Society of Robots - Robot Forum

Software => Software => Topic started by: lemontree on September 22, 2009, 05:07:34 AM

Title: Continuous Gray code optimization
Post by: lemontree on September 22, 2009, 05:07:34 AM

I found a paper about a really simple but effective algorithm for finding the minimum of a function recently.
I've put the paper and some sample code here:
http://code.google.com/p/lemontree/downloads/list (http://code.google.com/p/lemontree/downloads/list)
It kind of extends the idea of the Gray code (typically used in encoders etc.) to the continuous case and has
some nice scale free properties for optimization.
Sean O'Connor

Title: Re: Continuous Gray code optimization
Post by: RoboGeek on September 23, 2009, 12:03:12 PM

Hi. Hope this is not rude, but are you THE Sean O'Connor? ie slay?

Title: Re: Continuous Gray code optimization
Post by: lemontree on September 26, 2009, 11:18:07 PM

I am 'The' Sean O'Connor but not the guy who wrote slay. I know I have one namesake who writes a lot of games and another who has done a little bit of AI research.
My brother David has written a lot of games for EA and other companies. 
I've been trying out the Continuous Gray code for a few days now.  It is highly suited to open-ended evolution, co-evolution and dynamically changing or noisy environments  (ie the situations that robots would find themselves in). 
It is amazingly simple too.  It is well worth experimenting with. As it is very new you should feel free to try out any improvements to the algorithm you can think of.