Re: [jgap-users] Duplicate Chromosomes
Brought to you by:
klausikm
From: Klaus M. <jg...@kl...> - 2007-11-30 20:37:07
|
John, the reason for duplicate chromosomes is the BestChromosomesSelector that is constructed in DefaultConfiguration. In the latter class the parameter calls method setDoubletteChromosomesAllowed(true) from BestChromosomesSelector. If you replace this call with setDoubletteChromosomesAllowed(false) in DefaultConfiguration, then everything should work fine. I am currently enhancing the BestChromosomesSelector to clone chromsomomes even if the mentioned parameter is set to true. Although the change is completed, I'll have to adapt several test cases which is the difficult and longer-taking part of the task. Will be completed with the next release, which will appear this year asap. Maybe your other issue will disappear with that, too. I'll keep an eye onto it. Best Klaus > -----Original Message----- > From: jga...@li... > [mailto:jga...@li...] On Behalf > Of John Psaroudakis > Sent: Friday, November 23, 2007 4:14 PM > To: jga...@li... > Subject: [jgap-users] Duplicate Chromosomes > > I am using JGAP in my diploma thesis to create a GA used for Data > Mining and Rule Extracting. I have noticed that in the Population > object, there are many duplicate chromosomes, that is, > they point to > the same chromosome object. Does it have any relation with the > keepPopulationConstant parameter? Also, i have noticed > that whatever i > put in the keepPopulationConstant parameter, the population never > remains constant ! I start the evolution with > 100 chromosomes and after 70 generation, they reach > aproximately 270 > (which after some search i discovered that many of them > are duplicated > of existing chromosomes) > Thanks for your help in advance ! > > John Psaroudakis > > -------------------------------------------------------------- > ----------- > This SF.net email is sponsored by: Microsoft Defy all > challenges. Microsoft(R) Visual Studio 2005. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > jgap-users mailing list > jga...@li... > https://lists.sourceforge.net/lists/listinfo/jgap-users |