Virtual Genetic Evolution

Sleeperservice writes "This story at New Scientist describes how, using cell simulation in computers, evolution can be simulated. How long until we can work out what the DNA sequence for a Dragon should be I wonder?"

Comments

Could we use this to better estimate....

[mraymer]

...the chances of life evolving on non-Earth like planets? For example, could someone run a simulation to see if organisms could thrive on a planet all Earth life forms would perish on?

I've always been interested in this, because if this is possible, it would seriously increase the chances of life being elsewhere, since the odds of an "Earth-like" planet are supposedly not that great.

GIGO

[Ben Jackson]

Simulated environments are just too complex.

And genetic algorithms, knowing nothing of your problem domain, tend to find solutions that capitalize on facets of your environment that you hadn't even considered. Two examples from my experience:

I had a population of simulated organisms competing in a shared 2d grid for food, which appeared in a pile at a random location when the old food was depleted. While the organism had basic looking/moving operations to rely on, invariably some would discover that with enough organisms, the food moves enough that you can survive by just looking around until the food is in your line of sight, and then jumping on it. My arbitrary decision to place the food randomly formed the basis for an *entire species* of organisms (which didn't fare too well when some got smarter).

These same organisms used a stack to do their thinking. Looking and eating produced values, which could be used for simple branching. Out of sheer laziness, I designed the stack to allow infinite pops off an empty stack which would return false, and infinite pushes on a full stack which would discard the values. One memorable run produced a dominant species which relied on this stack behavior to implement COUNTING! It intentionally (well, purposefully) left crud on the stack in a main loop, relying on the filled-stack behavior to detect a certain number of iterations. The stacksize and the arena size happened to be comparable, and this is how it determined when to turn.

Re:GIGO

[wormbin]

I had a similar experience to this.

I had a friend who believed in numerology: specifically that showing that if some numbers could be converted to other numbers via simple equations then that proved that these numbers had some kind of divine association and thus revealed something about the universe. I thought this was incorrect (kind words) and decided to write a program that would take any set of input numbers, a set of operators, plus a desired output number and then use a genetic algorithm to find the smallest equation to link the two. The individuals in the population were equations and they would behave similar to a biological population in that they would sexually reproduce, mutate, inferior equations are pruned, etc.

The program could go through 1000 generations rather quickly and produced very small equations. I never found numbers that couldn't be related by a very small equation.

One time the best equation looked something like the following (I forget the exact equation so this is just a non-real example):

666*666*666+666*666+666=7

The equation was obviously incorrect so I had no idea why it was chosen as the best equation. I then ran it through my evaluation function and it actually worked! It took a bit of head scratching but it turned out that my evaluation function was ignoring overflow and underflow in operations thus large values would eventually become negative. The genetic algorithm took advantage of this fact and produced equations that cleverly used overflow and underflow!

I thought this was pretty cool but it also felt a little creepy.

Another bit of evidence

[Anonymous Coward]

Wouldn't future computers have precision limitations that would cause artifacts in the simulated world? Relativity and the passage of time, randomness of subatomic particles.

Also, the authors of the universe would have to cheat from time to time to deal with the limitations of whatever hardware they're using. Not wanting to simulate an entire universe, they just crudely simulate distant worlds as point sources of information and place them far enough so the approximation will never be realized.

Also, adjustments would be made from time to time as the simulation evolved. After all, such a universe would be a product of curiousity and it would be difficult to not want to tweak certain things and see how the system is affected.

But then you get into the murky business of whether or not the simulators are simulated. If simulation=reality, it becomes illogical to continue pondering the question.

This work was done a long time ago

[AffineTransform]

Dr. Dan Ashlock at Iowa State University has been evolving robots that push blocks around a 2-D world for some time now. Check out the BotWorld link on his research page for info. On a side note, I did a research project for one of his courses during Spring 2002, where I evolved robots for Robocode that were effective in destroying human-programmed robots.