Does anyone know a (preferably fast and or parallel) genetic algorithm library? I am looking for extreme ease of use. The only real thing I want to do is define the fitness function. Other than that, I just want to let the sucker run. I could write my own (in fact, I have before) -- but if I could just plug'n'play with some possible parallelism (most libraries seem to use MPI), I would be a very happy camper. Any recommendations?

I don't have any specific recommendations for libraries, but some general advice:

1) You'll typically need to make heavy changes to the fitness function, the simulation, and the way the GA works to get the results you want.
2) Letting the GA run randomly may not lead very interesting results unless you supervise it, so raw parallel computation power isn't necessarily the right thing.


These kind of algorithms are simple to write, just knock one up... It should only take a few hours (especially since you've done it before).

[Edited by - alexjc on November 20, 2007 4:50:54 AM]