emergent/swarm/evolutionary systems etc
Josiah Carlson
jcarlson at uci.edu
Mon Mar 29 23:46:17 EST 2004
[snip long and involved post]
What you have posted is currently generally known as a "Genetic
Algorithm". Searching for that term on Google should net you many links.
In terms of applicability, genetic algorithms are decent at
approximating solutions to a class of problems known as NP. You can
search through various resources for more information on NP.
Now, I say approximate, because if you look at the *space of possible
solutions*, genetic algorithms will tend to find local optimums. That
is, the space is huge, and genetic algorithms will find you a set of
decent solutions. These solutions are rarely, if ever, the true
optimal, but can, with the right crossover and mutation functions, be
pretty damn good.
In terms of using genetic algorithms to write programs, I haven't read
any papers in the last few years with any positive results in that
specific field. It is possible (in fact, quite likely) that I am
missing some sort of paper on the topic, but I wouldn't hold my breath.
So yeah, search for genetic algorithms and NP for their description, and
you'll probably find something you like.
Additionally, you may want to read Stephen Wolfram's "A New Kind of
Science". It is available on the internet for free.
- Josiah
More information about the Python-list
mailing list