Author Topic: 100 day sim  (Read 4741 times)

Offline EricL

  • Administrator
  • Bot God
  • *****
  • Posts: 2266
    • View Profile
100 day sim
« on: August 20, 2007, 09:53:03 PM »
This is a continuation of the sim described here.

I'll do some more analysis shortly, but a few thigns to note:

Here is the DNA of the "Best bot":

 start
 dist angle and
 -25 *.up dup ceil *.up pow .dn store
 52 mult <
 store
 store
 add pyth

Note that start command.  I didn't put it there, it evoled.  There used to be garbage ahead of it.  Over time, selection favorred a genome with the start statement earlier and earlier until we arrived at the opimal.  

Average DNA length has dropped.  The sim started with zerobots of length 30.  Now the average DNA length is below 20.  I talk about why this is the case here.
« Last Edit: August 20, 2007, 10:28:25 PM by EricL »
Many beers....

Offline googlyeyesultra

  • Bot Destroyer
  • ***
  • Posts: 109
    • View Profile
100 day sim
« Reply #1 on: August 20, 2007, 10:10:13 PM »
I'm eagerly awaiting the link. After all, since you don't have ridiculous numbers of deletions or any dna costs, all I can think of is more dna increases the likelyhood of deletrious mutations once you've managed to adapt to your situation.

Offline EricL

  • Administrator
  • Bot God
  • *****
  • Posts: 2266
    • View Profile
100 day sim
« Reply #2 on: August 20, 2007, 10:30:49 PM »
Found it (finially!).  Yup, you guessed it.  Here's the relevant text...

I never use DNA costs in my evo sims as I do not want any cost-based selective pressure towards smaller DNA. Maca is exactly correct - such costs will serve to shrink DNA length over time all else being equal.

I should point out that even without any DNA costs, selection pressures still operate indirectly on DNA length. A longer genome will mutate more often (mutations are proprotional to genome length). Since most mutations are deleterious, having a longer genome means your functionality is more likely to degrade due to mutations. Thus, selection tends to place slight downward pressure on genome length even without costs, all else being equal. Of course, this gets balanced out by useful adaptations which by definition must utilize DNA. So, if there is selection pressure to aquire new traits in your sim, the genome will grow in length all else being equal. If not, it will shrink, to the point where the code for critical adaptations starts to be impacted. Ignoring viruses, the average genome length in a sim without DNA costs will be the balance point of these two opposing forces.
« Last Edit: August 20, 2007, 10:32:20 PM by EricL »
Many beers....

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
100 day sim
« Reply #3 on: August 20, 2007, 10:48:24 PM »
Interesting, but I think the bot is at a dead end.  The DNA basically reduces to dist angle -1300 store.  Unless I miss my mark, it's basically doing dist angle .repro store

You're not seeing them move or shoot anymore, are you?  Unless the "best bot" is actually a genetic dead end and there are other species floating around, it looks like your bots are less sophisticated than they were the last time you showed us your sim.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
100 day sim
« Reply #4 on: August 20, 2007, 10:56:56 PM »
Let me amend.  It also looks like it's doing 0 0 pow .dn store.  So it should be moving a little bit each cycle.

Offline googlyeyesultra

  • Bot Destroyer
  • ***
  • Posts: 109
    • View Profile
100 day sim
« Reply #5 on: August 20, 2007, 10:58:44 PM »
Speaking of viruses, it would be nice if we could have an option somewhere to disable viruses (mostly for f2 or for evo sims, if we want to keep dna lengths under control, or if you don't want to disable them, some kind of adjustable cost associated with either .mkvirus or .vshoot).

Also, I've been running a new, randombot evo sim of my own. Quite interesting. Instead of learning to move forwards, they originally moved sideways and later changed to moving backwards. They then started storing values in backshoot, allowing their -1 feeding shots to be effective for feeding. Essentially, they're operating in reverse.
« Last Edit: August 20, 2007, 11:03:11 PM by googlyeyesultra »

Offline EricL

  • Administrator
  • Bot God
  • *****
  • Posts: 2266
    • View Profile
100 day sim
« Reply #6 on: August 20, 2007, 11:01:49 PM »
Nope, no more shooting or moving.  So sad...   I think I'll make the argument that the shooting and moving of the past was not so much selected for but rather not selected against.  What I think has happened more recently is that selection has favored fast reproduction over everything else.  What happens if you watch the sim for awhile is that there are bursts of reproduction when the bot density around a feeding veggy gets high enough.  They use reproduction as a means of movement and dispersal - the force of the birth tie.  Reproduction is kept in check by the body limitation (bots must have body > 2 for reproduction to be acheived).  So, what you see is a bunch of tiny bots with body at or below 2.  When they drift near a veggy and get a few nrg shots, their body increases a little and they reproduce.  I've evolved pond scum.

I should note that virus infection of my veggies were a constant problem in this sim until I implemented virus immunity (coming in 2.43b).  Now, the veggies are immune and the virus seems to have died out (no long-lived veggies to spread it!)
Many beers....

Offline googlyeyesultra

  • Bot Destroyer
  • ***
  • Posts: 109
    • View Profile
100 day sim
« Reply #7 on: August 20, 2007, 11:14:13 PM »
Once I got ones that could feed on their own, I removed "easy energy", set up dna-disabled veggies. I also set them up so that they wouldn't be fed more energy, and the vegies would be fed off of, die, and be repopulated. Even with 1 veggie with 30000 energy spawned per cycle, the bots are able to keep them around to 90-100. I'm using a small, dense field, so if you stop shooting, you stop gaining energy, and you get yourself shot to bits.

I get the whole rapid-repro thing, too. It's just they need shooting and moving just to get that 1-2 body.
« Last Edit: August 20, 2007, 11:16:53 PM by googlyeyesultra »

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
100 day sim
« Reply #8 on: August 20, 2007, 11:17:57 PM »
Yep, you've got yourself some pond scum.  Reminds me of the billions of years with just protists.  I think you might want to change things up a bit.  Thousands of generations and all they can do is reproduce...

Very disappointing

Offline googlyeyesultra

  • Bot Destroyer
  • ***
  • Posts: 109
    • View Profile
100 day sim
« Reply #9 on: August 21, 2007, 10:57:26 AM »
I think some of mine are using the birth tie to protect themselves; essentially, they try to reproduce every turn, but can't if they don't have two body or if they already have something right in front of them. So, by keeping your child, and your child's child right next to you in a chain for a while, you can build up around 10 body, and actually stockpile enough resources to survive a volley of stray shots. Also, that allows the bot to have somewhat more powerful shots, yet still be able to reproduce once the birthtie is gone. Apparently, evolution favors NOT severing the birth tie. And, of course, they don't shoot their children since they move backwards and fire backwards, yet reproduce forwards.

Also, I use fixed bot radii.
« Last Edit: August 21, 2007, 11:04:45 AM by googlyeyesultra »

Offline Peter

  • Bot God
  • *****
  • Posts: 1177
    • View Profile
100 day sim
« Reply #10 on: September 12, 2007, 10:40:47 AM »
A little gues, maybe the virus was the only thing that has evolved. By stopping veggies to spreid it it maybe died out. Strange it could'nt just use bots for that.
When I looked at the dna in the 1000 hour sim, with every gene the bot was moving and reproducing(and randomly spreiding virusus sometimes) exept the first of both(there where 2 species). If you put that DNA in a bot it just does nothing, just nothing. So in that sim if you deleted all virus genes you'd keep bots doing  nothing. If the virusses have died out maybe something like that has happened, you are left with bots doing nothing maybe some of them have learned to duplicate often or an leftover of the virus.

And further what has happened further, has the age dropped, is the population growing, have the dynamic costs risen of have they fallen down.  

If something strange has happened like the falling down of dynamic costs, then something strange has hapened leading to no competion between the bots. And maybe then something like a sudden kill of the virusses could have lead to this.
Oh my god, who the hell cares.