This is 2048 game clone for Octave/Matlab. Its main use is the study of playing strategies.
How to run one game session:
>game_random % choose random move each time
or
>game_corner % choose first of UP, RIGHT, DOWN, LEFT, whichever results in change
or
>game_corner2 % the same as previous, but UP and RIGHT are swapped after each move
or
>game_greedy % chooses the move with most points
Following runs all strategies 100 times and produces some diagrams about effectivess of each:
>game_plot
To run experimental Q-learning strategy:
>game_qtable
To plot results of 1000 iterations:
>results = game_qtable(1000);
>plot(results);
To save/load learned Q-table:
>global Q;
>save Q;
>load Q;
The game was tested in Octave, but probably runs in Matlab as well.