Welcome To Darwinbots > Newbie
coexistence
Numsgil:
It's still quite a feat! I don't think anyone's managed a simulation that long, at least not to my knowledge. Best of luck on your next simulation.
hdggDalton:
alright this is the setup i got running right now, splitting the map into 4 quarters
nothing too special yet as of 29.474 million cycles, the bots engage in opportunistic cannibalism where they hover around each other but don't try too hard to eat each other, like only shooting a few shots at a time. They're not very active movers, they just stay still when there's no food or other bots around. There's slight differences between the bots in each quarter but it's largely all in junk code/neutral mutations that doesn't do anything but still changes the color
This simulation almost faced a total extinction a few days ago because I made the barriers between the quarters too restrictive, which increased the risk of complete death in one quarter since it was really hard for a bot to repopulate it from another corner. What I think happened is that one corner died out from unfortunate starvation, then followed by another one, then another one, and all the plants in the died out corners contributed to the total energy (The sunlight turns off at a maximum upper limit of total energy to prevent the map from being overrun by plants) which caused even less energy to go into the plants into the corners with bots still alive, causing the other corners to die out quicker.
But a few bots managed to survive in the bottom left corner, when I got back from doing something they were down to a pitiful 30 since they had to rely on scraps from the few plants they could reach since the plants stopped spawning in since the amount of plants was at the maximum limit. Thankfully I kickstarted it back up by deleting those super restrictive barriers and manually moving some bots right next to giant piles of plants in each of the 4 corners. Their dna was still intact and unmolested by genetic drift, so they quickly repopulated and now i have a blooming simulation that's quite alive :D
hdggDalton:
oh my god guys, the bots are changing!!! they're becoming more greedy, with more average energy per bot. since i increased the graphing interval i can see some more long term trends, like a while ago there was heaps of population and diversity but somewhere along the way the population crashed and never recovered from it
hdggDalton:
oh dear it looks like a bot has discovered a way to gain immortality and cheat death
basically it spits out a child every frame but then eats it either on the same frame or in the very next frame, and somehow it has maintained maximum energy+body
--- Code: ---cond
*.eye6 *188 !~=
*.refeye << >=
start
25831 .up store
.eye5width store
806 <
logx *294 *.mysx *.tielen1
'''''''''''''''''''''''' Gene: 1 Ends at position 19 '''''''''''''''''''''''
'''''''''''''''''''''''' Gene: 2 Begins at position 20 '''''''''''''''''''''''
start
*.trefbody *.eyef 207 >=
*.myeye 806 <
logx store
*.hit *.tielen2 -6 .shoot store
15429 *.myeye *.refypos .shareshell store
337 xor
5 ++ *.refvenom sgnstore
substore
and
*.nrg 30511 >=
.repro floorstore
'''''''''''''''''''''''' Gene: 2 Ends at position 52 '''''''''''''''''''''''
--- End code ---
hdggDalton:
Hello everyone happy new year it's been a while but i got some interesting things to show you because ive been running the simulation all this time. First of all i kinda started a new simulation by picking out the best bots from the simulation, then just making it start with those bots, but I also edited their mutation rates because I screwed up by making the mutation magnitude too large. Like, a point mutation is usually only 1, but i edited it to be minimum 1, maximum 5 and it didn't work well because usually for cool new things to happen you need small mutations that over time work with more small mutations to create some new behavior or function. But with the large mutation magnitude, it was basically impossible for the bots to evolve significantly because it would change too many things at once, usually breaking the bot so that it does nothing and dies of starvation. Even if it doesn't kill the bot, the large mutations usually only add or change junk code that does absolutely nothing, and extremely convoluted junk code filled with nonsense too, so that it's extremely hard for additional mutations to end up being positive.
I want to simply describe what most bots do, because their behavior is quite unique. If they don't see anything, they just stay still. It's boring, and it's pretty much what the basic animal minimalis does anyways, but it's practical because the cost of movement is pretty high and especially since there are a lot of walls and no map wraparound, there's no guarantee of finding food by simply moving in a straight line, and in fact doing that can get you stuck in a corner where you're even less likely to find food and very likely to starve to death. If they see a plant, they go straight for it and eat it. If they see a bot, it seems like they try to circle around each other, and sometimes they attack each other with shots, but with not much dedication. They're pretty much opportunistic cannibals. Maybe the circling behavior is a result of bots trying to get behind the other to land free shots? Also they treat a corpse exactly like a plant and go straight for it.
So once I turned down the mutation magnitude, I started seeing some interesting mutations. First of all there was a species in the upper left quadrant which developed shells, which greatly helped them survive from the opportunistic cannibalism of basically everyone else (opportunistic cannibalism is when bots don't necessarily seek out other bots to kill and eat, but they will totally do so if one ends up in their sights) but they eventually died out. I believe this is because they made too much shell so they became too heavy and couldn't move anywhere to find food, so while they could resist cannibalism, they just starved to death. Second there was another species in the upper left quadrant some time after this which somehow stopped being cannibals and basically regained the ability to distinguish between plants and bots so that they won't attack bots. They still circle around each other and get pretty comfy with each other, so they still recognize bots, but won't attack them. I think I've said before that once animal minimalis loses its ability to recognize other bots to not attack them, it's impossible for them to regain that ability through mutations, but i was proven wrong and i'm delighted about it. Sadly i found them yesterday right as i was about to go to bed so i didn't take a picture of them, went to bed, and i forgot to take a picture of them again when i woke up and reopened the simulation and forgot all about them, and by the time i remembered the simulation had ran for a while and they apparently went extinct during that time. :( my guess is that they were killed with impunity by the opportunistic cannibals and they didn't fight back at all. Or it might be bad luck in finding food, as they were a pretty small population of about 10 bots
the first picture is an overall screenshot of the whole simulation, and the second picture is an example of what it looks like when bots have just finished eating up a clump of plants, they do this crazy circling battle royale where they try to eat each other
here is the best bot, with 61 descendants out of a population of 610, impressive!
--- Code: --- cond
*.tieang3 *.myaimdx *.eye8 !=
*.tin4 *.refeye >=
start
*.refage .strbody store
*.refveldx clearbool
29205 413 *.eye2 18838 *.myshoot store
*.refage 2 .memloc store
.tielen4 store
*.refypos dropbool
'''''''''''''''''''''''' Gene: 1 Ends at position 28 '''''''''''''''''''''''
'''''''''''''''''''''''' Gene: 2 Begins at position 29 '''''''''''''''''''''''
start
*.eye5
'''''''''''''''''''''''' Gene: 2 Ends at position 30 '''''''''''''''''''''''
'''''''''''''''''''''''' Gene: 3 Begins at position 31 '''''''''''''''''''''''
else
'''''''''''''''''''''''' Gene: 3 Ends at position 31 '''''''''''''''''''''''
'''''''''''''''''''''''' Gene: 4 Begins at position 32 '''''''''''''''''''''''
start
and
207 >=
*.robage *.hit -6 .shoot store
327 not
5 ~ substore
*293 angle *.veldx *.trefvelmydn *.nrg 26656 >
527 *.refsx .dn store
.repro store
924 27413 *.pleas'''''''''''''''''''''''' Gene: 4 Ends at position 61 '''''''''''''''''''''''
--- End code ---
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version