Text based games with numerous strategies including recursive and iterative minimax.
Based off of a CSC148 Assignment at UofT.
To play, run game_interface.py
The game_interface.py file relies on the generic Game and State classes in game.py and state.py, respectively. Therefore any game that can be made by subclassing these two classes can be played by game_interface.py. The games that can do this are generally two-player, sequential move, zero-sum, perfect-information games.
If you do make your own game using these subclasses, make sure to add the Game class to the playable_games dictionary on line 11 of game_interface.py.
All the minimax strategies work by checking all possible moves from the current state, and picking a move that results in the highest possible 'score', or a move where if both players play perfectly, the current player will win. If this is not possible it picks a move where a tie results, or a loss as a last resort.
Both the recursive and iterative minimax strategies work by, either recursively or iteratively, looking through all possible moves and picking a move.
Since it does look through ALL possible moves, it can take a long time to run; for example, running minimax on a new Stonehenge Game of board size 3 can take over 4 hours to run.
Memoization is similar to the recursive minimax strategy, however at each step of minimax, once a state's highest possible score has been found it gets added to a dictionary. Therefore, whenever we encounter a similiar state we already have it's score and don't have to do a recursive call on it.
This is one method of optimizing the recursive strategy to run faster/more efficiently.
Myopia is also similar to the recursive minimax strategy, however it limits the depth of recursion to 4. Once depth 4 is reached, it determines a 'best_guess' using a state's rough_outcome() strategy of whether we think the state will result in a win, tie, or loss.
This method sacrifices accuracy for speed.