Mutation Rate Modulation for Faster GAs


Welcome
Quick Background
Goal and Motivation
GA Setup Details
Expected Results
Actual Results
Code & Numbers
About

Code and Data for Download

Code

The source code, in C:

ga.c

ga.h

param.h

proj.c

util.c

util.h

makefile

Data sets

There are two types of datafiles. Each contains one record per line. Numbers are in ASCII format.

Each zip file contains a file called "avgRateFitness" which is a record of the average fitness of the rate-population for each epoch (line). There is only one number per epoch (line).

Each zip file also contains 100 files called ratePopulation*. Each file contains the 100 rate strings (one per line)in binary, with each set of eight bits separated by a space. There are five numbers on a line: the first four are the four eight-bit mutation rate encodings in binary, and the last one is the fitness value for that string during that epoch. There is one file per every ten epochs. The files contain only one epoch's data to keep the data manageable. The other 90% of the data is not recorded.

gaussian1.zip - The output files for the singly peaked Gaussian fitness function.

doublepeak1.zip - The output files for the first trial with a doubly-peaked Gaussian fitness function.

doublepeak2.zip - The output files for the second trial with a doubly-peaked Gaussian fitness function.


All contents © 2000 by Gabriel Au and Baldeep Sadhal.