I have a problem where I am running a series of genetic algorithm optimizations (one per month for twenty years for a total of 240 optimizations). I am aware that the genetic algorithm has parallel options built in, but I am wondering which would be the most efficient way to handle this problem: - Run each optimization inside a for loop with parallelism turned on for the genetic algorithm options, or - Run each optimization inside a parfor loop with parallelism turned off.
Also, I've noticed that the genetic algorithm takes a while to get started. Is this because the initial setup has to be done in serial, and only the fitness function evaluations are done in parallel? If so, it seems like the second option would be most efficient. Any input would be greatly appreciated!