VOLVE 5.0
| Newer Posts | Older Posts |
ALIFE: 1 Year of Alife Status
January 18th 2007The new 1 year of alife looks good, the new vision system is being used, and the amount of variation I am seeing is good. I am working on a 3D version, which uses the IRRLICHT 3d engine. The 3d version will allow you to explore a simulation in a first-person perspective (rather than looking down). It will have "spectate mode" in which you just look around at a running simulation. And player mode. In player mode you are an actual organism, and you are part of the simulation. You use Quake navigation keys. When you encounter a creature, you can "EAT" using the mouse button. If you are killed, you're energy is deposited at that spot as organic material. Hopefully, as you play evolution will take place and adapt to your EATING of them. You goal is to eradicate all creatures. But each each time you're killed you start over in a random spot (and all your hard earned energy is consumed by those organisms that most likely killed you). If all goes well, you can download various simulations (from 1 year of alife for example) and be able to experience the strength or weakness of the evolving creatures first hand. In theory these creatures should get harder to kill just by competing against themselves. So you as a player, are not needed to breed more difficult opponents. Anyway, the possibilities for gameplay modes are huge, and it should make this a fun game as a well as an alife sim. Here's some evolve KFORTH that uses vision:
{ -3 5 2dup NEAREST2 EAT pop OMOVE ?loop 1 EAT NEAREST -2 AGE 2* OMOVE ?loop 1 EAT 2 call 1 }
ALIFE: 1 Year of Alife Status
January 18th 2007The new 1 year of alife looks good, the new vision system is being used, and the amount of variation I am seeing is good. I am working on a 3D version, which uses the IRRLICHT 3d engine. The 3d version will allow you to explore a simulation in a first-person perspective (rather than looking down). It will have "spectate mode" in which you just look around at a running simulation. And player mode. In player mode you are an actual organism, and you are part of the simulation. You use Quake navigation keys. When you encounter a creature, you can "EAT" using the mouse button. If you are killed, you're energy is deposited at that spot as organic material. Hopefully, as you play evolution will take place and adapt to your EATING of them. You goal is to eradicate all creatures. But each each time you're killed you start over in a random spot (and all your hard earned energy is consumed by those organisms that most likely killed you). If all goes well, you can download various simulations (from 1 year of alife for example) and be able to experience the strength or weakness of the evolving creatures first hand. In theory these creatures should get harder to kill just by competing against themselves. So you as a player, are not needed to breed more difficult opponents. Anyway, the possibilities for gameplay modes are huge, and it should make this a fun game as a well as an alife sim. Here's some evolve KFORTH that uses vision:
{ -3 5 2dup NEAREST2 EAT pop OMOVE ?loop 1 EAT NEAREST -2 AGE 2* OMOVE ?loop 1 EAT 2 call 1 }
In Real or Artificial Life, Is Evolutionary Progress in a Closed system possible?
January 15th 2007Answer: YES. It can be trivially shown to be true using the following proof: [b]STEP 1:[/b] Define what constitutes Evolutionary progress. [b]STEP 2:[/b] Create a closed system with agents that can posses only one of two possible genes. The first gene is the [b]"DO NOTHING gene"[/b]. The other gene is the, "Perform the thing that satisfies the definition of evolutionary progress" gene (as specified in STEP 1). Called the [B]"X gene"[/B] for short. [b]STEP 3:[/b] Create agents with the [b]"DO NOTHING gene"[/b] . [b]STEP 4:[/b] Eventually, one of the agents will produce offspring where the [b]"DO NOTHING gene"[/b] has been mutated into the [b]"X gene"[/b]. And thus the closed system will have demonstrated evolutionary progress. Ergo, you have demonstrated that yes, in real or artificial life, evolutionary progress in a closed system possible. End proof. [i]NOTE: This proof works for [b]any[/b] definition "X" that one cares to use as demonstrating evolutionary progress (and can be expressed as computer code).[/i] [b]Example:[/b] Lets say you want a closed system to start out doing nothing, but then evolve to play chess. Then define the X-gene as "play chess". Now create an artificial life simulator that has a chess program built into it, and this chess program is activated when an agent aquires the "X-gene". Then start out your simulation with agents that only posses the "DO NOTHING gene". In a short time, you will have agents that can play chess. Wow, how very impressive! (sarcasm) This is why the alife-x-prize is a prize for an inherently trivial question. I win!!!! Pay me fuckors!!! See the prize website at: [url=http://www.panspermia.org/eprize.htm#whatsnew]http://www.panspermia.org/eprize.htm#whatsnew[/url]
Alife box rebooted: power outage
January 15th 2007Austin had a power outage for a couple of hours yesterday. My battery backup worked but ran out of juice a few minutes before the power was restored. I had expected it to last longer. It's a huge battery and it was only powering the box (no monitors etc...). I expected it to last 4 hours, but I guess I calculated the watt hours wrong. The job of the battery back up was to survive those breif power drop outs (of which it has survived many). So I am not totally bummed by this event. It's back up and running. I am sorry to lose the awsome 'uptime' of 100+ days. Alife simulator is also running where it left off. On a related note everytime I reboot one of my suse linux boxes, i need to play around with routing to make is "see" the internet. I finally figured out what settings to play with, but why? Once I configure using YaST why does it keep forgetting my network settings on reboot? Very annoying.
ALIFE: Evolve 4.8 released
January 8th 2007One of the best and most polished releases yet! Evolve 4.8 is out, get it now! It's free damn it! It has a much improved vision system (that evolution actually finds useful enough to select for) I went back to the old absolute code block adressing scheme (in contrast to the relative addressing). Improved the documentation a lot. Created some fun sample simulations to watch (especially 'intestines.evolve') One year of alife restarted! Wooohooo. NOTE: White balance on those coffee drip photos sucks :-(
ALIFE: Quest for Vision
January 3rd 2007My little alife simulator was cool, but nothing was evolving that would use vision. Vision would make these creatures exhibit the kind of action at a distance that vision permits. The kind of behaviours in which the creature examines its environment and acts accordingly. I am now experimenting with a new LOOK instruction. LOOK takes an (x, y) vector. Currently the these look vectors are actually terminated at the endpoint of the vector. Now I am testing a new approach where the vector projects to infinity (to the end of the bounds of the simulation universe). Preliminary results are good! Within only about an hour of simulation there are creatures using the LOOK-SPORE instruction to alter its movement. I am very encouraged by this new tweak.
| Newer Posts | Older Posts |