2048 environment and MCTS simulation

File contents for MCTS and GUI:
└── decision_tree
    ├── MCTS.exe
    ├── decision_tree.cpp
    ├── tree.txt
    └── Py_Master
        ├── yada yada
        ├── puzzle.py
        └── yada yada

How to use the MCTS inside GUI: Technically, just run puzzle.py and it should work. In practice, I have no idea how to fix the problems with absolute paths so you have to go into decision_tree.cpp, change the absolute path of tree.txt to match your dir architecture, build it (I used g++ decision_tree.cpp -o MCTS.exe), go into Py_Master -> puzzle.py and change the absolute paths of tree.txt and MCTS.exe to match your own.

Then run puzzle.py. The key N gives you the next board.

if you wish to use the MCTS without the GUI, you can simply edit tree.txt with the board you want it to evaluate and call the executable MCTS.exe in a bash. If the absolute paths are correct, the script will read the board and output the resulting move in the terminal.

How to use this environment:

1. Put the current state (4x4 matrix) into input.txt, followed by the move you wish to make

    0 - up
    1 - right
    2 - down
    3 - left

2. cd into the 2048-environment directory and run the following commands:

g++ move.cpp -o main  ->  compiles the program, run if you change the source-code in move.cpp.
./main                ->  runs the program.

3. In output.txt you will find:

 1 - legal move
-1 - illegal move

If the move was a legal one, it will also contain the new 4x4 matrix.

4. In score.txt you will find the attributed score of the move. This score means to give an approximate value as to how good the move was, but is tremendously biased towards having the biggest value in the bottom-right corner.