After the player places a tile in a column, the function runs and returns zeroes (the AI’s identifier) for all rows except the first one and where the player has put a tile. I’ve also observed that if printing the depth, row and column, the depth alternates once but then doesn’t for a long while. Example: 2, 3, 5 then 1, 6, 7 then a long line of just depth 2.
The minmax algorithm:
def minimax(depth, board, maximizing_player):
# Terminal states. #
# Low score for player win. #
if depth == 0 or is_terminal(board):
if win_check(board, player_id):
return -9999
# High score for AI win. #
elif win_check(board, enemy_id):
return +9999
# There is a draw, which equals 0. #
elif draw_check(board, enemy_id) or draw_check(board, player_id):
return 0
else:
return board_check(board, enemy_id)
# Gets valid locations. #
valid_locations = get_valid_locations(board)
# Will switch back and forth between maximizing and minimizing players. #
# Maximising Function #
if maximizing_player:
best_value = -99999999
# Checks all the valid locations and returns the highest evaluated one. #
# Error. #
for row, column in valid_locations:
print(depth, row, column)
# Creates a copy of the board, places random piece in and evaluates. #
board_copy = board.copy()
board_copy[row][column] = enemy_id
new_value = minimax(depth - 1, board_copy, False)
if new_value > best_value:
best_value = new_value
current_column = column
return best_value
# Minimising Function #
else:
best_value = +99999999
# Checks all valid positions and finds the lowest rated one. #
for row, column in valid_locations:
# Creates a copy of the board, places random piece in and evaluates. #
board_copy = board.copy()
print(depth, row, column)
board_copy[row][column] = player_id
new_value = minimax(depth - 1, board_copy, True)
if new_value < best_value:
best_value = new_value
current_column = column
return best_value
My valid locations function creates a tuple of the row and column.