Just a question on design about darwinning

[dorkrawk]

Hey! Thanks for checking out the library. You're totally right that many of these examples are probably wasting time recalculating fitness each time its used. Because I'd like Organism#fitness to be as flexible as possible I think the best solution would be to memoize the #fitness value. Something like:

def fitness
  @fitness ||= attr_a + attr_b + att_c
end

I'll update some of the examples with this pattern.

[dorkrawk]

Here's an updated example: cdd96c0

[illtellyoulater]

Thanks a lot for clarifying. I've experimented with darwinning to build a system which creates small musical (MIDI) themes, that evolve after being played and rated by the user, so much similar to your cookie example (but I had checked it out only later :) ). So I had temporarily put the rating code inside the #fitness method, and saw how for the same member/object I was asked to listen it and rate it many times (presumably as many times as the number of comparisons with other members it is going through?).
So yeah, I had fixed it by only calculating it first time and then just letting fitness return it for subsequent times, as you did above.. well maybe it could be fun if I could add it to this repo examples.. :) It's using https://github.com/arirusso/unimidi multi platform midi lib.

I'm seeing though, that it gets fastly stuck on the same tune. I've googled around a bit and read that in some conditions genetic algorithms tends to quickly get stuck to local optimum... but again I don't know much about this yet.. (thanks for having given me the opportunity to learn something on this fascinating subject with your gem!).

[dorkrawk]

That's a really cool project! One of my first genetic algorithm projects back in grad school worked with MIDI as well (but it was written in Python). Here's a small paper I published on it: http://socrs.cdm.depaul.edu/2011/program/papers/paper_13.pdf