The 2048 Puzzle AI Player is an intelligent agent designed to play the 2048 puzzle efficiently, utilizing advanced AI techniques like Expectiminimax and heuristics to achieve high scores. This ReadMe provides an overview of the project, including its purpose, algorithmic approach, and usage instructions
def getMove(self, grid):
best_value = maxsize * -1
for move in grid.getAvailableMoves():
temp_grid = grid.clone()
temp_grid.move(move)
value = self.min(Node(move=move, grid=temp_grid, depth=DEPTH-1))
if value > best_value:
best_move = movedef min(self, node, alpha, beta):
for child in children:
value = self.max(child, alpha, beta)
if value < beta:
beta = value
if alpha >= beta:
break
return betadef evaluate(grid):
number_of_blank_tiles = len(grid.getAvailableCells())
max_tile = grid.getMaxTile()
bonus = 10 if max_tile in [grid.map[0][0], grid.map[0][3]] else 0
return number_of_blank_tiles + bonuspython GameManager.py- PlayerAI_3.py: Implements AI logic with Expectiminimax and heuristics.
- ComputerAI_3.py: Controls computer moves by inserting tiles randomly.
- Grid_3.py: Handles all operations related to the game board (e.g., tile movement, merging).
- Displayer_3.py: Manages grid display.
- GameManager_3.py: Manages game flow, alternating between AI and computer turns.
The AI explores all possible game moves using the Expectiminimax algorithm. Each state is evaluated using a weighted function to prioritize empty tiles, grid monotonicity, and corner positions for high-value tiles.
def evaluate(grid):
# Return a weighted sum of the heuristic factors
return len(grid.getAvailableCells()) + monotonicity_score + corner_bonus