The code explores multiple toy systems: Rubik's Cube, N-puzzle, Wordgame. Our goal is to use multiple configurations of each system and understand if taking mutations can improve the configuration. To score a mutation, we use a $f$ score.

Each system has at least one specific $f$-score function where $f(x) = 1$ for a configuration $x$ represents the configuration is in the desired state.

To run the code on either the n-puzzle system or the rubik's cube system you would use the following functions.

get_mult_cube_results in cube_runner.py gives you the results for the Rubik's cube.

get_mult_puzzle_results in n-puzzle.py gives you the results for the N-puzzle.

Below are the standard configurations for both functions:

get_mult_cube_results(num_cubes, random_cube = False, num_splits = 20, depth = 20, plot=True, use_original_f_score=True, allow_repeats=False, solvable=True):

def get_mult_puzzle_results(num_puzzles, max_iterations = 100000, artificial_start_traps = True, num_splits = 20, depth = 20, plot=True, allow_repeats=True, use_solvable_puzzles=True):

num_cubes or num_puzzles specifies the number of configurations to start out with

num_splits splits all the randomly picked configurations based on their initial f-score. If there are 2 splits, the splits would be from 0-0.5 and 0.5-1.0

depth specifies the number of mutations per configuration

allow_repeats allows each configuration to pick a repeated configuration.

solvable lets the user pick either originally solvable or un-solvable configurations

To generate the figures:

• Figuers 1-4 can be generated with get_multiple_dfe_dist in n-puzzle.py
• Figure 5-6 can be generated with get_dfe_distribution in cube_runner.py
• Figures 7-8 can be gnereated with get_dfe_distribution in n-puzzle.py
• Figure 9 can be generated with get_multiple_dfe_dist in cube_runner.py
• Figure 10 can be generated with get_multiple_dfe_dist in n-puzzle.py