Something I've been wondering about

[Testlund]

The default values in the mutation rates are set to 5000. Does that mean the changes in the bots genome will take place after the simulation has run 5000 cycles?

[shvarz]

No, that means that each time a bot divides each command has a 1 in 5000 chance to be mutated.  


Well, maybe not exactly that, but this explanation is close enough...

[Numsgil]

As shvarz hinted, the truth is a bit more complex.

If all the fields are set to 5000, each command has a closer probability of 1 in 250 ish to mutate.

[Testlund]

Ok. Not sure I understand what it means anyway. Do you mean 1 change every 250 cycles or every 250 command? How many commands can a bot have every cycle? Doesn't every field represent individual commands not connected to each other? For instance, if all fields are set to 5000 and I change the Minor Deletion field to 1000, then only Minor deletion will occur every 1000, the rest should happen every 5000?

[Griz]

Nums is saying it is a probability ...
1 in 250 or 1 in 500 ... that a mutation will occur.
I don't think this has anything to do with cycles.

[PurpleYouko]

Has nothing to do with cycles.
It is all about probability.

The number you put in is used as a maximum for a random number generator.
5000 will generate a random integer value between 1 and 5000
100 will generate a random integer value between 1 and 100

When a robot reproduces, each command, operator and function in its DNA is parsed in sequence and a random number is generated. If that number is 1 then the value is mutated. if not then it is copied over exactly.

The lower you set the values in the mutations window, the more mutations you get. Set them all to 1 and every single point in the DNA will be mutated every time the robots reproduce. 1 chance in 1 is a 100% certainty
Set it to 1,000,000 and mutations become extremely rare. 1 chance in 1,000,000 is pretty improbable

[Numsgil]

Mutatations, unless they state otherwise, occurr at reproduction.

[Testlund]

Ok. Got it.