ECJ-INTEREST-L Archives

February 2006

ECJ-INTEREST-L@LISTSERV.GMU.EDU

Options: Use Monospaced Font
Show Text Part by Default
Condense Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Sender:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Date:
Sat, 4 Feb 2006 11:39:19 -0500
MIME-version:
1.0 (Apple Message framework v746.2)
Reply-To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Content-type:
text/plain; charset=US-ASCII; delsp=yes; format=flowed
Subject:
From:
Sean Luke <[log in to unmask]>
In-Reply-To:
Content-transfer-encoding:
7bit
Comments:
To: ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Parts/Attachments:
text/plain (74 lines)
On Feb 4, 2006, at 2:18 PM, Tony Hirst wrote:

> what can i actually use it to change?
> I've managed to chage the rand seed, num of gens etc in the control  
> tab,
> but don;t seem to be able to twek eg the pop size (posibly  
> understandably)
> or the crossover rate etc
>
> Is there a way of being able to :a) change these parameters in the
> console; b) save those changes? etc.

The GUI is in its infancy -- we still mostly use the command line  
ourselves.  You can change the parameters to shadowed versions of the  
parameters, and you were *supposed* to be able to change them to any  
value, (click on the cell in the Value column), but there are  
obviously bugs there.  Before I look into it, if anyone else can see  
something....

In the short run, the intent was to be able to edit the parameters  
via a JComboBox.  You type in a parameter or you pop up one that's  
shadowed in an overridden parameter file.

In the long run, you should be able to add new parameters as you like  
and/or save out changes.  That's totally unimplemented.


> i've also run into grief with the gui - eg on the apps/gui/ 
> coevolve2 demo
> i get:
> [Image:04022006_141425_1.png]
>
> any pointers on how to fix this?

This looks like JFreeChart has changed its method signatures yet  
again.  We'll have a new version of ECJ out fairly soon if you want  
to wait: otherwise you have to tweak a method here or there and  
recompile.  It's possible that a simple recompile without tweaks may  
be sufficient.  I might be able to provide a more current version of  
ECJ to you if necessary if you want to test against that.



> and a quesiton from a long time ago - is there a classifier system
> demo/example code anywhere i could play with?

Not at present [from GMU anyway].


> and finally finally, does the ecj project accept code contributions  
> from
> outside? i am an academic on a course team for a distance education
> university, writing a module on evolutionary computation, and am  
> looking
> to use ecj. If we do go with it (i need to check the license again,
> probably) then our s/w developers may have a tinker with graphiocal
> interfaces for ecj. If they do, i would like to think the code  
> could be
> contributed back...

ECLab has always gladly accepted (and received) contributed code,  
though we are fairly harsh in our requirement that such code adhere  
to ECJ's framework style, down to setup() and parameters and related  
documentation.

With the new release of ECJ, we'll have a contrib distribution  
avaialble obstensibly for less adherent contributed code, where you  
can stash stuff.  You'll be responsible for getting that code working  
under newer versions of ECJ when we change things, though we try hard  
to be backward compatible unless we really really need to change  
something.

Sean

ATOM RSS1 RSS2