ECJ-INTEREST-L Archives

April 2009

ECJ-INTEREST-L@LISTSERV.GMU.EDU

Options: Use Proportional 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:
Fri, 10 Apr 2009 14:00:12 -0400
MIME-version:
1.0 (Apple Message framework v930.3)
Reply-To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Content-type:
text/plain; charset=US-ASCII; format=flowed; delsp=yes
Subject:
From:
Keith Sullivan <[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 (43 lines)
Esmail,

This can be done within the ECJ framework.  You need to create custom  
version of breeder, mutation, etc. depending on what parameters you  
want to change.  Look at the es and gp packages for examples on how to  
change individual size and mutation parameters.   Another way to  
change the size of the individual is to change in your Evaluate method  
using, say, arrayCopy.

Keith


On Apr 10, 2009, at 12:01 PM, Esmail wrote:

> Greetings, a few questions regarding ECJ and its documentation.
>
> First of all, this all looks very impressive and I plan to spend  
> some time
> getting to know this system.
>
> I wonder if it is possible to dynamically change some of the  
> parameters
> during run time. For instance if I wanted to change the population  
> size
> of the probabilities for mutation or crossover during run time, is  
> this
> possible?
>
> What about the length of the vectors, ie. if I represent a potential
> solution made up of a vector of 10 bits, can I dynamically change this
> during run-time to 20 bits? Note, I am not talking about the  
> population
> size, but rather changing the size of the individuals.
>
> Thanks,
> Esmail
>
> ps: In tutorial #2, just above the section "Specify the Breeding
>    Pipeline" it says: "we'll use two-point crossover (so
>    crossover-prob doesn't really matter)" ? Why is this? Does this
>    have to do with how crossover is defined for integers and the fact
>    that the probability is set to 1?

ATOM RSS1 RSS2