• Announcements

    • khawk

      Download the Game Design and Indie Game Marketing Freebook   07/19/17

      GameDev.net and CRC Press have teamed up to bring a free ebook of content curated from top titles published by CRC Press. The freebook, Practices of Game Design & Indie Game Marketing, includes chapters from The Art of Game Design: A Book of Lenses, A Practical Guide to Indie Game Marketing, and An Architectural Approach to Level Design. The GameDev.net FreeBook is relevant to game designers, developers, and those interested in learning more about the challenges in game development. We know game development can be a tough discipline and business, so we picked several chapters from CRC Press titles that we thought would be of interest to you, the GameDev.net audience, in your journey to design, develop, and market your next game. The free ebook is available through CRC Press by clicking here. The Curated Books The Art of Game Design: A Book of Lenses, Second Edition, by Jesse Schell Presents 100+ sets of questions, or different lenses, for viewing a game’s design, encompassing diverse fields such as psychology, architecture, music, film, software engineering, theme park design, mathematics, anthropology, and more. Written by one of the world's top game designers, this book describes the deepest and most fundamental principles of game design, demonstrating how tactics used in board, card, and athletic games also work in video games. It provides practical instruction on creating world-class games that will be played again and again. View it here. A Practical Guide to Indie Game Marketing, by Joel Dreskin Marketing is an essential but too frequently overlooked or minimized component of the release plan for indie games. A Practical Guide to Indie Game Marketing provides you with the tools needed to build visibility and sell your indie games. With special focus on those developers with small budgets and limited staff and resources, this book is packed with tangible recommendations and techniques that you can put to use immediately. As a seasoned professional of the indie game arena, author Joel Dreskin gives you insight into practical, real-world experiences of marketing numerous successful games and also provides stories of the failures. View it here. An Architectural Approach to Level Design This is one of the first books to integrate architectural and spatial design theory with the field of level design. The book presents architectural techniques and theories for level designers to use in their own work. It connects architecture and level design in different ways that address the practical elements of how designers construct space and the experiential elements of how and why humans interact with this space. Throughout the text, readers learn skills for spatial layout, evoking emotion through gamespaces, and creating better levels through architectural theory. View it here. Learn more and download the ebook by clicking here. Did you know? GameDev.net and CRC Press also recently teamed up to bring GDNet+ Members up to a 20% discount on all CRC Press books. Learn more about this and other benefits here.
Sign in to follow this  
Followers 0
cryobuzz75

Negamax Ai for TicTacToe

13 posts in this topic

Hi all,

 

I'm doing a simple TicTacToe so that I can implement a Negamax algorithm which I can later use for other abstract games. However I'm encountering problems whereby the AI doesn't play the best move, and it loses constantly. My suspect is the static evaluation function. Here it is:

 

	public int getScore() 
	{
		int score = 0;
		
		CellState state = CellState.Empty;
			
		if ((cells[0] == cells[1] || cells[1] == cells[2]) && (cells[1] != CellState.Empty)) 
			state = cells[1];

		if ((cells[6] == cells[7] || cells[7] == cells[8]) && (cells[7] != CellState.Empty)) 
			state = cells[7];

		if ((cells[0] == cells[3] || cells[3] == cells[6]) && (cells[3] != CellState.Empty)) 
			state = cells[3];

		if ((cells[2] == cells[5] || cells[5] == cells[8]) && (cells[5] != CellState.Empty)) 
			state = cells[5];
		
		if (((cells[3] == cells[4] || cells[4] == cells[5]) && (cells[4] != CellState.Empty)) ||
		    ((cells[1] == cells[4] || cells[4] == cells[7]) && (cells[4] != CellState.Empty)) ||
			((cells[0] == cells[4] || cells[4] == cells[8]) && (cells[4] != CellState.Empty)) ||
			((cells[2] == cells[4] || cells[4] == cells[6]) && (cells[4] != CellState.Empty))) 
		{
			state = cells[4];
		}
		
		if (state == currentPlayer)
			score = getMaxScoreValue();
		
		else if (state == currentPlayer.getOpponent())
			score = -getMaxScoreValue();
		
		return score;
	}

 

Cells is just an array of size 9 that represent the board positions. CellState is just an enum with values {Empty(0), Player(1), Opponent(2)}. getmaxScoreValue just returns the highest score (65536),

 

Is this static function complete or am I missing other conditions?

 

Thanks,

C


 

Edited by cryobuzz75
0

Share this post


Link to post
Share on other sites

The evaluation function seems correct (although you only need to check `cells[4] != CellState.Empty' once in that long condition). The problem might be that you need to check those conditions in every node of the tree, not just the leaves. Maybe you can show us your search function (I would call it `NegaMax', but I don't know what you are calling it).

0

Share this post


Link to post
Share on other sites

All your conditions are only checking if two adjacent cells containt the same state.  I assume you mean to be checking for lines as you are assigning the max score to this.

So they should be

 

if ((cells[0] == cells[1] && cells[1] == cells[2]) && (cells[1] != CellState.Empty))
            state = cells[1];

 

the logical "and" replaces the logical "or"

1

Share this post


Link to post
Share on other sites

Here's my Negamax class:

 

public class Negamax
{
    IBoard board;               
    int maxDepth;
    private Move[] dummyMove = new Move[1];

    public Move GetBestMove(IBoard board, int depth)
    {
        maxDepth = depth;
        int alpha = -999999, beta = 999999;
        Move[] newMove = new Move[1];

	alphaBeta(board, depth, alpha, beta, newMove);

        return newMove[0];
    }

    private int alphaBeta(IBoard board, int depth, int alpha, int beta, Move[] bestMove)
    {
        bestMove[0] = null;

        if (depth == 0)
            return board.getScore();

        if (board.getWon() != 0)
            return board.getScore();
        
        List<Move> bestMoves = null;
        List<Move> moves = board.getMoves();

        if (depth == maxDepth)
        {
            bestMoves = new ArrayList<Move>();
            bestMoves.clear();
            bestMoves.add(moves.get(0));
        }

        int minimax = -board.getMaxScoreValue();
        int val;

        for (Move move : moves)
        {
            this.board = board.copy();
            this.board.makeMove(move, true);

            val = -alphaBeta(this.board, depth - 1, -beta, -alpha, dummyMove);
            move.setScore(val);
            
            if (val > minimax)
                minimax = val;

            if (depth == maxDepth)
            {
                if (val > bestMoves.get(0).getScore())
                {
                    bestMoves.clear();
                    bestMoves.add(move);
                }
                else if (val == bestMoves.get(0).getScore())
                    bestMoves.add(move);
            }
            else
            {
                if (val > alpha)
                    alpha = val;
                if (alpha >= beta)
                    return alpha;
            }
        }

        if (depth == maxDepth)
        {
            int rnd = MathUtils.random(bestMoves.size() - 1);
            bestMove[0] = bestMoves.get(rnd);
        }

        return minimax;
    }
}
0

Share this post


Link to post
Share on other sites

All your conditions are only checking if two adjacent cells containt the same state.  I assume you mean to be checking for lines as you are assigning the max score to this.

So they should be

 

if ((cells[0] == cells[1] && cells[1] == cells[2]) && (cells[1] != CellState.Empty))
            state = cells[1];

 

the logical "and" replaces the logical "or"

 

Ooops! I can't believe I missed that he was using the wrong operator. That's the first thing to fix. Then I wonder what `board.getWon()' does, since it must be almost identical.

0

Share this post


Link to post
Share on other sites

This is getWon():

 

	public int getWon() 
	{
	    //If any winning row has three values that are the same (and not EMPTY),
	    //then we have a winner
		for (byte c = 0; c < WINNING_POS.length; c++)
	    {
	        if ((cells[WINNING_POS[c][0]] != CellState.Empty) &&
	            (cells[WINNING_POS[c][0]] == cells[WINNING_POS[c][1]]) &&
	            (cells[WINNING_POS[c][1]] == cells[WINNING_POS[c][2]]))
	        {
	            return cells[WINNING_POS[c][0]].ordinal();
	        }
	    }
	 
	    //Since nobody has won, check for a draw (no empty squares left)
		byte empty = 0;
		for (byte c = 0; c < cells.length; c++)
		{
			if (cells[c] == CellState.Empty)
			{
				empty++;
				break;
			}
		}
		if (empty == 0)
			return 3;
		
	    //Since nobody has won and it isn't a tie, the game isn't over	
		return 0;
	}

 

getScore checks for 2 in a line. Even if I change maxscore with eg, 100, the AI still plays bad.

0

Share this post


Link to post
Share on other sites

I changed the operands in getScore() as suggested, and now the AI looks like it's playing good.

 

So my question is: Why should getScore() check for 3 in a line?

Edited by cryobuzz75
0

Share this post


Link to post
Share on other sites

So my question is: Why should getScore() check for 3 in a line?

 

It doesn't have to! If your search is looking for 3 in a line, you can just return a losing score when it happens: Since the 3-in-a-line condition just appeared, it must be that the opponent of the player to move won.

 

Tic-tac-toe is such a small game that you don't need an evaluation function at all. You can always just explore the tree to the end.

Edited by Álvaro
0

Share this post


Link to post
Share on other sites

I changed the operands in getScore() as suggested, and now the AI looks like it's playing good.

 

So my question is: Why should getScore() check for 3 in a line?

 

You evaluation function should generally return a value for a win/draw/loss at the very least.  For most games this isn't very useful because you aren't searching terminal nodes. However it still needs to be there so that the search function returns a value when it does find a terminal node.

 

For instance you were returning a score when you detected a getWon.  But this score was meaningless as it would just try to ensure there was at least one adjacent symbol and not find an actual winning position.

 

So on top of scoring a win/loss/draw you can add other criteria such as in your initial version of the scoring that detected useful features (adjacent symbols).  So that non-terminal nodes that are search will return a useful evaluation. 

 

So you can still test your negamax function on tic tac toe - but you would have to limit its depth so that you can see if the negamax parts are working correctly.  The tree is so small that without limiting the depth you are probably searching all the way to terminal nodes from the beginning.  Try limiting the depth to 3 or 4.

0

Share this post


Link to post
Share on other sites

Thanks for the insight. So basically I start the scoring by calling getWon() and return maxscore, -maxscore or 0 depending if the game is won, lost or drawn. If this is not the case, the evluation function will check of other combinations.

 

I will implement this in the connect4 I'm doing to make sure that negamax works fine before adding extensions to it.

0

Share this post


Link to post
Share on other sites

does anyone here knows how to measure the efficiency of negamax? sorry newbie here

 

What do you mean by "efficiency"? And why do you want to measure it?

0

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0