Recent GA Features
Newer GAs (those developed in the '90s) feature several
improvements which provide dramatic improvements. Try to
employ a GA that utilizes:
- Parallel populations
- Tournament selection
- Creep mutations
- Niching or sharing
- Elitism
GA Links
These links are not in any way meant to be canonical. They are
merely those sites which we have used in one form or another.
- MSU GARAGe
- Michigan State's Genetic Algorithms Research and Applications Group.
Developers of GALOPPS, the public domain parallel GA software
we've used for much of our GA work. This software evolves
better solutions faster than sequential GAs. Highly
recommended.
For recommendations on settings for parallel GA tuning
parameters, see the following paper: An
Efficient Migration Scheme for Subpopulation-Based Asynchronously Parallel
Genetic Algorithms
- SUGAL
Genetic Algorithms Library
- A very slick GA available for the MS Windows operating
system. Still undergoing development.
- Some
Simple GA Codes
- Nothing beats simple, easy to compile code when you're
just starting out. I recommend the GENESIS package.
-
GA Archive- a newsletter
- This is an index of back issues of a moderated newsletter
covering GA topics.
- Messy
GA code
- This is an ftp site for code provided by the Illinois Genetic Algorithms
Laboratory (IlliGAL). Code for a messy GA (variable length
genes) is available here.