kareypyer / busy-beaver-genome

A genetic algorithm to grow and test Turing Machines in search for Busy Beaver candidates

Vinnie Monaco

December, 2008

Part of the Busy Beaver Genome Project for Dr. Benjamin's CS385 Artificial Intelligence

Quickstart

All of the code should be Common Lisp compatible. It was built and tested with SBCL.

1. Change the project-root global parameter to the root of the project.

2. Make sure the results directory exists in the project root.

3. Edit the run time variables in main.lisp to change: -the size of the population -number of generations -mutation rate -machine states to search -shift limit to run any machine to -weights for the fitness function -number of searches to perform

   They will all already be defined. Change them to alter the search.

4. Compile and load main.lisp. This will compile and load every other file needed automatically, as long at the project-root is correct.

5. Run the function (run) to run the algorithm with standard output or (run-with-fitness-output) to run the algorithm with the fitness scores of each generation output to the results directory.

6. The function will return 'DONE when complete. Check the results directory for the evolutionary and fitness score results of each search performed. Those files will be overwritten if another search is performed without moving them, so rename or move anything worth keeping.