Darwinbots Forum

Bots and Simulations => Simulation Emporium => Topic started by: Endy on February 13, 2009, 10:30:28 PM

Title: Any other Zerobot Sims?
Post by: Endy on February 13, 2009, 10:30:28 PM
Still letting my zerobots do their thing, the problem is they seem to have plateaued in their evolution. They reproduce, they tie, they shoot and they've learned to modify their shootval based on the body of the bot they're tied to. I'd like to move them forward again though.

Injecting DNA from other bots has worked well in the past(tie ability and reproduction) so I'm hoping someone else will post a sim with novel zerobot DNA I can mix with theirs.

ps

It seems like its been forever since I last posted. Finally finished with the Navy, so things were a bit hectic  
Title: Any other Zerobot Sims?
Post by: Numsgil on February 14, 2009, 06:27:53 AM
Do you have viruses enabled?  If not, I would try the predator idea.  Throw in a few predators for a few hundred cycles to cull the herd a bit.
Title: Any other Zerobot Sims?
Post by: Testlund on February 14, 2009, 09:16:26 AM
I've been trying two times running a new zerobot sim with the latest version, where I've disabled vision, started with 1000 vegs and 1000 bots. Somehow I doubt the touch senses work as good as vision, in that when a bot touches another bot it should be able to do the same thing as if it saw that bot, if the dna has evolved the instructions. So far not a single tie from any bot and very slow reproduction. The bots just sit there doing nothing while the population is slowly declining.
I'm thinking about starting a new sim with the default bots just to compare.
Title: Any other Zerobot Sims?
Post by: Numsgil on February 14, 2009, 10:07:39 AM
Vision was really central to the program in its earlier incarnations.  Without it, there's no guarantee that the simulation is balanced well enough to sustain life.  Still it would be interesting to see if anything develops, so don't let me discourage you
Title: Any other Zerobot Sims?
Post by: Endy on February 14, 2009, 03:37:21 PM
I've got viruses enabled, I've just never actually seen them  

What settings did you use to get them to develop?

Vision doesn't seem strictly necessary, mine have managed fine with getting info from ties, though it took them forever and a day to develop that much. I had to help them with feeder veggies until they were efficient enough on their own, though.

You can normally get some good starting DNA just by hyper mutating a single zerobot. You just set Point mutations to 1 instead of 5000.
Title: Any other Zerobot Sims?
Post by: Numsgil on February 15, 2009, 09:52:53 AM
It shouldn't take that much effort to get a virus going.  The few zerobot sims I've done, plus Testlund's experience with them, seem to indicate that they should just be appearing as one of the first things to develop.  It's weird if you don't see any.  Maybe costs have been changed for viruses relatively recently.
Title: Any other Zerobot Sims?
Post by: Testlund on February 15, 2009, 12:30:36 PM
Quote from: Endy
Vision doesn't seem strictly necessary, mine have managed fine with getting info from ties, though it took them forever and a day to develop that much. I had to help them with feeder veggies until they were efficient enough on their own, though.

You can normally get some good starting DNA just by hyper mutating a single zerobot. You just set Point mutations to 1 instead of 5000.

Good to hear. I guess I'll just have to be patient then. I started with point mutations set to 32, which is quite low.
Title: Any other Zerobot Sims?
Post by: Endy on February 15, 2009, 12:55:49 PM
The basic stuff is easy to evolve, the mutation settings seem to limit higher numbers from entering the genome. Even with point mutations set to 1 and a range of 1-999 it takes forever. I took another look, and the bot mine got it's repro ability was a "random bot" not a strict zerobot like mine was. Still going to keep it, since it has so much time invested into it now.

Went looking and found some of the other zerobot sims, in the evolution and internet sharing sims. Going to see if I can't make a more interesting hybrid with some of their DNA.
Title: Any other Zerobot Sims?
Post by: Endy on February 19, 2009, 10:20:38 AM
They started pulling forwards again  

They're up to 0.85 of standard costs. Might just be able to enter them in as an F1 bot soon
Title: Any other Zerobot Sims?
Post by: Numsgil on February 19, 2009, 10:59:01 AM
That would be pretty cool.
Title: Any other Zerobot Sims?
Post by: Moonfisher on March 02, 2009, 04:09:32 PM
I'm working on a zero bot, but so far it wouldn't be able to survive in your sim, and it only rarely sends out any viruses.
So far it's at 25% F1 conditions with 50 alge. I'll post it when it get's closer to regular F1 settings, currently having problems with big titans killing half the population because waste isn't working properly and the field is so small. That and alge that get onfected with a large chunk of dna, making it shoot and produce poison and shell.
The alge can't reproduce because of the veggy repop settings, since F1 settings allow veggies to get too strong, with the high energy gain and low cost of poison and shell. But I haven't managed to do anything about viruses. I'm thinking of using a different veggy, one that keeps it's code clean from viruses and maybe kills off bots with too much waste. Just wanted to use a zero bot as veggy to make sure some virus didn't accidentaly spread handcrafted code into the zero bot. I guess if the alges virus defence is strong enough it shouldn't be a problem, but you never know...
I don't think increasing the sim size would help much since waste doesn't kill off the big non reproducing bots. And they're preventing the zbot from achieving a larger size, I see it trying, but then some titans come along and kill off any bot large enough to get hit...
And the bot needs to get larger for me to increase costs any further... been killing off the big bots manualy so far, but getting real tired of it.
Title: Any other Zerobot Sims?
Post by: Endy on March 02, 2009, 10:09:22 PM
Zerobot veggies serving as the main food supply is an interesting idea. I wound up turning off veggie mutations and preventing them from being infected, to try and avoid the problem of artificial code entering their dna.

You can just enable age costs to weed out the non-reproductive big bots. Just set it to change to 32000 at some age.

Evolution *should* be able to handle it by itself. Mine had a mass based firing condition for awhile. It eventually went away after the mini's were able to swarm the big bots.
Title: Any other Zerobot Sims?
Post by: Endy on March 06, 2009, 12:20:25 PM
Victory  

They've finally made it to F1 costs!

[attachment=1173:NewStart.zip]

I was noticing last night that their population had been repeatedly cycling between 200 and 400, apparently independent of costs, so I decided to chance it and ramped costs up from 0.85 to 1. The have managed to survive overnight and I think it's safe to admit them into F1.

At some point I want to pick apart their dna and try to figure out exactly how parts of it work.  Some of it is obvious, but most is pretty esoteric.

Code: [Select]
''''''''''''''''''''''''  Gene:  1 Begins at position  1  '''''''''''''''''''''''
 cond
 dropbool
 -24 -749 -49 *0 inc
 12 0 dup *.trefbody mod false
 else
 .shoot inc

''''''''''''''''''''''''  Gene:  1 Ends at position  16  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  2 Begins at position  17  '''''''''''''''''''''''
 start

''''''''''''''''''''''''  Gene:  2 Ends at position  17  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  3 Begins at position  18  '''''''''''''''''''''''
 start
 7 true
 236 298 282 317 mult dropbool
 240 383 *370 angle xor
 <=
 400 rnd & add store
 *.aimright sub -38 dec
 %=
 -44 inc
 ~ 48
''''''''''''''''''''''''  Gene:  3 Ends at position  46  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  4 Begins at position  47  '''''''''''''''''''''''
 start
 dec
 1282 angle %=
 *-29 -- =
 9 -36
''''''''''''''''''''''''  Gene:  4 Ends at position  56  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  5 Begins at position  57  '''''''''''''''''''''''
 start
 =
 >=
 and
 <
 dec
 inc
 2 .sx store
 pow *.focuseye ++ sub stop
''''''''''''''''''''''''  Gene:  5 Ends at position  71  '''''''''''''''''''''''
 .robage store

''''''''''''''''''''''''  Gene:  6 Begins at position  74  '''''''''''''''''''''''
 else
 | <
 -45 inc
 store
 * .timer store
 & >

''''''''''''''''''''''''  Gene:  6 Ends at position  84  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  7 Begins at position  85  '''''''''''''''''''''''
 start
 .shoot dec
 & sqr not
 ~ or

''''''''''''''''''''''''  Gene:  7 Ends at position  92  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  8 Begins at position  93  '''''''''''''''''''''''
 cond
 inc
 floor 2 -1 ^
''''''''''''''''''''''''  Gene:  8 Ends at position  98  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  9 Begins at position  99  '''''''''''''''''''''''
 cond

''''''''''''''''''''''''  Gene:  9 Ends at position  99  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  10 Begins at position  100  '''''''''''''''''''''''
 cond
 sqr ++ -219 overbool
 start
 store
 108 181
''''''''''''''''''''''''  Gene:  10 Ends at position  108  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  11 Begins at position  109  '''''''''''''''''''''''
 cond
 dec
 false
 *-222 mult 1067 overbool
 *.aimleft else
 * dropbool
 7
''''''''''''''''''''''''  Gene:  11 Ends at position  120  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  12 Begins at position  121  '''''''''''''''''''''''
 start
 dec
 *.dn and
 << 59 * -18 16 clear mod <=
 and
 <<''''''''''''''''''''''''  Gene:  12 Ends at position  134  '''''''''''''''''''''''
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 06, 2009, 12:41:08 PM
That is very impressive.
The fact that the DNA looks so organized is perplexing, I was kind of expecting it to be something more confusing yet still equivalent in effect to what a hand-coded bot might do.
I will be looking forward to any more progress you make.
Title: Any other Zerobot Sims?
Post by: Numsgil on March 06, 2009, 07:35:04 PM
It looks organized because of the DNA deparsing code.  Everytime the code detects a cond or a start it provides some comments about a gene.  And everytime it finds a conditional or store/inc/dec, it starts a new line.  Makes even the most convoluted DNA somewhat easier to understand.

Endy, have you tried running it in F1 yet?  Can it get past the first bot yet?
Title: Any other Zerobot Sims?
Post by: Endy on March 06, 2009, 10:00:38 PM
I'm thinking some of the DNA compacting is being done to cut down on the odds of deadly mutations occurring, "2 .sx store" is harder to effect than something like "1 ++ .sx or store".

The One trounced it  

It gets eaten alive by tie feeding and sharefeeding. After several fights it can pull off a win against C_Ancestralis.  

I'm going to try and either evolve it to sharefeed or evolve a strict sharefeeder zerobot and combine their DNA. They're still having trouble finding food, the few times C_Ancestralis won were when the zerobot didn't find food. Not really sure how to get them to evolve to some of these things.

Title: Any other Zerobot Sims?
Post by: Prsn828 on March 07, 2009, 09:30:18 AM
Quote from: Endy
I'm thinking some of the DNA compacting is being done to cut down on the odds of deadly mutations occurring, "2 .sx store" is harder to effect than something like "1 ++ .sx or store".

The One trounced it  

It gets eaten alive by tie feeding and sharefeeding. After several fights it can pull off a win against C_Ancestralis.  

I'm going to try and either evolve it to sharefeed or evolve a strict sharefeeder zerobot and combine their DNA. They're still having trouble finding food, the few times C_Ancestralis won were when the zerobot didn't find food. Not really sure how to get them to evolve to some of these things.

I have noticed that in my zerobot evolution sim there is a common theme going on where the bots that begin to tie to the veggies (even if they don't feed) are the ones that are most successful.  My sim has monster veggies though...  Anyway, what I am getting at is that the problem with tie-feeding slaughters can probably be fixed by evolving your bots with large veggies that can do nothing but be moved by other bots.  On the other hand, this could also weaken your bots in a lot of other areas, as it isn't the most challenging of environments unless the bots start fighting amongst themselves.
Title: Any other Zerobot Sims?
Post by: Endy on March 07, 2009, 11:25:23 AM
Yeah, learning to tie does seem to be a common trait. I'm guessing your veggies feed the other bots and so they're attempting to remain close in order to maximize the chance of being fed.

My current thinking is that their ability to shoot is undermining their ability to learn any other feeding methods.

I'm planning on either using a shotgun approach using hypermutation or wiping out their shooting ability while raising the costs on shot formation to encourage them to evolve tie or share feeding. Once I get the desired DNA I can then induce viruses to splice them together so they have both abilities.
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 07, 2009, 11:48:12 AM
Quote from: Endy
Yeah, learning to tie does seem to be a common trait. I'm guessing your veggies feed the other bots and so they're attempting to remain close in order to maximize the chance of being fed.

My current thinking is that their ability to shoot is undermining their ability to learn any other feeding methods.

I'm planning on either using a shotgun approach using hypermutation or wiping out their shooting ability while raising the costs on shot formation to encourage them to evolve tie or share feeding. Once I get the desired DNA I can then induce viruses to splice them together so they have both abilities.

Actually, my veggies kinda just sit there.  I am making the zerobots do all the work
Right now they have decided to spin and shoot constantly, and when they get enough energy they reproduce, creating a temporary tie to the veggy and flinging it across the simulation to be fed on by the other bots.  Now that I think about it, this is actually a good example of unconscious teamwork.
Title: Any other Zerobot Sims?
Post by: Endy on March 07, 2009, 01:09:18 PM
It is neat how they evolve to work together to overcome problems. Kind of like a primitive colony

Could you upload the sim? Curious now what it looks like.

I think I'm going to start using Insertion mutations instead of Point to introduce more novelty. Point mutations tend to be in the same number range as before, so in a zerobot it tends to have mostly lower range memlocs. Insertion seems to rove over the whole spectrum so should be more useful. It kills the sim when it's set at one, so I'm going to try a scaled back run in hopes they make use of the more advanced dna.
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 07, 2009, 01:29:23 PM
Quote from: Endy
It is neat how they evolve to work together to overcome problems. Kind of like a primitive colony

Could you upload the sim? Curious now what it looks like.

I think I'm going to start using Insertion mutations instead of Point to introduce more novelty. Point mutations tend to be in the same number range as before, so in a zerobot it tends to have mostly lower range memlocs. Insertion seems to rove over the whole spectrum so should be more useful. It kills the sim when it's set at one, so I'm going to try a scaled back run in hopes they make use of the more advanced dna.

Sure, here is one of the recent saves.
To be honest, I didn't even change any of the specific mutation settings, just the overall mutations depending on how things seemed to be going (When they begin to make progress I scale down mutations to avoid Muller's Ratchet, but if the diversity crashes I crank it up really high.)

Every once in a while I find I need to change things around.  Sometimes I change the number of Veggies, other times I just tweak the costs or mutations. I also had to hand-feed the bots a few times, because at first they were just sitting still fire at nothing. Now they move so it isn't a real issue anymore.
Title: Any other Zerobot Sims?
Post by: Endy on March 07, 2009, 03:16:50 PM
Quote
Every once in a while I find I need to change things around. Sometimes I change the number of Veggies, other times I just tweak the costs or mutations. I also had to hand-feed the bots a few times, because at first they were just sitting still fire at nothing. Now they move so it isn't a real issue anymore.

I've always seen it as slowly increasing the challenges they face. I imagine RL evolution had much the same thing occurring, life having progressive challenges it evolved to overcome.

It's odd how they're making poison.  Could just be random or they are trying to defend themselves.
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 07, 2009, 11:03:29 PM
Quote from: Endy
Quote
Every once in a while I find I need to change things around. Sometimes I change the number of Veggies, other times I just tweak the costs or mutations. I also had to hand-feed the bots a few times, because at first they were just sitting still fire at nothing. Now they move so it isn't a real issue anymore.

I've always seen it as slowly increasing the challenges they face. I imagine RL evolution had much the same thing occurring, life having progressive challenges it evolved to overcome.

It's odd how they're making poison.  Could just be random or they are trying to defend themselves.

Now that you mention it, you are right. I never really thought about that, but they do make poison, and I have had that and shell be created in several separate zerobot sims. It actually seems to happen a lot. I don't know about how well increasing challenges applies to real life though. It seems more like evolution has a tendency to fight itself, always trying to improve the predator, then the prey, and then the predator again. If you think in that way, we are the predator, setting the costs for our prey, the bots. Only difference is that we don't evolve and they do... now that is kind of creepy
Title: Any other Zerobot Sims?
Post by: Endy on March 08, 2009, 05:11:48 PM
Well, we're still evolving, just much slower than the bots

I think the main stresses for us are STD's and the various population migrations going on. From a species view evolution is attempting to realize every possible combination that can survive. The more that we learn to cure/fix the more genetically diverse our population should become(assuming we don't change the genome itself).
Title: Any other Zerobot Sims?
Post by: Numsgil on March 08, 2009, 09:30:30 PM
I definitely think in the next 60 years (probably much sooner), we'll start tinkering with our DNA in earnest.  It's actually one of the non-AI ways to reach the technological singularity (http://en.wikipedia.org/wiki/Technological_singularity), with successively smarter generations of humans understanding how to modify their genome even more.
Title: Any other Zerobot Sims?
Post by: Moonfisher on March 13, 2009, 05:14:50 PM
The bots in my sim are now at regular F1 costs, but currently with 80 alge in the sim. (Going to cut down slowly to end up like regular F1 settings)
Hadn't even noticed there was a checkbox for disabling viruses in a bot species, but the waste managing part helped keep big berthas to a manageable amount and allowed the bots to evolve to a larger size.
I only run the sim in the evening when I'm home, so it'll probably be a while yet before I start working on getting it to actualy beat F1 bots

The bots mostly move in the same direction since this reduces the amount of conflicts, and makes big berthas less likely to kill off it's own kind.
At a certain point in their life they start locking themselves in place and sometimes tie to the alge. This helps keep the alge in place for everyone to feed on them, and the behavior has been around since the earliest stages of the sim. Whenever the population dropped too far leaving few older bots to hold the alge in place, the younger bots would start to die out, so it seems like the bots aren't able to track the alge properly and depend on the older bots for feeding.
The older bots don't seem to gain much from being locked since they're facing the wrong way their shots don't hit the alge they're blocking unless they tie to an alge and draw it in, but this seems to be random they're just as likely to push it away.
There was a subspecies wich lasted for about 400K cycles, wich would lock itself very early and spind and shoot, this way by grouping together they could herd the alge as the other species moved them into the locked clusters of bots. This seemed to work well enough for both species, but only for a while, the clusters moved around slowly over generations and had problems when hitting the corners of the toroidial map.
But they did coexist for a long time, saved the sim, fogured maybe I could tweek it and increase sim size and stuff like that to make them brach out even more.

And concerning the more recent topic, I think the current "evolution" of humans points towards groups that are more likely to become sperm donors and groups that have a lot of children for religious, cultural or biological reasons, but probably mostly the 2 first mentioned. So my guess would be that our "sim" has reached a point where almost anything will survive and the biodiversity within our specie is increasing dramaticaly. And I don't think a higher intelect would be the mutation to trigger the next evolutional step, and the "correct" mutation would also depend on the part of the world you're in.
And even if people could pick the traits of their children I'm not sure intelect would be the most prominent direction, intelectuals tend to have fewer children. I'm sure they'd make some realy smart people if they can, and even if they don't have children their knowledge will help create even smarter shildren, leaving a different kind of heritage in the genome. But I'm not sure this would become the "comon" genome, I think the majority may prefer to stay as they are, knowledge does not equal happyness or a large amount of decendants, so not everyone will persue it and those who do are no more likely to become the majority over generations.
I think it's hard to say where we go from here evolutionwise... medical science is only just starting to get serious so I'm not sure the imune system will play a huge role for much longer, not sure intelect is a driving factor and it's not entirely bound one genome, a large part comes from your culture and parents. Faster reproduction in some way could mean a lot in western countries... maybe... but in an overpopulated place someone why requires less food may be more liekly to get by. I think in the end the one who adapts the fastest would be the safest bet, our world changes faster than ever before and whoever gets with the program the fastest gets by. So more powerfull epigenetic memmory... that would be my guess.
Title: Any other Zerobot Sims?
Post by: Endy on March 13, 2009, 10:47:49 PM
Two zero bots at F1 settings, I see a fight coming on

It is neat how the different sims evolve. You'd think we'd see more similar results, but each one winds up different.
Title: Any other Zerobot Sims?
Post by: Endy on March 15, 2009, 09:57:37 PM
A higher number of insertion mutations seems to be working for increasing the amount of novel DNA. Might be too high at 1000, but they seem to be evolving better overall.

My current computer keeps threatening to meltdown when running sims. I'm going to try and get a new screen to use with a laptop, so I can start up again.
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 20, 2009, 11:34:01 AM
I've started running two ZB sims on my machines. One is running a standard 'let-things-work-themselves-out' sim with 500 bots and a few hundred static veggies. The other one I have to toy with because it runs on an older computer. That one started out with a single 87 gene zerobot that I hyper-mutated with over 1000 mutations, none of which did anything at all.

Right now the big sim is still at 499 bots (one died early on from costs, methinks.) and as I typed this a second bot developed a shooting gene, this one can actually feed! The first shooter used -1 shots that did jack nothing. I'm quite pleased, as this is the first gene that does something besides slowly move around in this sim! It's currently at 17 hours and a hair over 2 million, 57 thousand cycles. ( I like my 3.2GHz CPU. Great Christmas gift.) All it needs is to get a movement gene and develop reproduction, and then this is going to get really interesting.

The (current, mind you. these bots were developed before this sim was run) second sim hasn't done much the past 10 hours. A bot developed one-way movement, so I copied the DNA code and made around 50 of them. Not much has changed since the sim ran, except more bots have become stationary. What their DNA is having them do is move to the right on this relatively small toroidal map. While not much is turning out per hour (it was just over 166 thousand cycles 5 minutes ago) it makes a better erosion sim than an evo sim, because they seem to have eroded the veggies and static bots into a fairly straight and small formation. That gave me a little laugh. Most developments in this sim that have occurred simply changed the orientation of the bot a hair, and that might still only be impacts. I don't expect to see anything new for a while there simply due to processor speed.


This sim is running using standard F1 costs set to a 0X multiplier so they don't die right now. Once reproduction evolves and shooting comes with it, I'll be able to ratchet the costs up and increase the veggies. At the moment they do nothing, except one bot out of the 499 remaining is capable of feeding off of them. It simply cant move or reproduce.





Also, I'm afraid to end the sim and load a save because I don't want to go through the "______'s DNA parsing is incorrect. Ignoring parameters." error 500 times. Any way to fix this?
Title: Any other Zerobot Sims?
Post by: Endy on March 20, 2009, 01:14:50 PM
Just copy the dna from the robdna menu into a blank text file.  

Ctrl-C still works, even though it's not really obvious.
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 20, 2009, 02:28:17 PM
Quote from: Endy
Just copy the dna from the robdna menu into a blank text file.  

Ctrl-C still works, even though it's not really obvious.
No, I mean when I load a saved sim. There's no way to do that with 499 bots. (I think it's down to 497 now. Poor wandering bastards.)

Unrelated; The 500 bot sim now shows 41 shots per cycle. I know of three of the bots that shoot, but now there's a fourth one. I just can't find the little bugger, and it's driving me mad! Now again at 2.56 million cycles, there are now five shooters. Found the new one, but number four continues to elude me.
Title: Any other Zerobot Sims?
Post by: Endy on March 20, 2009, 09:03:50 PM
That's odd, it shouldn't be doing that from a saved sim. You might want to try restarting with just the shooter strains.
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 21, 2009, 04:55:39 PM
Well, I'll have to load a save now, my brother closed the sim to play Fallout 3. Thing is, I wanted to wait until one of the bots developed reproduction, then splice all three of the genes into a bot with a long genome to leave room for new functions. So now I have to find a way to get rid of that damn error, or I lose over 3 million cycles and two days of evolution.

EDIT; Sweet, saved sims work. Though still no repro. How long does reproduction usually take to develop, in cycles? Once it appears, I want to turn costs on to weed out all the crap. Also, how long before any kind of eye function can be expected?
Title: Any other Zerobot Sims?
Post by: Testlund on March 21, 2009, 05:26:55 PM
Quote from: Ta-183
How long does reproduction usually take to develop, in cycles?

I think somewhere within 10 M cycles you would have a stable reproduction. I've had some bad luck with my own sims after several tries where the population just slowly decline. Any bot that tries to evolve a useful function dies. It wasn't like that before. I will NOT give up on using costs! It SHOULD work despite costs!  
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 21, 2009, 08:18:58 PM
10 million cycles, huh? Lets see.... 43 hours and I'm at a hair over 4 million cycles. I like this computer. Shouldn't be too long then, especially with 488 bots. (Yes, more died from shots. I find it comical how half of the shooters are using info shots.)

EDIT; I'm going another route on my second PC. Due to an outdated CPU, standard zerobot evolution methods are unfeasible. I plan on using existing, basic zerobot genes and splicing them together with additional random genes to facilitate faster evolution. Does anyone have a first generation zerobot reproduction gene? I have movement genes and shooting genes from my own zerobots, so I need additional zerobot DNA to keep it inline with the idea of evolving a pure zerobot.

EDIT 3/22: Great, shots are up to 108 per cycle and I've lost count of the shooters. I'm loosing bots at a rate of 10 bots every million cycles. Hopefully reproduction will save the day. Then I'll be able to end the sim and splice some genes.
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 22, 2009, 01:32:11 PM
Success! 5.4 million cycles and the counter reads two births. I located the parent of one and obtained it's DNA. It appears to be fine, the child bot is still alive and I have seen this bot fire small numbers of shots. Can anyone decipher this?

Code: [Select]
stop
 swap 27 0 38 -20 -26 24 add 0 -3 0 -11
''''''''''''''''''''''''  Gene:  1 Begins at position  14  '''''''''''''''''''''''
 cond
 -20 0 -17
''''''''''''''''''''''''  Gene:  1 Ends at position  17  '''''''''''''''''''''''

''''''''''''''''''''''''  Gene:  2 Begins at position  18  '''''''''''''''''''''''
 cond
 -1 -4 9 0 -23 4 9 11 -4 0 12 -7 -11 0 *.up -15 clearbool
 -4 start
 13 -4 0 8 -5 stop
''''''''''''''''''''''''  Gene:  2 Ends at position  43  '''''''''''''''''''''''
 ~ stop
 -2 -2 0 0 *.timer 20 -6 38 0 *.robage -5 -13 16 0 -7 0 ~ pow -3 0 0 *13 0 0 0 0 -18 floor 0 -15 false
 0 <
 35 -6 -29 3 0 0 -3
''''''''''''''''''''''''  Gene:  3 Begins at position  86  '''''''''''''''''''''''
 start

''''''''''''''''''''''''  Gene:  3 Ends at position  86  '''''''''''''''''''''''
''''''''''''''''''''''''  Gene:  4 Begins at position  87  '''''''''''''''''''''''
 start
 dist inc
 8
''''''''''''''''''''''''  Gene:  4 Ends at position  90  '''''''''''''''''''''''

''''''''''''''''''''''''  Gene:  5 Begins at position  91  '''''''''''''''''''''''
 cond
 -15 -11 0 27 add 0 -11 -25 -22 0 0 5
''''''''''''''''''''''''  Gene:  5 Ends at position  103  '''''''''''''''''''''''

''''''''''''''''''''''''  Gene:  6 Begins at position  104  '''''''''''''''''''''''
 cond
 | 0 0 0 15 13
''''''''''''''''''''''''  Gene:  6 Ends at position  110  '''''''''''''''''''''''

''''''''''''''''''''''''  Gene:  7 Begins at position  111  '''''''''''''''''''''''
 cond
 -5 !=
 5 and
 3 -22 0 0 1 0 11 overbool
 !~=
 -3 7 14 -3 else
''''''''''''''''''''''''  Gene:  7 Ends at position  129  '''''''''''''''''''''''


EDIT: HOLY CRAP IT SPAWNED MORE AND ONE TIED TO A VEGGIE! I am so please with the progress, I am toning down the mutation rates and will be activating costs soon.
Title: Any other Zerobot Sims?
Post by: Numsgil on March 22, 2009, 04:33:46 PM
I believe it boils down to:

8 -5 dist inc

Basically it stores into an arbitrary location in memory based on how far it is from (8, -5) (Assuming I'm remembering how dist works).
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 22, 2009, 04:41:25 PM
It also doesn't seem to be as much of a success as I thought. I'm still running the main sim but I'm also evolving this bot on the other computer. I haven't spliced and DNA yet because I can't determine what triggers reproduction right now. It's a bit odd, but if I shake one it might spawn two bots. Or not. It's sketchy, but with a bit more mutating it might become usable.
Title: Any other Zerobot Sims?
Post by: Testlund on March 22, 2009, 07:30:56 PM
What you would want to look for is behavior that exists through several generations. Otherwise it usually disappears after awhile. You need a lot of offspring to prevent good genes from getting mutated.
Title: Any other Zerobot Sims?
Post by: Numsgil on March 22, 2009, 08:12:35 PM
Think of it this way: once a bot reproduces, you don't necessarily have a reproducing bot strain.  It was a single isolated success.  However this strain of bots is more likely than other strains to reproduce again in the future.  For something like this the bot has to be at just the right position to reproduce.  That's highly unlikely (but better than impossible).  But maybe it'll mutate to do 8 -5 dist rnd inc.  Adding in that rnd command makes it fuzzier, so it doesn't have to be at the exact right spot.  Such a bot would be more likely to reproduce.  Keep it up and you'll get a bot which can reproduce quite regularly.

It's just a slow, drawn out process, with teeny tiny steps.  As it should be   Welcome to the world of Darwinian selection lol
Title: Any other Zerobot Sims?
Post by: Endy on March 23, 2009, 10:21:41 PM
Ran a test on it and it will repeatedly reproduce if you turn Brownian motion on. Just have to make sure and ween them off it once they evolve a little bit further.
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 23, 2009, 10:25:11 PM
Thanks endy.
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 24, 2009, 09:56:39 AM
Just wanted to point out that there is no reproduction in the DNA you posted earlier.
As often seems to be the case, zerobots like to find weird ways to reproduce, and I have never seen a .repro in the DNA of a zerobot...
Any ideas on why?
Title: Any other Zerobot Sims?
Post by: Ta-183 on March 24, 2009, 12:30:27 PM
Actually, yes. Even with my severely limited knowledge of DNA code, I can tell how they reproduce. In DNA, every command has a corresponding number. For example, .shoot is 7 if I recall correctly. I'm probably wrong, but I think I remeber .repro being 50.
Title: Any other Zerobot Sims?
Post by: Testlund on March 24, 2009, 03:37:24 PM
As far as I've seen in my own sims the most common command present in the genes of a reproducing bot is inc. According to the bot help file this command only increments a stored value in memory. Eventually that value will pass the reproduction value and the bot reproduces.
The reason why .repro is so rare and inc is so common as a reproduction method I think is because DB isn't totally random. It is somewhat seeded where you have a set pattern how the mutations will occur. There was a discussion here where we talked about making DB more random. I hope this can be implemented in DB3.
Title: Any other Zerobot Sims?
Post by: Prsn828 on March 24, 2009, 03:47:16 PM
Quote from: Testlund
As far as I've seen in my own sims the most common command present in the genes of a reproducing bot is inc. According to the bot help file this command only increments a stored value in memory. Eventually that value will pass the reproduction value and the bot reproduces.
The reason why .repro is so rare and inc is so common as a reproduction method I think is because DB isn't totally random. It is somewhat seeded where you have a set pattern how the mutations will occur. There was a discussion here where we talked about making DB more random. I hope this can be implemented in DB3.

Actually, DB3 is almost definitely going to be more random.
Since I wrote the mutation code myself, I should know

In the mutation code I allowed for four input values that will determine how likely each kind of command/value is to be the result for the mutation.
In theory it should be one of the options in the mutation settings.

Also, I gave no preference at all to any one command over another, so .repro is just as likely as .setdx
Title: Any other Zerobot Sims?
Post by: Endy on March 25, 2009, 09:08:59 AM
There is something weird I've noticed since EricL's changes. Point mutations tend to be in the lower ranges by default, ie. 1 becomes 2, or inc becomes 1. Unless you fiddle with the ranges it gives by default pretty poor quality mutations for just point.

Insertion works better, but requires reproduction before it can be used.

Repro - I think they tend to use dist/angle/rnd since they don't risk a mutation causing reproduction continually occur or not at all. It's also likely easier to store/inc a random range than evolve the exact value of 300.

Once they have, "semi-randomly reproduce" it's just a matter of juggling the numbers that determine the rate and energy to give offspring.
Title: Any other Zerobot Sims?
Post by: Testlund on March 25, 2009, 03:02:02 PM
Quote from: Prsn828
Actually, DB3 is almost definitely going to be more random.
Since I wrote the mutation code myself, I should know

Kewl!