Darwinbots Forum

Code center => Bugs and fixes => Confirmed & Outstanding bugs => Topic started by: Testlund on January 11, 2014, 10:14:09 AM

Title: Mutations going crazy!
Post by: Testlund on January 11, 2014, 10:14:09 AM
I'm just running with default values with all mutations, including the new mutation features and within only 185000 cycles the bots DNA have totally changed.

For one bot I got only numbers and another one of the same species I got text DNA with a few numbers. This doesn't seem right. I have the mutations slider set to 1 / 32 too, so I think mutations should happen a LOT slower than this!

Title: Re: Mutations going crazy!
Post by: Botsareus on January 11, 2014, 05:08:07 PM
Turn OFF Delta2 mutations. Or set the change values as low as possible. I just have a thing for crazy mutations, they work for me.  :P
Basically what is happening is your mutation rates (themselves) are changing very fast.
Title: Re: Mutations going crazy!
Post by: Peter on January 11, 2014, 06:41:19 PM
Does it happen often delta mutations adjusting their own delta mutation rate can create crazy mutation rates?
Title: Re: Mutations going crazy!
Post by: Botsareus on January 12, 2014, 10:00:43 AM
Well first of, Delta2 mutations do not adjust there own delta2 mutation rates. Delta mutations do. But, delta mutations (old feature as apposed to delta2 which I added recently) are relatively rare. Also, delta mutations do not adjust the mutation run lengths at all. Feel free to use them though, I am not forcing this stuff on anybody and never will.
Title: Re: Mutations going crazy!
Post by: Testlund on January 12, 2014, 10:29:29 AM
I see this will take some time to learn. I was assuming the rates of the new mutations were similar to the old mutations. Maybe I should try one mutation type at a time and see what they do.

When you say "set the change values low" you mean higher numbers, right?

However if you continue to run this sim you should get an overflow error. Yay.  :dry:
Title: Re: Mutations going crazy!
Post by: Botsareus on January 12, 2014, 11:09:12 AM
Ok here we go,

Probability:

change of mutation % is how often the 'change of probability' happens

exponential is set default to 10 ^ (1 /1 )

This means that on reproduction the rates can get multiplied by 10 or divided by 10 for this specific robot.

liner is set to zero

This means that on reproduction the rates can get added by 0 or subtracted by 0 for this specific robot.

Mean / Stddiv changes the "mean stddiv" values instead of the probability values.

What to change is only liner, and changes the 'what to change' value

The interval for point mutations is the interval at which these are being done for point mutations (for point mutations it is not happening on reproduction, it is happening every time the interval is reached based on age)

edit: I would love for some one to write a proper wiki article on these...



What overflow error?  :blink: :cussing:
Title: Re: Mutations going crazy!
Post by: Botsareus on January 12, 2014, 11:22:07 AM
I am going to run your sim as is until Tuesday to see if I get an overflow error...

edit: I found one! I found one!
Title: Re: Mutations going crazy!
Post by: Botsareus on January 12, 2014, 12:55:16 PM
Testlund, I want you to go to Delta2 mutation settings and set "Delta2 cycle interval for point mutations" to zero and exit db and restart the sim by clicking 'startnew'. It will temporary fix the error while I debug it. This will also stabilize your simulation a little as this feature was primarily designed for zerobots.
Title: Re: Mutations going crazy!
Post by: Testlund on January 12, 2014, 07:17:57 PM
Ok, I tried this now and started a new sim and I got some more weirdness. Animal Minimalis is supposed to spin and search for food, but now it just sits there like a dead veggie. Also all DNA immediately changed to numbers as shown in this screenshot. Is this a feature? I would prefer text DNA so I can understand it.

Edit: The robots textfile has been overwritten too with these numbers!
Title: Re: Mutations going crazy!
Post by: Botsareus on January 14, 2014, 11:08:45 AM
How the hell did that happen if your total mutations is zero? Also, it should not overwrite the text file at all, very strange. Attached is a working animal minimalis file, replace it and run the whole thing again. It looks like your file was just bad to begin with.

Can I get your settings file and global.gset file?




Good news is, that 10th (or 12th was it?) overflow bug is now fixed with version 2.46.02BetaB in announcements.
Title: Re: Mutations going crazy!
Post by: Botsareus on January 14, 2014, 11:18:19 AM
Also, here are some (sane) values you can use for Delta2 other then using Delta and disable the feature completely.
Title: Re: Mutations going crazy!
Post by: Botsareus on January 14, 2014, 11:58:34 AM
Dude, I know what I am looking at. I knew the pattern looked familiar.
You somehow managed to load the robots .mrate file instead of the .txt file. Every time you edit the mutations the .mrate file is generated along side the .txt file. This will allow exporting robots from the simulation while preserving the mutation rates the program picked for them during the simulation (if you understand the concept that each robot in the simulation can have unique mutation rates)

How did you manage to load a .mrate file instead of a .txt yet it clearly says .txt?
Title: Re: Mutations going crazy!
Post by: Testlund on January 15, 2014, 08:30:44 AM
If that is .mrate code than what's happening is that DB writes that code into the bot.txt files too!

Here's what the 2 bot files looks like now:

Alga_Minimalis_v2.txt
_____________________

Code: [Select]
80
80
1892
3
1
1892
1
0
1892
3
1
1892
1
0
1892
250
75
1892
3
1
1892
1
0
1892
500
150
1892
250
75
1892
1
0
1892
1
0

Animal_Minimalis.txt
____________________

Code: [Select]
80
80
4816
3
1
4816
1
0
4816
3
1
4816
1
0
4816
250
75
4816
3
1
4816
1
0
4816
500
150
4816
250
75
4816
1
0
4816
1
0

People better backup their robot folder now!

Here's a zip file with my sim files:
Title: Re: Mutations going crazy!
Post by: Botsareus on January 15, 2014, 04:49:06 PM
Testlund, how are you saving the robot to get it to write to the .txt file? Very Strange...

Can you isolate how that happens, or at least guide me trough the steps you did?

edit:

Quote
People better backup their robot folder now!

Don't freak out, I'll skype you If I have to but we will get this fixed.
On the flip side, I kinda want to move on to other stuff and I am getting depression. Am I being too nice to this guys following every bloody bug they post or at-least trying to?
Title: Re: Mutations going crazy!
Post by: Testlund on January 16, 2014, 07:13:28 AM
I'm not doing anything strange. Just trying to run a sim the same as before. It's incredible if I'm the only one having this problem! The problem began when I ran the sim you suggested to create the settings file that you wanted. It began with version Darwin2.46.02BetaA.

I'm busy today but I will try later to run an earlier version to see if it makes any difference, so it's not the main install that has become corrupted in some way. My guess is that version Darwin2.46.02BetaA writes the .mrate data in the .txt files too.

Hang in there! I think there are only two things left to solve now. This problem and that run-time error 6.  ;)
Title: Re: Mutations going crazy!
Post by: Shadowgod2 on January 16, 2014, 09:37:03 AM
it happened to my animal min too a weak ago but it hasn't happened sense.
Title: Re: Mutations going crazy!
Post by: Botsareus on January 16, 2014, 05:30:04 PM
I may have something here, but I am drained today and going on a date tomorrow. Saturday is our best bet. Once again, thank you for keeping me sane guys.
Title: Re: Mutations going crazy!
Post by: rwill128 on January 16, 2014, 08:24:19 PM
You gotta take care of yourself first buddy! We can always roll back to an earlier version if things are messed up. Best of luck and stay well.
Title: Re: Mutations going crazy!
Post by: Shadowgod2 on January 17, 2014, 12:31:16 AM
couldn't have said it any better and good luck and have fun on your date :)
Title: Re: Mutations going crazy!
Post by: Testlund on January 17, 2014, 03:00:56 PM
Bots, I think you should take this weekend off and do what you feel you would like to do. You have no obligation to anybody here. You do this on your free time and I think you've done a great job! Lots of interesting improvements! I would say DB is pretty much finished now. It's the best A-life software that exist! It doesn't look like it's going to have any competition anytime soon. :-)

I just post bug reports to help out with the development of DB which doesn't mean people have to jump at it. You fix it when you have time. I also suspect I have a demon in my computer but that's my problem. Lol! I'm like a bug magnet for some reason.  :P
Title: Re: Mutations going crazy!
Post by: Shadowgod2 on January 17, 2014, 11:41:32 PM
unfortunately so am i but i usually work around them if it's an error type thing. :)
Title: Re: Mutations going crazy!
Post by: Botsareus on January 18, 2014, 10:40:06 AM
Quote
It doesn't look like it's going to have any competition anytime soon. :-)

Well, except from Numsgil in one year...
Title: Re: Mutations going crazy!
Post by: Botsareus on January 18, 2014, 10:54:50 AM
That mrate problem was caused for the following reason:

The program redirects to the robots folder if it can not find the robot in its current folder. The mrates data (on save) should redirect as well. However it was redirecting to a .txt file instead of a .mrate file (ding V8)

Also, I think I was a able to stabilize point mutations a little more to prevent overflows by reducing how much max time (in cycles) the program is allowed to make between point mutations.

Anyway, check out announcements post of version B to find version C.

Have fun. I am going to take the advice and take the rest of the day off.
Title: Re: Mutations going crazy!
Post by: Botsareus on January 19, 2014, 01:37:36 PM
On a serious note: Did that fix it?
Title: Re: Mutations going crazy!
Post by: Shadowgod2 on January 19, 2014, 11:44:33 PM
i only had it happen to me once and can't recreate it though i think it did testlund any issues on your end?