Darwinbots Forum
Welcome To Darwinbots => Newbie => Topic started by: Peter on June 24, 2007, 04:09:41 PM
-
Hello everybody, I'm new and I've got some questions.
I don't speak english everyday, so my grammar isn't that well, I hope you can understand it.
How can you use a virus to hurt an enemy, but how can you let the virus know when it's inside your own bot, and than to deactivate itself if it is.
This steals slowly the energy of a bot. I only want it to steal energy from the enemy's.
Now there's a second question if you stop the motherbot from firing the virus in this way, can you use another gen to create a virus with it.
'Small but with increasing numbers inside a cell will steal more and more nrg.
cond
start
.vshoot inc
*.thisgene .mkvirus store
stop
I've got this virus out of the wiki, if you were asking. I just want to now how to properly use it.
-
The idea is to use conditionless programming to multiply the destination of the store to make it 0 when you don't want it to activate. See this (http://www.darwinbots.com/WikiManual/index.php?title=Conditionless_Bots) page on the wiki.
-
The idea is to use conditionless programming to multiply the destination of the store to make it 0 when you don't want it to activate. See this (http://www.darwinbots.com/WikiManual/index.php?title=Conditionless_Bots) page on the wiki.
Well, thanks I don't get completly how this can help me, but I found at the virus wiki something (very simple) to put into the bot. Simple and works.
cond
*.robage 0 =
start
7 989 store
stop
bot
cond
7 *989 !=
virus
Well, to say the truth, I don't get much from the conditionless bots section.
20 .up *.hit mult store
.hit *.hit mult dec
All I see is a number some stars(or how do you call these(*) and some words. Store it is storing, hit it is hit, mult to multiply, up to go up I gues. Dec ?? no idea. Why would this, so says the explanation go up when it's hit. When up is in front, it looks to me kinda strange if he is only going up when hit is true, becouse up is in front of the code.
-
If the asterisks (* aka stars ), dec, and other DNA bits throw you for a loop, it might be a good idea to read through the documentation of the DNA before starting with something like viruses. Walk before you run Check out this (http://www.darwinbots.com/WikiManual/index.php?title=Robot_DNA) wiki page and look up any commands that you don't understand.
-
20 .up *.hit mult store
.hit *.hit mult dec
Conditionless bots use maths to do stuff rather than conditions. It's up to you which method you choose in the end.
Asterisks (*) reads you the value of a location so *.hit gives you the value of the .hit location (sysvar 201) (sysvar = system variable)
The value of .hit, will be either 1 or 0 so the first line (20 .up *.hit mult store) will only move if *.hit returns the value of 1 (otherwise .up (sysvar 1) will be multiplied by 0 and return sysvar 0 which doesn't exist)
The second line uses dec (decrease) to subtract 1 from .hit but only if *.hit = 1, (otherwise multiply sysvar by 0 again)
As a maths equation it would read (20 (.up *.hit mult) store) and ((.hit *.hit mult) dec)
The maths is all done in Polish Reverse Notation by the way so instead of;
1 + 2
it's;
1 2 +
Bit strange to start with but you'll soon get used to it!
Neither dec (decrease) or inc (increase) need a store to change a sysvar unlike most other operations.
Hope that helps a bit.
sysvars can be found here (http://www.darwinbots.com/WikiManual/index.php?title=Sysvars)
-
Well, thanks for the help, Numsgil an Jez. You're right first learn to walk, before you can run. Maybe I'am trying something too fast.
If have got some another question's
First one, in the first gene I have in the robot I've got mkslime and mkshell, the strange thing is the shell is fastly going down and the mkslime is stable. Is'nt it supposed to be reversed(mkslime going down).
cond
*.robage 0 =
start
7 989 store
200 .mkslime store
200 .mkshell store
stop
My code in the first gene.
There is another question. I have created a shortbot, first based on animal minimalis from Numsgil, later I created a new one myself. The question is, my robot is beating up the whole shortbot-league, I'am just starting thus it would be strange to have a good bot already,is there any rule I overlooked. In the rules of the shortbots stands there aren't any rules exept having no more then 6 genes. Maybe there are some other rules somewhere .
Edit:
I was speaking to fast, I was simulating a sim(logic) with mine and the league bots and every time I won, I just set up a shortbot league, I was beaten by furiae(I knew I missed one). My virus wasn't working well on it. The only bot jet.
Third question.
I've got this code to keep shell at 200. Why isn't it working well. There's nothing happening
1 .mkshell 200 *.shell sub mult store
-
If that gene you posted only sets mkslime and mkshell when robage = 0 then I'd imagine that you are using F2 or shotbots not tiebots. Slime reacts to being hit by ties and shell reacts to being hit by shots so I'm guessing it's (shell) getting eroded from stray shots. (Ties tend to be more accurate than shots as well).
Henk set the rules for the shortbot league and that's the long and short of it! If you have a bot that you think might do well in the league please feel free to post it in The Starting Gate (http://www.darwinbots.com/Forum/index.php?showforum=37) so it can be given an official ranking.
You should be able to download the official league easily enough but if you have problems please say so.
It wouldn't be the first 'first' bot to go straight to the top of it's relevant league btw, there are lots of members here who have pulled the same trick...
I think your code would only work if *.shell = 199 otherwise it would be changing the sysvar .mkshell to a different sysvar
(1 (.mkshell (200 *.shell sub) mult) store)
200 *.shell sub .mkshell store ---- (200 *.shell sub) .mkshell store
Might work better but it's getting a bit late so I'll check that tomorrow!
-
Shell is becoming smaller without it being atacked by others, slime is staying stable. In the wiki these are reversed, slime is supposed to become smaller and shell is stable. If I ad slime, shell is being added and so the other way.
I've entered my bot, I'am not sure if it's doing very good in the league there where some compititors missing, in the league I downloaded
About the code, Well thanks it works, it's beginning to make sense. At least I'am again thinking I get it.
And to keep you busy (or someone else)
Right now I'am running a zerobot-simulation, the robots can already reproduce. Now they can reproduce I've put them in another sim, becouse the mutation rates in the first are way too high for a gene to stay stable.
What are the maximum mutation rates for the evolution of a bot, to speed up the sim and keeping the original reproduce genes more or less intact.
For point
delta
copy
insertion
reversal
minor deletion
major deletion
I'am having trouble with it are low values(high mutation) is overmutating the population. At high values (low mutating) I'am thinking, there's nothing happening and there is'nt happening really much.
Furthermore, for the creation of reproduction genes I have used the smallest room with 100 zerobots. Is it for the further evolution better to have a larger room with more (developed) zerobots or does'nt it care. If so what's the best, I'am at least not using the maximum room, I want to see the kind of stupidity of the zerobots.
Well, anybody to anwer this (stupid)question, Thanks, oh and thanks to Jez for answering the last question.
-
Mutation rates are something of an art form. 1 in 5000 is a reasonable starting point. In general it's better to have mutation rates too low than too high. Too low just means it'll take longer. Too high means that everything will fall apart as the genome won't be stable enough.
When you're first starting out, most of the settings won't matter because the bots can't do much more than reproduce. Costs and feeding rate are about the only two things that make a difference. So feel free to set up pretty much whatever environment you want, and see if you can get your bots to evolev inside it.
-
The bots missing in the shortbot league download should just be the ones I couldn't find. When I get round to entering your bot in the shortbot league, and there are a couple of bots in the starting gate already that will get entered in the league first, I'll be using the same file.
When I do that I'll also check the shell and slime thing and adjust the wiki entry, there have been a couple of minor changes in the past to shell and slime so it's quite possible the wiki entry never got updated. Well spotted!
-
I think I'am not patient enough for the zerobot-evolution, All that happens is a bot is suddenly moving(jay it's moving) it takes the oppertunity to reproduce and reproduce till it is forbidden by the program to reproduce, then it's killed by the dynamic costs I inplented. Sometimes one of the larger sipplings(probably couldn't reproduce due to space)survives and the same thing is going on, till they're all gone. Then another one is moving doing exact desame thing. I know there will come a time they will be smarter but that's probably costing much time. Even the "virus"-bots in the 1000 hour simulation from EricL aren't very smart and doing random stuff. And that's 1000 HOURS.
My question for a simulation I want to start with instead of one I want to let two specie's survive. With the use of an double alga. It isn't working I can't figure out what's wrong.
So what's wrong. If someone want's to tell me. I just don't get it.
' Alga shell/poison protection
'
' Vegetable
'
' This robot has three genes,
' necessary for it to reproduce,
' make a choice to take poison/shell,
' And create poison/shell protection.
' Energy is given to it by the program,
' by checking the "autotroph" box.
cond
*.robage 0 =
start
1 rnd 49 store
end
cond
1 *49 =
start
1000 *.shell sub .mkshell store
else
1000 *.poison sub .strpoison store
end
cond
*.nrg
6000
>
start
50
.repro
store
15
.aimdx
store
stop
end
-
Stop not End, the program only sees the first gene!
End is only used once at the end of all the genes.
Pause your sim, select one of the bots on screen then right click on it and choose 'show robot info' now click 'dna' and you can see the genes the bot is using.
-
Thanks
Oh , now I feel stupid I've got to use stop instead of end.
And there's even another question.
cond
*.numties 1 =
start
1 .tienum store
314 2 *42 mult 1 add mult .fixang store 'allebei dezelfde kant op!!
50 .sharenrg store
stop
I was trying to create a multibot and I can't get them to go to the same side
*42 supposed to be either 1 or 0. One of the two has 1 the other 0.
This is what I'am trying to do.
O--O
O_O
| . |
When I try they are doing this
O_O---
|
The anwser will probably be simple but I just don't get to the solution.
-
314 2 *42 mult 1 add mult .fixang store 'allebei dezelfde kant op!!
Looks like it's working to me. One turns a quarter circle and the other turns half a circle. But I think you probably wanted something more like this instead:
314 -1 *42 mult 1 add mult .fixang store 'allebei dezelfde kant op!!
-
It is not working with me. And what am I exactly seeing wrong in the code
314 ((2 *42 mult) 1 add) mult
314*((2*1)+1)=942
314*((2*0)+1)=314
This is what I am seeing in my code.
314 ((-1 *42 mult) 1 add) mult
314*((-1*1)+1)=0
314*((-1*0)+1)=314
This is what I am seeing in your posted code.
What am I seeing wrong.
Practical they're doing desame in the simulation. How does this work.
In fact I posted this twice, the first time it got stuck, is this happening more often.
-
I don't see anything wrong with what you are seeing in the code. Are you letting the ties harden first? That would cause a problem. Perhaps Nums will spot something when he looks again, he's much better at the maths bit than I am.
Post the code you are working on if you can't figure it out still and I'll have another look, it's often easier when you can see the whole bot and what it's doing.
Sometimes there are problems posting here, not very often though, I think it's just the Invision forums getting too busy, at least that's the message I got when I tried to post something earlier!
-
Well you asked for the code here you get it.
' reproducing while in singe form
cond
*.numties 0 =
*.robage 0 !=
start
7 42 store
50 .repro store
'1.readtie store
stop
' repro second
'cond
'*.numties 1 =
'*.robage 0 !=
'stop
'50 .repro store 'not used jet
'stop
' Child connect to mother
cond
*.robage 0 =
start
.tie inc
stop
' sharing energie in organism
cond
*.numties 1 =
start
1 .tienum store
314 1 *42 mult 1 add mult .fixang store 'both siding desame direction.
50 .sharenrg store
stop
' body strengthening
cond
*.nrg 2000 >
start
100 .strbody store
stop
'waste desposal
cond
*.waste 10 >
start
*.waste .shootval store
-4 .shoot store
stop
Can you now tell what's wrong?
-
It is not working with me. And what am I exactly seeing wrong in the code
314 ((2 *42 mult) 1 add) mult
314*((2*1)+1)=942
314*((2*0)+1)=314
This is what I am seeing in my code.
314 ((-1 *42 mult) 1 add) mult
314*((-1*1)+1)=0
314*((-1*0)+1)=314
This is what I am seeing in your posted code.
What am I seeing wrong.
Practical they're doing desame in the simulation. How does this work.
In fact I posted this twice, the first time it got stuck, is this happening more often.
Yeah. you're right. Guess I haven't been programming bots enough lately.
-
fixang always seems to be set to 32000, at least according to the console
(select and rclick on bot and open console, type '? .fixang' in console
without quotation marks and it will return the value).
Why fixang is remaining at 32000 and not being changed when you enter a new
value is a mystery.
This raises the question of why your bot is turning at all and why, when they do turn, they are
not turning by the same amount.
Although in fact your bot is only setting its direction after about 30 turns,
by which time the ties should be hardened anyway and it shouldn't be able to
turn.
I have just checked this against Dimacheri(MB)(Shen) and it is suffering from
similar problems.I think the reason that bot works and manages to change the
angle of its ties is simply because it moves.
I suggest we wait and see what Nums has to say about the problem, if it's a
really easy problem to find and fix then we could be lucky, otherwise we'll
just have to be patient.
The only mistake I noticed in your code btw was *42 was either 0 or 7 but that would have just meant your bot should have turned 3.75 rotations instead of .75 of a rotation!
-
How is the 7 coming there. , strangely enough it's doing as one, that's a coinfidence.
So, I gues there's a mistake in the code, maybe there's one of the reasons multibots doesn't work well. But a well now I've got to wait onto the fixing, allrigth.
Another question, this time not about bots, but I feel I've got to keep asking questions .
I gues I have'nt logged out yesterday, are you not automatic logged out, I was still logged in when I came here back. The question if it wasn't clear.
Are you here automatic logged out or do you stay in until you log out, how does it exactly work in here.
(I know almost for sure I have broken a record of asking question's in row in here, I hope you don't mind )
-
No probs with asking lots of questions!!
The forum remembers you and your password for you until you log out. If you log out then next time you visit you'll have to enter your name and password again.
-
Allright if it's not a problem here is another. The 'permanent' ties in multibots, are they supposed to be permanent. They are in fact not permanent in my simulation.
Is this code in a modefied bot like the one posted earlier, I am sure it will do desame in the posted bot.
cond
*.numties 1 =
*.multi 1 = and
'0 *43 = and
start
1 .tienum store
628 .fixang store
50 .sharenrg store
'1 43 store
stop
This is a permanent (yellow) tie.
cond
*.numties 1 =
*.multi 1 = and
0 *43 = and
start
1 .tienum store
628 .fixang store
50 .sharenrg store
1 43 store
stop
Here you have a not-permanent (red) tie. (it's one cycle yellow)
-
To be honest, this is a bit out of my ken. Eric primarily was the one who set up the ties after I broke them in 2.40. You might want to try looking at his change logs in the bugs and fixes forum, and searching for fixang. Try checking out this post (http://www.darwinbots.com/Forum/index.php?showtopic=1623).
Also try your bot out in 2.37 and see if there's a difference. That'll give us a clue as to wether it's the bot or the program.
-
Well I found this in
2.42.9c Buddy Drop Download 18)
Bots should still be able to fix the angle of a tie they did not create by explicitly using .fixang.
This is the only sentence I found that was going about fixang. Nothing about becoming a permanent tie or something like it.
It did desame in 2.37.6 so I gues it's normal. I was just guesing that a permanent tie supposed to be permanent. So it is'nt.
-
Nums;
Tried Peters bot in 2.37.6 - 2.37.7b and 2.37.7g but the bot crashed them all.
Triangulus in 3.27.6 and 2.37.g manages to form triangle but fails to in 2.43 (and 2.42.7 the earliest 2.42 .exe I have)
fixang still reads 32000 in 3.27.6 though so that is probably not be the problem.
Are there other tests I can try for you or anything I can do to help narrow down what is causing this? I'm not really a MB person anyway so not sure how to progress further in finding problem.
Peter, what do you mean "I was just guesing that a permanent tie supposed to be permanent. So it is'nt."?
A permanent tie is formed after a set number of turns, after that you shouldn't be able to change its position using .fixang, however you can still use .tieang
(tieang ----- Ties harden after a while. Whatever angle and length that they have at that point becomes permanent. .tiang lets me temporarily bend the angle by the value that I store. It springs back though.) -from ingame dna help
Permanent ties, as far as I know, can still be broken, what they can't be is moved or stretched in the same way. Also it doesn't matter which bot makes the ties, both bots can use them as long as they know the address, it's always been like that.
Or were you thinking of something else?
-
Nums;
Tried Peters bot in 2.37.6 - 2.37.7b and 2.37.7g but the bot crashed them all.
Triangulus in 3.27.6 and 2.37.g manages to form triangle but fails to in 2.43 (and 2.42.7 the earliest 2.42 .exe I have)
fixang still reads 32000 in 3.27.6 though so that is probably not be the problem.
I know that you're speaking to nums but when I tried it has'nt crashed.
Edit: to be sure, I've copyd the code here into a textfile and simulated it in 2.43 and 2.37.6, my advice is put an 'end' at the end, 2.43 doesn't care but 2.37.6 does. Wow they're moving without they are supposed to move.
Peter, what do you mean "I was just guesing that a permanent tie supposed to be permanent. So it is'nt."?
A permanent tie is formed after a set number of turns, after that you shouldn't be able to change its position using .fixang, however you can still use .tieang
(tieang ----- Ties harden after a while. Whatever angle and length that they have at that point becomes permanent. .tiang lets me temporarily bend the angle by the value that I store. It springs back though.) -from ingame dna help
Check one of my earlier posts, I am not using tieang to harden the ties, but it hardens. There is nowhere is the code tieang, you can check at the complete code and you will see.
I gues you're quoting from the wiki (I've seen that somewhere) I wil be quoting this, the full story behind the little sentence quote. tieang is setting the angle of the one that did'nt shot the tie, for the one that shot the tie it springs back.
Upgrade version's from EricL.
18) Prety serious tweaks to ha3A11ed tie physics. Briefly, when a tie hardens, it's length gets set to whetever is is at the time it hardens. The angle of the tie relative to the bot that created the tie gets fixed at that end but the angle at the other end, the end connected to the victim bot, continues to be free to rotate. This makes sense to me in that a bot should be able to control his end of the tie, but not the other end. People may hate this, but it really helsp tie physics I think. Bots should still be able to fix the angle of a tie they did not create by explicitly using .fixang. I also added some slop in the tietorque and tiehooke routines so that small angle deviations and small tie length deviations from the fixed values don't result in forces being translated to the tied bots. This helps get rid of funny harmonics in tightly tied clusters of bots. There are still cases where lightweight bots can vibrate or get whipped around, but I think things are much much better than they were. Also added some limits on the effect of torque values on .aim.
19) Removed .nrg < 0 check in releasebod() so that -6 shots work against corpses.
Maybe I have something more to add, in the code I am using in a multibot.
This is the code, if you didn't now.
cond
*.numties 1 =
*.multi 1 = and
'0 *43 = and
start
1 .tienum store
628 .fixang store
50 .sharenrg store
'1 43 store
stop
Here fixang is executed every cycle, the 'head' part is staying on his place. The 'body' part can constantly move due to using this code. I am not sure if this code is 'overruling' the other or it is a mistake in the program.
la (or 'le' I have never been good at french) code.
*.eye4 *.eye6 sub .aimsx store
-
I was never very much of a MB person either, so take this with a grain of salt. I believe fixang is supposed to work with hard ties (and hard ties only). I think there's another command that sets a "temporary" angle that the ties then try to fix.
Peter, if your bot still seems weird in 2.37 then it's probably your bot. 2.37 was probably the most stable, MB friendly version we have. I guess you should just keep playing with things until you get the angles you want, and then we can work backwards and see what sort of values do what.
-
According to PY's ingame help file;
fixang ----- This one lets me permanently change the angle between the tie and myself. Zero should make me face you while 628 (half a circle) should make me face directly away from you.
tieang ----- Ties harden after a while. Whatever angle and length that they have at that point becomes permanent. .tiang lets me temporarily bend the angle by the value that I store. It springs back though.
-
Well, strangely the tie is temperarely hardening for 1 cycle with that gene I use(without tieang). If the gene activates every cycle it stays hardened. If the gene is activated one cycle the tie is one cycle hardened. It isn't really a problem, but I just found it a little strange that the hardened tie can become a normal tie anytime, maybe I will use tieang in my code, see what will happen. But almost desame is happening 3.37.6 so that probably alright and it will be supposed to act like that.
Is it true that in the beginning that the movement of multibots is being multiplied(or anything else strange). I've had strange things in my multibot I have put a timer into it (first 50 cycles do nothing) and it was just acting perfect.
If I have understuud, the whole tie-system is being trown overboard in 2.5 and a new version is becoming created. If I may ask, how far is version 2.5 already, and when is it possible coming out. Maybe I can help with some things. I have'nt really got much programming experience, in fact all the the programming languege I've got is a few weeks(if it were weeks) playing with the gamemaker-languege. I will in one of the coming days take a look at the code, if I understand it, maybe I can help with some things. I will PM you if I feel I can do something.
-
That sounds strange!
As I understood it ties should harden after a certain amount of time, once a tie is hard, beyond deleting it and remaking, it should stay hard. Thanks for testing that in previous versions of DB btw.
MB's should gain a cost advantage by being MB's, movement cost for example should be reduced. (Something I read earlier in wiki I think).
In version 2.5 (I presume you are talking about the C++ portal) Numsgill (AFAIK) is intending to make quite a few changes, a lot of which are regarding the introduction of proper physics with regards to ties etc. /me looks around for Num's to confirm this!
The whole C++ thing though is a big job and doesn't, as yet, have a timeline. If you feel like helping out in the meantime though perhaps you would like to take a look at the VB code and see if you can spot any of the current bugs that are 'bugging' us!
(Again, Num's can guide you in the right direction here, I do have the VB code somewhere but only used it to help spot some of the bugs when EricL was fixing stuff).
Any help, even if it is just spotting bugs in the first place, is always apprecitated, DB is a group project after all. We can thank Carlo (the original programmer) for that!
-
So, you think it's strange too, I am not the only one. Well I think it is only something strange but it is not bad the multibots does'nt work less becouse of it.
About the cost advantages
I quote from the wiki.
MB Advantages
[edit]
Pre 2.4
The MBs have several advantages over a normal Single Bot(SB). Most of these are in the form of cheaper actions (costs of actions divided by number of other bots the current cell is connected to).
[edit]
Post 2.4
(I believe I have gotten rid of these artificial advantages, but I will have to check. --Numsgil 13:16, 18 Sep 2005 (MST))
-
2.5 is something like 95% done. It's just missing bot debugging tools and multibot physics. But I abandoned it when it became clear that it wasn't going to be easily maintainable. At the moment I'm researching physics so I can be in a position to implement the more complex physical reactions correctly. Also, 2.43 has quite a few features added to it that the C++ source doesn't. On the other hand, the physics are much nicer in the C++ code.
When I start again it'll probably be from scratch using C#. I'll probably start with the physics and work the rest of the program into that. I should be in a position to start sometime in the next few months. If you want to start tinkering around int the VB source, you'll find it very easy. I actually learned VB working on this project.