README

**************************** MapReducing Genetic Algorithms ***********************

Author: Abhishek Verma (verma7@illinois.edu)

This package contains the source code for MapReducing Simple and Compact Genetic
Algorithms.

For details, please read: 
1) Abhishek Verma, Xavier Llorà David E. Goldberg and Roy H. Campbell, "Scaling Genetic Algorithms using MapReduce". Ninth International Conference on Intelligent Systems Design and Applications (ISDA), pp.13-18, Pisa, Italy, Nov, 2009. (http://verma7.com/pdfs/ISDA09_MR_GA.pdf)
2) Abhishek Verma, Xavier Llorà Shivaram Venkataram, David E. Goldberg and Roy H. Campbell, "Scaling eCGA Model Building via Data-Intensive Computing". IEEE Congress on Evolutionary Computation (CEC), Barcelona, Spain, July 2010. (http://verma7.com/pdfs/CEC10_MR_ECGA.pdf)
3) Xavier Llorà Abhishek Verma, Roy H. Campbell, and David E. Goldberg, When Huge is Routine: Scaling Genetic Algorithms and Estimation of Distribution Algorithms via Data-Intensive Computing. Parallel and Distributed Computational Intelligence, Natural Series of Computational Intelligence (SCI) 269, pp. 11-41, Springer Berlin/Heidelberg, 2010. (http://verma7.com/pdfs/Springer10_Meandre.pdf)

Requirement: Hadoop 0.19 or higher (and it's dependencies)

###################################################################################
You may need to modify the variable "rootDir" in MapReduce.java and CGA.java to
point to your home directory or some directory on the HDFS that you have read 
and write permissions on.

To compile the class files and the jar files, run

$ ant compile jar

This will create a build and bin directory for the class files, by compiling 
against Hadoop 0.19 jar located in the lib directory. Also, it will create 
ga.jar which MapReduces simple genetic algorithm and cga.jar which 
MapReduces compact genetic algorithm.

###################################################################################
To execute, go to the root Hadoop directory and run, 

$ hadoop jar ga.jar <nMaps> <nReducers> <variables> <nIterations> <popTimesNlogN>

where <nMaps> and <nReducers> is the number of mappers and reducers you want to 
start, <variables> is the size of the ONEMAX problem that this program solves, 
<nIterations> is the number of iterations you want to run and <popTimesNlogN> sets
the size of the population to a multiple of n log n, where n is the number of 
variables.

For example, 
$ hadoop jar ga.jar 4 4 10000 100 10

###################################################################################
To run compact genetic algorithm version, do

$ hadoop jar cga.jar <nMaps> <nReducers> <vectorSize> <nIterations> <popTimesNLogN>
 <tournamentSize> <iter>

where the arguments passed are similar to the previous command, except
<tournamentSize> is the size of the tournament and <iter> is the iteration number
you want the algorithm to continue for. 

For example,
$ hadoop jar cga.jar 4 4 1000 10 10 4 0
###################################################################################