialPopulation. When say 50% give an infeasible solution, then n/2 jobs are dispatched of which 50% fail, then n/4 are dispatched, etc., until the full population has a solution. This will mean that many cores are idle.

It would be better to dispatch as many jobs as possible (nr of cores) until there are more than n feasible solutions.