PEA - An EA library in Java

The PEA library was created to suit the needs of two Master's theses (V. A. Valko: Self-calibration in Evolutionary Algorithms: adaptive population size, and A. T. Vigh: Behaviour Analyses of Multi-parent Recombination Operators in Genetic Algorithms). As PEA was written in the Java language, it should run on all platforms that have a Java Runtime Environment (JRE) installed.

From this page the complete source code of the PEA package can be downloaded, including the parameter files of the experiments described in the following paper:
A. E. Eiben, A. T. Valko, and W. Hordijk. An Experimental Analysis of Multi-Parent Crossover in Genetic Algorithms. In preparation, 2007.

Download PEA

The files are available in a gzip'd tar file: pea.tar.gz (1.12MB).

To install PEA, follow these steps

This description assumes Unix machines. On other operating systems do the appropriate commands for your system.

1. Make sure there is a JRE on your system. If not, download and install one first, e.g. from java.sun.com.
   


2. Unzip and untar the PEA package:
   
   > gunzip pea.tar.gz
   > tar xf pea.tar
   
   Change to the pea directory:
   
   > cd pea
   


3. PEA comes compiled, but if you want to compile it again you need a java compiler, e.g. javac from Sun. Otherwise, go to step 4.
   
   Build PEA
   
   > make clean
   > make
   


4. Set the environment variables by running
   
   > ./setvars.sh
   
   or modify your classpath directly by writing into your .profile:
   
   export PEA_HOME=[home directory of pea]
   export CLASSPATH=${PEA_HOME}/lib/pea.jar:${CLASSPATH}
   
   Optionally, if you want to use the MersenneTwister random number generator from CERN, read the README file in opt/src/pea/random and also append
   
   ${PEA_HOME}/lib/colt.jar
   
   to the classpath.