Genetic Adaptive Systems LAB (GASLAB)

Home	Code	Publications	Email Us	People	Projects	Links

---

These are genetic algorithm implementations and support tools for using them. The tools are preliminary versions of an advanced integrated system for GA application and monitoring. The code is free for educational use.

• GOOGU (Generic Object-Oriented Graphing Utility) is a Java-based tool for monitoring GA progress (or almost anything else, for that matter) in real time or after action. It can be run as an application or over the net as an applet. Download [1.3.5, October 1998]. There is documentation.
• GAGA (General Architecture for Genetic Algorithms) is the companion to GOOGU. It provides an extensive and powerful framework for using GAs in diverse applications. It offers developers the ability to choose many features and parameters of the GA, and will offer a setup/configurator as well as standard and island-model parallel modes using PVM. GAGA is currently available in C and C++ versions; the C version is more full-featured and complete; the C++ version is elegant but esoteric. Download C version or C++ version
• A simple GA tarred and gzipped. This is a C version of Goldberg's simple GA code from his book. No warranties. Buy the Book. It explains the code and is a clear, informative, easy to read introduction to Genetic Algorithms. David Goldberg's wry sense of humor is evident throughout.

• Augmented GA and a corresponding input file. This code comes with a kind of elitist selection (CHC as described in Eshelman's 1989 paper), a generalized n-point crossover operator, linear scaling, and other goodies. Use this if you are looking for something that will tend to quickly find good solutions (in case the simple GA doesn't).

• Petrol: A simple parallel GA using PVM . This is a parallel version of Goldberg's simple GA. There is a help file. Designed and implemented by Eric Blood .

• Apetrol: augmented version of the parallel GA code above . My modifications of Eric's code.

• CIGAR: Case Injected Genetic AlgoRithm A robust genetic algorithm that uses a case-base of past experience to improve performance on new (similar to previosly attempted) problems. Can also run as a genetic algorithm. Publications describing CIGAR

• GAS: Generic Application Server (or Genetic Algorithm application Server) . Deploy Applications (especially genetic algorithm applications) on the web. Works with GOOGU and the various GAs above to visualize and monitor a running application. Very flexible but a pain to configure -- we are working on a configurator.

• Simple Classifier System code. Tarred and Compressed. Not ours.

For more information: gaslab@cs.unr.edu