Darwinbots Forum
Welcome To Darwinbots => Newbie => Topic started by: Greven on April 27, 2005, 10:24:54 AM
-
I had just written a very long post, about my background in Artificial Life!
And what happened? My computer did shutdown, because of low battery
(Do always remember to plug it in, and do not remove you possibly warning!)
OMG :angry: :angry: :angry: :angry: I AM SO ANGRY!!!!!!!!!!!!!! :angry: :angry: :angry: :angry:
:shoot: GREVEN!
So it would only be some of it.
My story begins 5 years ago. At the age at 16 I discovered AL.
I did so AL-programming based on 'Tierra' by Tom Ray & 'Avida' by CalTech, later on 'Echo' by Holland.
A few years ago I found DarwinBots among others (DarwinsPond etc.). No help at that time, I lost my interest fast, dispoinnted I deleted it, I could see it had potential, but with no help....
January 2005 I found DarwinBots once more! This time with help, and very improved!
I have read book after book, article after article on artificial life and search the web thin for artificial life simulations/software! And I must say that the most 'living' software is DB. :D
(This is a very short version, alot is left out! Because I am still :angry: )
~Greven!
-
Welcome Greven!
Glad to see new people flowing in!
-
Welcome!
-
Hi, glad to see you like DB
-
Welcome!
Out of curiosity what are the other programs like? Anything we could learn from them good or otherwise?
I've tried some of the simpler Alife stuff, but once I got hooked on DB, I've tended to focus solely on it.
Endy :)
-
Thanx all B)
I'm on work right now, so it will be very short.
The first 3 or so is purely based on 'Tierra' or 'Avida'.
The fourth on 'Echo' and 'Tierra'/'Avida', a combination of the two, not very succesful.
The fifth based again on 'Tierra'/'Avida', very much like them.
My sixth were made just before I re-discovered DarwinBots, based on 'Echo' and lot of my own ideas, a few other amatuer AL (Ant simulation among others, can't remember all of them), and 'E-den'.
This AL is called 'SoL' (Sun in danish), short for 'Simulation of Life'.
'SoL' containes graphs, genalogic-tree, SEED-value function (very important for me, and every ALifer), a counter for all different bots (if a daughter is different from its mother, it will increment this counter and save the genome to a file) and alot of other stuff.
'SoL' is not about behavior as in DB, but on survival, meaning if a 'bot' is over its treshold for reproduction, it will reproduce. Bots fights each other for food, the winner of the fights is determined by the genome, and do not mean complete death of one or the other, just like DB. (very short description)
For me speed is essentiel, and I did optimize my code to around between 600-1000 cycles/sec, for about 400 bots (cant remember the exact numbers), for a genome of length 100.
The genome consist of 8 different DNA-commands, A, B, C, D, a, b, c and d.
Different combiantion of the 8 DNA-commands gives different functions etc. (like in the real world, 3 DNA-bases gives a amino-acid and many aminoacids gives a protein etc.)
This system is very versatile, and I think it would be very nice to implement in DB, and gives better mutations, though it would be very hard to do it.
'SoL' is not finished yet, and I think it never will be.
-
IMO the graphs and other ways of analyzing the simulation are sorely missing from the current DB. In some ways it's just too hard to figure out what's going on in the current simulations.
When I have a few seconds I'll figure out what I like with the other ALsims and see what can be applied into DB.
-
I agreed on this! We need alot more info, maybe an option to turn it off if you dont want it.
-
Totaly Man , I am working on "Pusher Bots" , simuler to "Sol" , in the fact that there is infinate! energy! (meaning if a 'bot' is over its treshold for reproduction)
But I keep it tight on space ,and my dna still looks like computer code
For now actualy I have finished a new controls hack and playing cr
anyway I like that Project Sol good stuff.
-
And Yea I like to see DB save every robots history tree just like smexe does. (It does not delete executed files of the hard drive)
Edit: Stuff to do post for PY or NUM (yea like its ever going to happen)
-
Hmm, my sims go through about 50,000 bots for each million cycles... The tree created would be un-usable and would take up tons of space if DNA of each bot is saved.
-
My tree only took succesfull 'bots', meaning if they got over a certain treshold they would be saved, and incooperated into the tree. Much better, than saving every mutant, which would be crazy, 'cos some of my sims reach beyond 1,000,000 bots, and only 500 or so were treshold genomes, meaning they got at least 10 (I think) or over in population, meaning they were more or less better fit than others (I know it is rather arbitary, but still the only way to do it, when there is no exact fintness function)
-
Botsareus do you have a hp on 'PusherBots'?
-
My tree only took succesfull 'bots', meaning if they got over a certain treshold they would be saved, and incooperated into the tree. Much better, than saving every mutant, which would be crazy, 'cos some of my sims reach beyond 1,000,000 bots, and only 500 or so were treshold genomes, meaning they got at least 10 (I think) or over in population, meaning they were more or less better fit than others (I know it is rather arbitary, but still the only way to do it, when there is no exact fintness function)
You're right that it's a little arbitrary. I would stear away from any artificial fitness function. The point of DB is that there is nothing that decides fitness beyond a bots ability to survive, just like in real life.
When you create artificial criteria, you can generally speed up evolution towards that simple goal, but you'll rarely, if ever, see anything really interesting or new that you didn't expect to see.
-
Whats "hp"?? you mean help files or hyper link? I resently wrote a "Read Me"...
I only got the physics done really and some code-editor work, so not really mutch too see yet.
-
You're right that it's a little arbitrary. I would stear away from any artificial fitness function. The point of DB is that there is nothing that decides fitness beyond a bots ability to survive, just like in real life.
When you create artificial criteria, you can generally speed up evolution towards that simple goal, but you'll rarely, if ever, see anything really interesting or new that you didn't expect to see.
I dont think you understand me Num.
I did mean there is no fitness function, but to decide which genome to save to disc, you have to make arbitary decisions etc.
I did mean if a certain genome's abundance got over 10 (or so) it would be considered more or less succesful and saved to the genalogical (?) tree, because genome less than 10 would mean it had a hard time to survive.
-
Whats "hp"?? you mean help files or hyper link? I resently wrote a "Read Me"...
A mean a homepage! Were is the Read Me
-
Its not uploaded to my website yet bc its not done yet. All you can find on my home page is an old virsion of a self-mutating-application that reproduces and opens copys until you run out of memory:
[a href=\"http://plasmic007.tripod.com/smexe.html\" target=\"_blank\"]http://plasmic007.tripod.com/smexe.html (http://%20%5burl=\"http://plasmic007.tripod.com/smexe.html\"%5dhttp://plasmic007.tripod.com/smexe.html) [/a]
O yea it was sesigned for windows xp and 600x800 resolution , with no other programs open. And the taskbar on the buttom of the screen. Otherwise it wont properly shutdown its own fetal crashes.
and here is the source code under construction for PusherBots (You will notice I beleave in small spaces to save on speed; But I will have some insane mutations in there so you will still get a lot of waird stuff; The most important thing about it , even if the dna code is blank , the robots will still be forces to reproduce, The only way the robots can really kill themselfs is if they really drift off beiond the edges of the circle; Talk about crazy mutations: The Parant does not survive reproduction. )
Topic on Pusher Bots (http://s9.invisionfree.com/DarwinBots_Forum/index.php?showtopic=70)