public void TestMCTStree()
{
WeightVectorAll weights = new WeightVectorAll { Corner = 0, Empty_cells = 0, Highest_tile = 0, Monotonicity = 0, Points =0, Smoothness = 0, Snake = 1, Trapped_penalty = 0 };
int timeLimit = 100;
int[][] state1 = new int[][] {
new int[]{1024,16,0,0},
new int[]{4,32,2,0},
new int[]{64,16,0,0},
new int[]{16,16,2,2}
};
int[][] state2 = new int[][] {
new int[]{2,0,2,0},
new int[]{8,2,0,0},
new int[]{16,8,4,0},
new int[]{64,4,4,0}
};
int[][] state3 = new int[][] {
new int[]{16,16,16,4},
new int[]{64,4,0,0},
new int[]{8,0,2,0},
new int[]{16,0,0,0}
};
int[][] state4 = new int[][] {
new int[]{0,0,0,8},
new int[]{0,0,16,16},
new int[]{2,0,32,32},
new int[]{2,4,16,8}
};
Console.WriteLine("Testing state1:");
GameEngine gameEngine = new GameEngine();
Minimax minimax = new Minimax(gameEngine, 0);
Expectimax expectimax = new Expectimax(gameEngine, 0);
MonteCarlo mcts = new MonteCarlo(gameEngine);
Move minimaxMove = minimax.IterativeDeepening(new State(state1, CalculateScore(state1), GameEngine.PLAYER), timeLimit);
Move expectimaxMove = expectimax.IterativeDeepening(new State(state1, CalculateScore(state1), GameEngine.PLAYER), timeLimit, weights);
Move mctsMove = (mcts.TimeLimitedMCTS(new State(state1, CalculateScore(state1), GameEngine.PLAYER), timeLimit)).GeneratingMove;
Console.WriteLine("Minimax move chosen: " + ((PlayerMove)minimaxMove).Direction);
Console.WriteLine("Expectimax move chosen: " + ((PlayerMove)expectimaxMove).Direction);
Console.WriteLine("MCTS move chosen: " + ((PlayerMove)mctsMove).Direction);
Console.WriteLine("Testing state2:");
minimaxMove = minimax.IterativeDeepening(new State(state2, CalculateScore(state2), GameEngine.PLAYER), timeLimit);
expectimaxMove = expectimax.IterativeDeepening(new State(state2, CalculateScore(state2), GameEngine.PLAYER), timeLimit, weights);
mctsMove = (mcts.TimeLimitedMCTS(new State(state2, CalculateScore(state2), GameEngine.PLAYER), timeLimit)).GeneratingMove;
Console.WriteLine("Minimax move chosen: " + ((PlayerMove)minimaxMove).Direction);
Console.WriteLine("Expectimax move chosen: " + ((PlayerMove)expectimaxMove).Direction);
Console.WriteLine("MCTS move chosen: " + ((PlayerMove)mctsMove).Direction);
Console.WriteLine("Testing state3:");
minimaxMove = minimax.IterativeDeepening(new State(state3, CalculateScore(state3), GameEngine.PLAYER), timeLimit);
expectimaxMove = expectimax.IterativeDeepening(new State(state3, CalculateScore(state3), GameEngine.PLAYER), timeLimit, weights);
mctsMove = (mcts.TimeLimitedMCTS(new State(state3, CalculateScore(state3), GameEngine.PLAYER), timeLimit)).GeneratingMove;
Console.WriteLine("Minimax move chosen: " + ((PlayerMove)minimaxMove).Direction);
Console.WriteLine("Expectimax move chosen: " + ((PlayerMove)expectimaxMove).Direction);
Console.WriteLine("MCTS move chosen: " + ((PlayerMove)mctsMove).Direction);
Console.WriteLine("Testing state4:");
minimaxMove = minimax.IterativeDeepening(new State(state4, CalculateScore(state4), GameEngine.PLAYER), timeLimit);
expectimaxMove = expectimax.IterativeDeepening(new State(state4, CalculateScore(state4), GameEngine.PLAYER), timeLimit, weights);
mctsMove = (mcts.TimeLimitedMCTS(new State(state4, CalculateScore(state4), GameEngine.PLAYER), timeLimit)).GeneratingMove;
Console.WriteLine("Minimax move chosen: " + ((PlayerMove)minimaxMove).Direction);
Console.WriteLine("Expectimax move chosen: " + ((PlayerMove)expectimaxMove).Direction);
Console.WriteLine("MCTS move chosen: " + ((PlayerMove)mctsMove).Direction);
}
// Runs an entire game using the given AI type to decide on moves
private static State RunAIGame(AI_TYPE AItype, bool print, int depth = 0, int timeLimit = 0, int iterationLimit = 0)
{
GameEngine game = new GameEngine();
State end = null;
if (AItype == AI_TYPE.CLASSIC_MINIMAX)
{
Minimax minimax = new Minimax(game, depth);
end = minimax.RunClassicMinimax(print);
}
else if (AItype == AI_TYPE.ALPHA_BETA)
{
Minimax minimax = new Minimax(game, depth);
end = minimax.RunAlphaBeta(print);
}
else if (AItype == AI_TYPE.ITERATIVE_DEEPENING_ALPHA_BETA)
{
Minimax minimax = new Minimax(game, depth);
end = minimax.RunIterativeDeepeningAlphaBeta(print, timeLimit);
}
else if (AItype == AI_TYPE.PARALLEL_ALPHA_BETA)
{
Minimax minimax = new Minimax(game, depth);
end = minimax.RunParallelAlphaBeta(print);
}
else if (AItype == AI_TYPE.PARALLEL_ITERATIVE_DEEPENING_ALPHA_BETA)
{
Minimax minimax = new Minimax(game, depth);
end = minimax.RunParallelIterativeDeepeningAlphaBeta(print, timeLimit);
}
else if (AItype == AI_TYPE.CLASSIC_EXPECTIMAX)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunClassicExpectimax(print, weights);
}
else if (AItype == AI_TYPE.EXPECTIMAX_STAR1)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunStar1Expectimax(print, weights);
}
else if (AItype == AI_TYPE.EXPECTIMAX_STAR1_FW_PRUNING)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunStar1WithUnlikelyPruning(print, weights);
}
else if (AItype == AI_TYPE.ITERATIVE_DEEPENING_EXPECTIMAX)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunIterativeDeepeningExpectimax(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.PARALLEL_EXPECTIMAX)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunParallelClassicExpectimax(print, weights);
}
else if (AItype == AI_TYPE.PARALLEL_ITERATIVE_DEEPENING_EXPECTIMAX)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunParallelIterativeDeepeningExpectimax(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.TT_ITERATIVE_DEEPENING_EXPECTIMAX)
{
Expectimax exptectimax = new Expectimax(game, depth);
end = exptectimax.RunTTExpectimax(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.TT_ITERATIVE_DEEPENING_STAR1)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunTTStar1(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.EXPECTIMAX_WITH_ALL_IMPROVEMENTS)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunTTIterativeDeepeningExpectimaxWithStar1andForwardPruning(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.EXPECTIMAX_WITH_ALL_IMPROVEMENTS_NO_FORWARDPRUNING)
{
Expectimax expectimax = new Expectimax(game, depth);
end = expectimax.RunTTIterativeDeepeningExpectimaxWithStar1(print, timeLimit, weights);
}
else if (AItype == AI_TYPE.ITERATION_LIMITED_MCTS)
{
MonteCarlo MCTS = new MonteCarlo(game);
end = MCTS.RunIterationLimitedMCTS(print, iterationLimit);
}
else if (AItype == AI_TYPE.TIME_LIMITED_MCTS)
{
MonteCarlo MCTS = new MonteCarlo(game);
end = MCTS.RunTimeLimitedMCTS(print, timeLimit);
}
else if (AItype == AI_TYPE.ROOT_PARALLEL_ITERATION_LIMITED_MCTS)
{
MonteCarlo MCTS = new MonteCarlo(game);
end = MCTS.RunRootParallelizationIterationLimitedMCTS(print, iterationLimit, NUM_THREADS);
}
else if (AItype == AI_TYPE.ROOT_PARALLEL_TIME_LIMITED_MCTS)
{
MonteCarlo MCTS = new MonteCarlo(game);
end = MCTS.RunRootParallelizationTimeLimitedMCTS(print, timeLimit, NUM_THREADS);
}
else if (AItype == AI_TYPE.EXPECTIMAX_MCTS_TIME_LIMITED)
{
HeuristicLearning HL = new HeuristicLearning(game);
end = HL.RunExpectimaxMCTStimeLimited(print, depth, timeLimit);
}
else if (AItype == AI_TYPE.EXPECTIMAX_MCTS_WITH_SIMULATIONS_TIME_LIMITED)
{
HeuristicLearning HL = new HeuristicLearning(game);
end = HL.RunExpectimaxMCTSwithSimulations(print, depth, timeLimit);
}
else if (AItype == AI_TYPE.FINAL_COMBI)
{
HeuristicLearning HL = new HeuristicLearning(game);
end = HL.RunParallelizationMCTSExpectimaxCombi(print, depth, timeLimit);
}
else
{
throw new Exception();
}
if (print)
{
Console.WriteLine("GAME OVER!\nFinal score: " + game.scoreController.getScore());
}
return end;
}
