August 2006


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

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

Print Reply
Robert Baruch <[log in to unmask]>
Reply To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Wed, 30 Aug 2006 09:57:24 -0400
text/plain (60 lines)
Hi Jurie,

I hate messing around with makefiles, which is why I use a  
development environment such as JBuilder or Netbeans. Then I just say  
"compile everything", and it does it!

For JFreeChart, unzip the package and look in the lib directory.  
You'll need to include jfreechart-1.0.1.jar and maybe  
jcommon-1.0.0.jar and gnujaxp.jar in your classpath when you compile  
and when you run.

For itext, it should just come in a jar file, so you need to include  
that in your classpath for compiling and running.

A development environment makes that easier -- you just add those jar  
files to the required libraries, and it will take care of the rest.

I think genetic programming is good for evaluation functions. In  
fact, that's all genetic programming really does -- generate  
functions, whether that function is an evaluation function, an  
equation, a function which moves Pac-Man in a game, whatever. Of  
course, I've only used GP, and not any other paradigm, so I could be  
biased :)

When you have an evaluation function, I'm pretty sure you mean some  
function which returns a number which is higher if the board position  
is more favorable. GP is perfect for that. But the purpose of an  
evaluation function is usually to generate a bunch of hypothetical  
board positions, compare the evaluation function (possibly to some  
number of moves), and choose the best move based on the results.

What I did was evolve a function which would actually play the game  
-- it would make a move. It didn't use minimax or anything like an  
evaluation function. I guess it depends on whether your game is more  
analytic (less reliant on chance) or probabilistic (more reliant on  


On Aug 30, 2006, at 3:37 AM, Jurie Oberholzer wrote:

> Hi Rob
> Thanks so much for your help, needless to say I am still  
> struggeling to get the makefile to run to completion. I am not sure  
> where i should be unzipping the JFreechart files to?  I keep  
> getting the error that  package org.jfree,chart does not exist.  I  
> get a total of five errors when I run the makefile.
> With regard to your other comment, I am looking at using the  
> genetic programming paradigm and that is why I have chosen to use  
> ecj. Is ecj my best option? When you speak about evolving players  
> what exactly do you mean as I am looking at generating an  
> evaluation function mainly.
> Thanks for all your help.
> Jurie