Parallel Genetic algorithm implementations

Design and implement a simple parallel genetic algorithm using master slave model. A simple genetic algorithm uses fitness proportional selection, one-point crossover, and point mutation. Use this simple parallel genetic algorithm to investigate the relationship between commputation time and speedup. Define your own fitness computation (evaluation function).

I would like to see a graphs of speedup (1/t_r) versus the number of processors (n) where t_r is the relative running time for n processors assuming that the running time for 1 processor is 1. Plot these graphs for n = 1, 2, 4 processors and for fitness computation times that range from much less than one second to one second or more. I would like to see at least three graphs for three different computation times in the above range. Once again, make up your own fitness functions.

Put these graphs on your web page and send me the URL.

You will need to get accounts on gaspac.cs.unr.edu. Please see me for this.

• Description of Chris Miles' Parallel GA implemenation with code.

• You could use MPICH MPI or LAM MPI. These are two implementation of the MPI standard. MPICH is installed on Gaspac, LAM MPI has to be downloaded, compiled, and installed in your gaspac (or other) account.

Good Luck