mcts-ai

mcts-ai

This library allows you to create your own simple Monte Carlo Tree Search AI

Usage

Require the module

const MCTS = require("mcts-ai").MCTS;

To create a game, call the MCTS constructor

let game = MCTS();

By default it creates a 3x3 board with the ai taking 100 milliseconds per move.

Calling findNextMove() generates the "optimal" move for the current player

game.findNextMove();

Calling performMove(player, p) creates a specified player's piece at the target location

game.performMove(1, [1,2]);

Calling getBoard() returns the current board state

// returns
   [[0,0,0],
    [0,0,0],
    [0,0,0]]

Calling checkStatus() returns the status of the board:

-1: incomplete
0: tie
1: player 1 victory
2: player 2 victory

Options

You can customize the game by giving it different amounts of time for the ai to find next moves

let game = MCTS(1000);

You can also customize the game by customizing board size. For boards where height and width are not equal to three you will also have to give the ai a function to test for game status. The board is an array matrix with each row of the board represented by an array. 0 represents an empty spot. 1 a spot taken by player 1. 2 a spot taken by player two.

let game = MCTS(null, 5, function() {
	// write function to check game status
	// access board through this.boardValues
	// return -1 for incomplete game
	// return 0 for tie game
	// return 1 for player 1 victory
	// return 2 for player 2 victory
})

Note

Currently Only works for two player board games where all pieces are identical.