/// <summary>
/// Asks the game engine next (valid) move given a game position
/// The board assumes following standard move notation:
/// v
/// A B C D E F G H I
/// [0 1 2 3 4 5 6 7 8] (first index)
/// 1[0]
/// 2[1]
/// 3[2] w K
/// 4[3] K w
/// 5[4]
/// 6[5]
/// >7[6] x
///
/// {Column;Line}
/// E.g.: 'w' on D3 in game notation will map to {3,2}, and 'x' on I7 to {8,6}
/// </summary>
/// <param name="game">a 2D board with integer values: 0 for white 1 for black and -1 for empty tiles. First index for the column, second index for the line</param>
/// <param name="level">an integer value to set the level of the IA, 5 normally</param>
/// <param name="whiteTurn">true if white players turn, false otherwise</param>
/// <returns>The column and line indices. Will return {-1,-1} as PASS if no possible move </returns>
public Tuple <int, int> GetNextMove(int[,] game, int level, bool whiteTurn)
{
int[,] gameState = new int[9, 7];
for (int i = 0; i < 9; i++)
{
for (int j = 0; j < 7; j++)
{
gameState[i, j] = game[i, j];
}
}
BoardIA logic = new BoardIA
{
GameBoard = gameState,
PlayerTurn = whiteTurn ? Player.White : Player.Black
};
List <IntPosition> possibleMoves = logic.GetAllPossibleMoves();
if (possibleMoves.Count == 0)
{
return(new Tuple <int, int>(-1, -1));
}
var nextMove = AlphaBeta(logic, level, int.MinValue, true); // Always minimum value (int.MinValue) and we maximize (true), whatever the color...
return(new Tuple <int, int>(nextMove.Item2.Column, nextMove.Item2.Row));
}
/// <summary>
/// Implementation of negamax algorithm with alpha beta pruning. A node is represented by a game board.
/// </summary>
/// <param name="nodeBoard">A particular game state</param>
/// <param name="depth">The maximum tree depth</param>
/// <param name="parentValue">The parent node value</param>
/// <param name="maximizingPlayer">Whether node is a maximizimer or minimizer. True if maximizer. </param>
/// <returns>A tuple containing the value as it's first item and the position corresponding to the best
/// predicted play as it's second item. </returns>
private Tuple <int, IntPosition> AlphaBeta(BoardIA nodeBoard, int depth, int parentValue, bool maximizingPlayer)
{
if (depth == 0 || nodeBoard.IsTerminal())
{
var heuristicValue = nodeBoard.GetHeuristicValue();
return(new Tuple <int, IntPosition>(heuristicValue, new IntPosition(-1, -1)));
}
else
{
int bestValue = maximizingPlayer ? -int.MaxValue : int.MaxValue;
IntPosition bestMove = new IntPosition(-1, -1);
var childPositions = nodeBoard.GetAllPossibleMoves();
foreach (var child in childPositions)
{
var childValue = AlphaBeta(PosToBoard(child, nodeBoard), depth - 1, bestValue, !maximizingPlayer);
int minOrMax = maximizingPlayer ? 1 : -1;
if (childValue.Item1 * minOrMax > bestValue * minOrMax)
{
bestValue = childValue.Item1;
bestMove = child;
if (bestValue * minOrMax > parentValue * minOrMax)
{
break;
}
}
}
return(new Tuple <int, IntPosition>(bestValue, bestMove));
}
}
/// <summary>
/// Creates an AIBoard class given the position of a move to play
/// and the board where the move has to be played.
/// </summary>
/// <param name="position">The position of the pawn to play on the board</param>
/// <param name="sourceBoard">The board where the move takes place</param>
/// <returns>A new AIBoard with the updated game state</returns>
private BoardIA PosToBoard(IntPosition position, BoardIA sourceBoard)
{
int[,] gameState = new int[9, 7];
for (int i = 0; i < 9; i++)
{
for (int j = 0; j < 7; j++)
{
gameState[i, j] = sourceBoard.GameBoard[i, j];
}
}
BoardIA newBoard = new BoardIA
{
GameBoard = gameState,
PlayerTurn = sourceBoard.PlayerTurn
};
newBoard.PlayMove(position.Column, position.Row, sourceBoard.PlayerTurn == Player.White);
newBoard.SwitchPlayer();
return(newBoard);
}