developerWorks: Genetic Algorithms Simulate a Multi-Celled Organism
Oct 28, 2004, 05:30 (0 Talkback[s])
(Other stories by Teodor Zlatanov)
How to Help Your Business Become an AI Early Adopter
"My first two articles on genetic algorithms (GA) with Perl
observed the mutations and life cycles of single cells whose
fitness was entirely dependent on their own DNA. This article will
show how to simulate a multi-celled organism. The particular
application will be to generate letter puzzles, rated by their
complexity and correctness. You should refer to the first two
article for a GA background.
"The individual cells will be letter tiles in the puzzle. Their
fitness will be determined in combination with all the other cells,
so cell DNA alone won't mean much before it's been applied in
context. Furthermore, the DNA will have to be long but not complex.
It will just tell us what letters any particular cell likes to
connect to, and it will of course tell us the letter of this
particular cell (which can also be a blank tile)..."