Step 1: Unzip the folder containing this readme and the code if not done already. Step 2: Navigate inside of the innermost AI2048 folder. Step 3: Run the main.py file. Step 4 (Optional): The user can make small modifications to the main.py file to play different AI's at different depths. The main.py file contains the relevant explanations, but for good measure they are included here as well.
By default, running the main.py file will play the game with our expectimax AI at depth 3. To change to AI to use different search algorithms, simply modify the arguments passed to the timeAI() function in main.py. There are three arguments that can be modified, they are the following:
AI argument: Specifies the type of AI. Possibilites are "ExpectimaxOpti", "MiniMax", "MonteCarlo", "RandomAI". NB: Upper case matters.
Depth argument: Specifies the search depth for each AI. Irrelevant for randomAI. For MonteCarlo this is not the simulation depth(which is hardcoded to 2) but rather the number of times it runs through the Selection, Expansion, Simulation and Backpropagation phases, so the number of iterations.
In our report Results section, a selection of our logged data is presented. More data is however available, and can be accessed in the data/data.csv file.