Comments on two-way reproduction optimality?

[Dario Pellegrini]

Hi,

I saw the code that spawns a child by randomly mixing the genome of two parents and I was wondering if the two-way mixing achieves the fastest improvement of the behavior gen after gen.
Do you have any experience/comments on one-way reproduction (the genome of a single parent is copied and mutated) or multi-way reproduction (the genome from three or more parents is mixed and mutated)?

Thank you for your time and patience.

[David R. Miller]

That's an interesting question. There's a parameter in the biosim4.ini config file that selects asexual (one parent) or sexual (two parent) reproduction. You can even flip that parameter while the simulation is running. From a few brief experiments, I got the impression that two-parent reproduction usually resulted in faster adaptation to environmental changes than one-parent, but not always. That setting seemed to interact with other parameters in ways that were not immediately clear to me. Having more than two parents would require some tweaks to the code and would allow some interesting experiments.