commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Al Chou <hotfusion...@yahoo.com>
Subject [math] Fw: Genetic Algorithms
Date Tue, 01 Apr 2008 15:35:22 GMT
Glad to see you guys getting off the ground.  I also did a bit more looking around for existing
projects with compatible licenses and came across Evolvica http://evolvica.org/ .  I approached
Andreas Rummier, its sole author, about contributing to Commons-Math and got the following
encouraging response.  Andreas has read this thread thus far and plans to join this mailing
list.

Al

----- Forwarded Message ----
> From: Andreas Rummler
> To: hotfusionman@yahoo.com
> Sent: Tuesday, April 1, 2008 12:56:52 AM
> Subject: Re: Fw: [math] Genetic Algorithms
> 
> Al Chou schrieb:
> 
> Hello Al,
> 
> > ...
> > I admit I haven't looked deeply at Evolvica's features, but they do seem to
> > address at least some of the wishes listed below.  Please let us know if you
> > would be interested in joining our community.
> > ...
> 
> To make a short quick answer, yes, I would be intested ... ;-)
> 
> A longer answer concerns the plans the plans for a GA framework. I looked up the
> thread Phil started on the mailing list, so I probably know what you want
> achieve with such a framework.
> 
> Evolvica has been around for quite some time, the main work has been done back
> in 2004/2005 during my time at the university. I have to admit that the frame-
> work never gained great attention, maybe because I didn't spend much effort in
> marketing. The current state can be circumscribed as "nearly abandoned", because
> I never achieved to form a community around Evolvica, which actively contributes
> to the framework. I'm certain Commons Math would be a much better platform for
> that and for this reason I would be glad to contribute the code to you. And I
> could offer to volunteer to get things back to life, although my time is
> somewhat limited (I'm not working at a university any longer). The big advantage
> is, that I already spent a lot of thoughts on design questions on GA frameworks.
> 
> I think nearly all of the features Phil stated in his post are already satisfied
> by Evolvica, although some things might be simplified. The overall structure
> consists of three parts: a core which is able to execute arbitrary algorithms,
> no matter if they are genetic or not (BTW, the correct term is "evolutionary",
> "genetic" means a special kind only), a class library containing many genetic
> operators and an Eclipse plugin which helps you to design and configure
> algorithms in a graphical way. I see the first two parts in the scope of Commons
> Math. At this point some something else comes to my mind: how complex should a
> framework grow to not go beyond the this scope? I see Evolvica as a framework
> for expert users, already being familiar with evolutionary algorithms. Wouldn't
> that raise the bar for wide adoption too high? Should a user be able to use an
> algorithm with a fixed structure out of the box? This may work with GA, but in
> most cases it will not. So, a user has to design the structure of the algorithm
> AND the operations contained in there - quite a lot of effort ...
> 
> But besides, I'm very curious about how you will proceed here ... ;-)
> 
> Best regards,
> 
> Andreas





      ____________________________________________________________________________________
You rock. That's why Blockbuster's offering you one month of Blockbuster Total Access, No
Cost.  
http://tc.deals.yahoo.com/tc/blockbuster/text5.com

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message