Sasha ([info]alexanderwait) wrote,
@ 2004-09-12 10:20:00
Previous Entry  Add to memories!  Tell a Friend  Next Entry
Live from ALife IX...
I am listening to an interesting talk about a computational model of DNA computing-- at the ALife IX conference-- this morning. My talk An Evolutionary Approach Generates Human Competitive Corewar programs seems to have gone OK. I was the first speaker at the Artificial Chemistry workshop. In a few minutes I'll be having lunch with my old adviser--Gilles Brassard-- who is flying to boston just to hear my tutorial this afternoon. If you are interested, please, sign up to my low traffic mailing list.

I also have a poster here. Have a look:

(Read 10 comments) - (Post a new comment)

Re: Dimensionality and universality
[info]shamebear
2004-12-26 12:14 am UTC (link)
I agree that Wolfram isn't much help here. I must also admit to know too little about graphs resembling euclidean spaces to make use of that approach.

However I liked the approach taken by for instance Kauffman in his book "at home in the universe" Consider a grid of "lightbulbs" where the input to each lightbulb is the state of the other lightbulbs it is connected to. Each set of inputs produce either light on or light off as output. Kauffman shows the "interesting" range to be at the "edge of chaos" for certain settings. A setting of K = 2 connections to other lightbulbs appear good. If K is increased, another parameter P should be adjusted to compensate.

P can be defined as (Number of input-combinations that give ON as output)/(Total number of input-combinations) and is a measure of the bias the lightbulb has towards one setting. If this bias is strong (P near 0 or near 1), the network may still be in the interesting range if K is high.

In a 2D cellular automata, each cell usually have four neighbours. in 3D it has six. That means our "lightbulb" should show a strong bias to compensate. (Kauffman also advices to stick to so-called "canalyzing" boolean functions)

These seem like good practical advices. While Kauffman talks of abstract spaces, not real-life euclidean, that's not too bad! Kauffman claims that life started as an "autocatalytic process", a certain web of chemical reactions in the primordial soup. Each cell then get to symbolise a certain molecule and the flow of nutrients could symbolise the relative amounts of molecules as one is spent to produce another. When it's all jumbled in a soup anyway, the elements exact position in the bowl is secondary.

"interesting" behaviour could be several instances of the same nearly closed webs of reactions (different individuals of the same species), showing homeostasis yet never stopping their evolution.

If these webs could be identified and displayed properly, very interesting behaviour would be allowed to occur on the screen without basing it in a physical space.

(Reply to this) (Parent)(Thread)
Re: Dimensionality and universality
[info]alexanderwait
2004-12-27 03:59 pm UTC (link)
"While Kauffman talks of abstract spaces, not real-life euclidean, that's not too bad! Kauffman claims that life started as an "autocatalytic process", a certain web of chemical reactions in the primordial soup. Each cell then get to symbolise a certain molecule and the flow of nutrients could symbolise the relative amounts of molecules as one is spent to produce another. When it's all jumbled in a soup anyway, the elements exact position in the bowl is secondary."

There have been lots of attempts to build evolving, artificial worlds with "a certain web of chemical reactions." But these webs are very hard to analyze. It's much simpler, frankly, to use 3D euclidian space.

(Reply to this) (Parent)

(Read 10 comments) - (Post a new comment)

Create an Account
Forgot your login or password?
Login w/ OpenID
English • Español • Deutsch • Русский…