Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!newsstand.cit.cornell.edu!ub!dsinc!spool.mu.edu!munnari.OZ.AU!news.mel.connect.com.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!matilda.vut.edu.au!rhh
From: rhh@matilda.vut.edu.au (Robert Hinterding)
Subject: Re: Q. Multiple chromosones
Message-ID: <Do7Hr1.Lt@matilda.vut.edu.au>
Organization: Victoria University of Technology
References: <4him10$dpq@c14dn2.pri.agt.ab.ca> <4i20b8$isu@soleil.uvsq.fr>
Date: Wed, 13 Mar 1996 12:37:48 GMT
Lines: 70

Olivier Chocron <chocron> writes:

>Hello, Malcolm J. Jones & all

>mjones2@ent.agt.ab.ca wrote:
>>A natural gene has multiple chromosones.  My question is what advantage does this offer the
>>natural creature.  From this question I then ask if this advantage might be transferable to
>>my artificial gene (and artificial creature).
>>
>>My thoughts:
>>1)  No advantage - multiple chromosones happened by chance with no cost accruing to the creature
>>2)  "Critical" features encoded at the head of each chromosone, thus ensuring that they are
>>not corrupted by faulty crossover.
>>3)  Advantage served by "Massive Parallel Processing".
>>
>>Has anyone investigated this?  Does anyone have other thoughts?
>>
>>Malcolm J. Jones
>>TELUS Corporation


>--
>I would rather say that the genome ( or genetic material) has multiple
>chromosomes.I am pleased that someone asked the question as I am asking to myself
>the same question for about two months.
>The answer of that question is undoubtly important for Artificial Evolution.

>As a first response, I would say (as I have ever done in this Newsgroup) that
>it allows an additional schuffling of genetic material during sexual reproduction
>while separating each homologous chromosome (Meiose) in diploid reproductive
>cells.For N=2n chomosomes in the diploid cell, it will result in
>two n chromosomes haploid reproductive cell which yield 2^n combinations.

>This kind of recombination will effectively bring a more parallel search as inside
>each individuals, ther will be several evolving chromosomes..
>It seems that it allows also good chromosomes not to be disrupted by the xover
>process while trying to optimize other chromosomes.

>For now, I have not tryed to use this advantage because diploid representation 
>bring some others difficulties and parameters to adjust as dominance.

It is nice to be able to agree with Olivier for a change.

Apart from the increase in crossove there are other advantages.  Using 
multiple chromosomes enables you to map different aspects of a problem
to different chromosomes.

For papers that discuss the use of multiple chromosomes see:
Juliff, K., A Multi-chromosome Genetic Algorithm for Pallet Loading,
Proceedings of the 5th ICGA,1993, pp467-473.

Hinterding, R. & Juliff, K., A Genetic Algorithm for Stock Cutting: An
exploration of Mapping Schemes, Technical Report 24COMP3,Victoria University
of Technology, 1993.

And how the above was improved by going to single chromosome representation

Hinterding, R., & Khan, L., Genetic Alorithms for Cutting Stock Problems:
with and without contiguity, Lecture Notes in AI, Vol ?, 1995.
(see my home page for the full reference, and copies of the last two
papers)

Cheers,
Robert

-- 
Robert Hinterding                       Email: rhh@matilda.vut.edu.au 
VICTORIA UNIVERSITY OF TECHNOLOGY       Fax:   +61 3 9688 4050
P.O. Box 14428, Melb Mail Centre        Phone: +61 3 9688 4686              
AUSTRALIA 3000                          Home Page: http://dingo.vut.edu.au/~rhh
