CMU Artificial Intelligence Repository
GAGS: Genetic algorithm application generator and C++ class
GAGS 0.92 (Genetic Algorithms from Granada, Spain) is a system designed
to make it simpler to design a genetic algorithm. It consists of a C++
class library for programming genetic algorithms. But from a user
point of view, it is a genetic algorithm application generator. The
application generator takes a fitness function written in C and
surrounds with enough code to produce a complete program. It then
compiles it and runs it for you.
GAGS includes the following features:
- A "chromosome" hierarchy, with variable length chromosomes.
- Selection procedures include steady-state, tournament, and
- Crossover is 2-point. Uniform crossover is also supported in
the library, but the apps generator does not know about it.
- Mutation is bit-flip, with user-supplied frequency. Other
operators, like gene duplication or elimination, and gene
transposition, are also included in the library, but not in
the apps generator.
- Best fitness and individual are printed on STDOUT or to a file.
- Most GA parameters are inside a configuration file, so that
different operators and frequencies can be tested on a
fitness function. Only the fitness function and the output
options are fixed, once the program has been generated and
- Fitness evaluation using training files.
- Graphics output through gnuplot.
The application generator gags.pl is written in perl, so this language
must also be installed before using GAGS.
Version: 0.92 (23-AUG-94)
Requires: C++, PERL
Updated: Thu Oct 13 18:05:23 1994
CD-ROM: Prime Time Freeware for AI, Issue 1-1
Author(s): J.J. Merelo Guervos
Authors!Guervos, C++!Code, GA Application Generator, GAGS,
gagsprog.pgz Gzipped postscript for the programmer's manual.
Last Web update on Mon Feb 13 10:22:48 1995