coevolution

[Anthony Liekens]

Dear carl, and other GAlib users,

together with some of our students, we are working on several extensions
of the GAlib library. We already have a diploid library and a
coevolutionary library, and are currently working on a bigger project to
create a library for parallellization under MPI (which is nearing
completion.

Our student Anton Crombach worked on a "DGAlib" which contains several
classes for working with diploid GAs and coevolutionary GAs. A tarball
that contains the dump of the CVS repository of his work can be found at
my local web server (this is a replacement for GAlib, since it contains
all GAlib classes):

http://pc-04.bmt.tue.nl/dgalib/dgalib.tar.gz

If you unpack this tarball, you can find the classes needed for
coevolution in directory "dgalib/dga/coevolution". A version of Hillis'
work [Hill90, Hill92] as a test for the coevolutionary classes can be
found in directory "dgalib/tests/hillis/experiment".

His report that discusses the design of both diploid and coevolutionary
libraries can be found at 

http://pc-04.bmt.tue.nl/dgalib/report.ps.gz

A reference of the diploid and coevolutionary classes can be found here:

http://pc-04.bmt.tue.nl/dgalib/reference.ps.gz

A reference of the implemented experiment of Hillis' work can be found
here:

http://pc-04.bmt.tue.nl/dgalib/experiment.ps.gz

Please note that Anton provided us with excellent work on these
libraries, and really deserved his "9 out of 10" mark.

I hope these libraries can help you with your coevolutionary
implementations, and please let us know if there are any interesting
result coming from your experiments, or comments you mights have
concerning the library.

[Hill90] W. D. Hillis, 1990, "Co-evolving Parasites Improve Simulated
Evolution as an Optimization Procedure", Physica D 42, pp 228-234

[Hill92] W. D. Hillis, 1992, "Co-evolving Parasites Improve Simulated
Evolution as an Optimization Procedure", Artificial Life II, pp 313-323

On Wed, 2002-11-06 at 20:52, carl keller wrote:
>    I want to write a coevolutionary GA using GAlib.
>    Does anyone have pointers. 

-- 
Anthony Liekens <a.m.l.liekens at tue.nl>
Faculty of BioMedical Engineering
Technische Universiteit Eindhoven
Eindhoven, Netherlands