It now shows colour gradients for different values of the genes, and mutation doesn't cause it to crash.Here are some of the things that I hope to do with the program as I continue to work on it.
1) Demonstrate with no selection that bots breed. Done!
2) Then demonstrate natural selection for breedspan. Done! - run several times to compare outcomes.
3) Compare different grid sizes - 10,20,40,80 - compare outcomes/convergence. What is the effect, if any, of different population size on natural selection?
4) Impact of mutation on outcomes/convergence. The proportion of events that have mutations can be controlled, as can the maximum change that a mutation can introduce. Is there an optimum mutation rate that allows development but doesn't wreck existing functionality?
5) Impact of wandering - allow bots to swap places. Does this improve "evolution"?
6) Impact of virility/latency. Currently, a fit individual could spawn several surrounding children every cycle. How about having a limited number of offspring per cycle - or even one offspring per several cycles.
7) Limited evolutionary resources - increasing fitness in one area decreases fitness in other areas.
8) Impact of environment - select for distance from given gene to sum of x and y - and then other more complex interactions.