Code center > Bugs and fixes

Mutation rates again

(1/1)

shvarz:
I made a bot to test mutation rates.  Here is the code:


--- Code: ---cond
*.robage 100 =
start
50 .repro store
stop

cond
1 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
2 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
3 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
4 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
5 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
6 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
7 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
8 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
9 1 >
start
1 2 3 4 5 6 7 8 9 0
stop

cond
10 1 >
start
1 2 3 4 5 6 7 8 9 0
stop
end
--- End code ---


Basically the bot is supposed to divide only once and then I could see what the "average mutations" graph was showing.

I set the rates so that it predicted 1:3 bots being mutated, run the sim and the graph showed ~.015.  That sort of made sense, because 50% of bots were parents and 50% of bots had 1/3 of them mutated.

Then I increased the mutation rates so that it predicted 1:2 bots being mutated.  The graph still showed ~.15.  That was troubling.

Then I increased the mutation rates so that it predicted 1:1 bots being mutated.  The graph showed ~2.  Not 0.2, but 2.  This is very weird.  Something is screwy there.  Or am I missing something?

Numsgil:
I'm experiencing similar curious effects.  I'm working on it now.

Navigation

[0] Message Index