Hi

[Schigum]

Hi,
my name is Sebastian, I'm 21 years old and from Germany.
I was reading about artificial intellegence/life and searched for some evolution-simulation, when I found Darwinbot. I was pretty impressed by the program and its possibilities (from "simple" evo-sims to competitions in leagues) and I sit in front of my screen for hours, trying to figure out working conditions and costs.
After I found some "Div by Zero - Errors" I decided to take a look at the Forum, found a new version and... cried a few tears, cause my 15hrs-sim didn't work fine with the new version   .
But I like it even more now *g*
I'll try to do my part in finding bugs and so on.

So far,
Sebastian

[shvarz]

Welcome to the DB, Sebastian!

[EricL]

Hi there, welcome to DB!

I'd be curious to know what version you were using before.   Thanks!

[Schigum]

I'd be curious to know what version you were using before.   Thanks!

Mh, it was an 2.42.9 without an extention, think I downloaded it from the download-link at the Wiki about a week ago. Still have the DarwinBot.exe File, but none of the Sims wich didn't work fine.
One of the "Div by Zeros" was the one with the logarythmic agecost increasement allready fixed, another one only occured if the movement vectors where disabled, but it seemed to happen randomly, sometimes after a few seconds, sometimes after an hour. And I was able to continue the sim by loading the error-file (still have one of the error-files).
Now I'm using 2.49.9s, where this "random" Div by Zero not happend yet.

[EricL]

Okay, great.  The crashes in 2.42.9 don't surprise me - I've fixed quite a few since then - but the sim file formats should be forward and backwards compatable.  I'm surprised you had problems loading a sim produced by 2.42.9 into 2.42.9s (was it a problem in loading or something that happened later?).  That should not happen.  If you find the old sim or can re-create the issue, I'd be very interested in taking a look at a sim produced by 2.42.9 that has problems in 2.42.9s+.

Thanks.

[Jez]

Hiya Sebastian, glad you joined us.

Hope you can recover some of the problem sim saves from your recycle bin for Eric to have a look at!

[Testlund]

Welcome to the forum, Sebastian! You should grab 2.42.9t, it's allmost perfect. If you have problems finding out the right costs I can tell you how I do it. I just start with one bot, usually Animal Minimalis, R Fisannis or a zerobot with a certain amount of energy. Then I put just one cost in one field and let it run to see how long the bot will last without energy. I then tweak the various costs until I find a balance where the bot will die after a certain amount of cycles. I can upload a save for a couple of evosims I'm running if you want more details, but maybe you prefer to do it on your own.  

[Schigum]

Hi Testlund,
meanwhile I have found "working" conditions for sims by try and error.
But I would love to have some other constellations of costs and physics, to compare with, eventhough I have still problems with applying the costs (posted at the bugs-forum).

[Testlund]

Alright, you can check this one out if you want. Hopefully the bots are saved in the sim, or I need to provide those too. After you load this you need to go into the UI and check the following on the general tab for NRG Provider: Disable vision, Disable voluntary movement and Disable reproduction. Those settings aren't saved in the sim.

[Testlund]

Damn, I messed that up. Gave you the wrong file. The above will kill all your bots after awhile. Sorry. Try this one instead.   By the way I meant the settings on the Species tab, but this time it looked like it was saved in the sim.

Also the costs here are balanced for day/night cycles set to 25000 incase you want to use that, but is not necessary. It will just be more taxing on the bots if you use it. Note that this is a new concept of an evosim that Eric mensioned recently, which is supposed to force heterotrophs to evolve better. I'll upload my autotroph sim later.

[Testlund]

Here's my other one. It contains 1000 autotrophs only and it's been running for 250000 cycles, but you could start anew if you want.

[Schigum]

Thanks alot,
by the way, how many Bots/Vegs do you usually use in an evolution-sim?
I tried to keep the number pretty low (summary not more than maybe 400) to keep the simulationspeed at above 50 cycles per second. Looking at your 1000 Veggs/Bots I feel pretty inpatient

[Testlund]

Well, the more bots you have the bigger the chance one well develop to do something useful. If you have a dual-core machine you can run two instances with 500 bots each connected with teleporters, to speed things up. I don't have dual-core on the machine where I'm running this sim, but the speed is still decent in the beginning, 4,5 cycles/sec. It will slow down once the bots have developed a lot of dna and start doing stuff. Also I use to put drifting shapes in at a later stage which will slow it down even further, but I'm the patient kind. It helps to have two computers for this stuff, so you can run it non-stop. Last time it took a couple of weeks before I got a good population of bots.
The heterotroph kind of sim I'm running on this computer, on the other hand, doesn't seem to do that well. I only got 100 bots left. I guess it needs more tweaking.