ECJ-INTEREST-L Archives

June 2008

ECJ-INTEREST-L@LISTSERV.GMU.EDU

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
Subject:
From:
David R White <[log in to unmask]>
Reply To:
ECJ Evolutionary Computation Toolkit <[log in to unmask]>
Date:
Fri, 27 Jun 2008 12:49:36 +0200
Content-Type:
text/plain
Parts/Attachments:
text/plain (45 lines)
Thanks for the reply.

I've implemented the statistics chain method, which works fine. 

Cheers

David

Sean Luke wrote:
> Interesting question.  prepareToEvaluate has a different purpose than 
> what you're looking for.  What you want to do is have the prototype 
> Problem generate its test cases prior to being cloned each 
> generation.  You could do this by overriding the 
> Evaluator.evaluatePopulation method to call some method on the Problem 
> and then calling super.evaluatePopulation.  Alternatively you could 
> stick in a Statistics subclass into the Statistics chain which 
> overrides its preEvaluationStatistics method to call the method on the 
> Problem.
>
> Sean
>
> On Jun 24, 2008, at 10:42 AM, David R White wrote:
>
>> Hi,
>>
>> I would like to create a single set of test cases for all individuals 
>> in a generation, and still be able to use multithreaded evaluation.
>> Is it possible to do this in an elegant manner?  If I use 
>> "prepareToEvaluate" then it will be called for each Problem instance 
>> that is cloned, i.e. once for each thread, and as my test cases are 
>> probabilistically selected I will end up with a different set of 
>> tests cases for each thread, which I don't want!
>> Any help appreciated,
>>
>> Thanks
>>
>> David
>>
>> -- 
>>
>> David R White
>> PhD Student
>> Dept of Computer Science
>> University of York, UK.

ATOM RSS1 RSS2