Darwinbots Forum

Code center => Suggestions => Topic started by: Botsareus on October 27, 2005, 07:18:17 PM

Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 07:18:17 PM
Someone try running FirstBot with mutation rates set to "5", I know I am crazy, but nothing prevents a user from doing this so it eather needs to be ristricted or fixed...

...What happens is after a while the robots mutate dna is mutating so much dna that it is simply too slow to run the simulation. Its like its frozen...


I already fixed it by "caping" how much "dnamakespace" will work with... on my new virsion...
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 07:21:02 PM
How would we even fix this?

I mean, of course it's going to slow down the program!  Mutations are generally written to be easy to debug, not fast.  Mutations are among the slowest code.

I think you could run bots with 10 000 DNA and have no noticable slowdown, but bots with 100 mutations each generation are going to slow the program down alot.
Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 07:24:00 PM
Anyway , I found a 'tie overflow' problem, and some more memory storage thats above 32000... not enough time now so...
Title: Mutating too much Rubble; Bug...
Post by: Ulciscor on October 27, 2005, 07:42:05 PM
Not really a bug. In fact not at all  :blink:

{Marked for removal}
Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 08:05:40 PM
exsept when the program is frozen , its the same thing as a crash , so its completly unusable...

Unusable programs - No problem

Why does Num always make it sound less troublesome?

Btw: Num is right when he sayed (bau translation) If a bot runs with 10000 dna its actualy still usable.

But he left out the part of what happens when a robot with 10000 mutates with "5" (do I dare say "2") mutation rates
Title: Mutating too much Rubble; Bug...
Post by: Ulciscor on October 27, 2005, 08:06:37 PM
Is there any way to fix this though? The program will slow down when the DNA is too large. That's not a bug it's a fact.
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 08:07:27 PM
I don't think it's the DNA being too large.

Someone make a bot with 10 000 DNA and see how well it runs.

I think it's the number of mutations.
Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 08:10:00 PM
exactly Num. When a program trys to mutate something like 10000dna it simply can not complete in under 5 hours or so.
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 08:11:52 PM
Right, so don't do that :P

Seriously, I don't see how you can fix it.  Or even why you should waste your time trying when 99% of the time the slowdown is in other parts of the program.
Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 08:15:11 PM
Simple,

I need this to see how TheONE with evolve with its mutation rates oxilorating(wrong spelling, py) between x1 and x27 so I can intergrade the HH I was talking about , and evolve first bot more productively (that is takes me less then 10 years)


If my vision of people using my methods in the future is falce , then yes I am simply wasting my time typing letters and numbers on a keyboard on the internet.


EDIT: MAJOR EDIT, I am to ZzzZz its THEONE ,
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 08:16:56 PM
Well, if you're set in your ways and don't want to change, you'll need to find an application called a "profiler".

Then you run the profiler and the program.  The profiler will tell you where you're spending all your clock cycles at.
Title: Mutating too much Rubble; Bug...
Post by: Botsareus on October 27, 2005, 08:20:57 PM
It is on "makespace" , and I already "caped" it. I was justing getting everone to do it, gess it did not work.


I know you like your advanced sounding high tech debug stuff num.

But all you really need to do, is press "break" a couple of times, more 60% that you will get lucky. So after 3 trys you will get it without any "profiler". another reason I like vb so much.
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 08:23:18 PM
Well then good!

But no, I don't think we'll be capping anything in the regular program.
Title: Mutating too much Rubble; Bug...
Post by: Ulciscor on October 27, 2005, 08:25:06 PM
OK I don't think this classes as a bug. Maybe it should be moved to suggestions if people want to discuss mutation of varying DNA length.
Title: Mutating too much Rubble; Bug...
Post by: Numsgil on October 27, 2005, 09:14:11 PM
Pro'lly.
Title: Mutating too much Rubble; Bug...
Post by: Ulciscor on October 27, 2005, 09:14:41 PM
{Moved}
Title: Mutating too much Rubble; Bug...
Post by: Endy on October 27, 2005, 10:27:17 PM
Why would you want to use such a high mutation rate? This is a kind of unrealistic thing to even be talking about. Yes, a user could select for super high mutation rates, but its simply illogical to do so, nothing is very likely to evolve under such circumstances.