@fmder ok, thanks, missed the part where the individuals were reevaluated. Do you think there would be another way of storing the fitness value at first evaluation. My optimisation use a lot of computing resources, reevaluating the individuals in the history is much too expensive
@shivamkalra Eh, probably not. You really want either:
To parallelize the fitness calculation itself, in which case the MPI would go in your evaluation function.
To have multiple fitness functions running in parallel, which means you want an MPI based map function. Either of these would be pretty easy I think, but they aren't what I have.
@shivamkalra That said, if you want the code, I'll be happy to give it do you. What's the best way to do that? I'm planning to open source the whole thing at some point, but I haven't quite gotten there yet, so I don't want to do a github repo yet.