def calculate_probabilities_for_player(
board: Board,
player_state: PlayerState,
depth: int,
step_offset: int = 1,
probability_cutoff: float = 0,
global_probability_factor: float = 1.
) -> Tuple[np.ndarray, np.ndarray]:
"""
Returns tuple of numpy arrays:
- probability of reaching the given cells in the [depth] next steps
- minimum amount of steps needed for the given player to reach the cell
"""
probabilities = np.zeros((board.height, board.width))
min_player_steps = np.ones(
(board.height, board.width)) * __INIT_VALUE_PLAYER_STEPS
valid_player_state_tuples = []
for action in PlayerAction:
new_player_state = player_state.copy()
new_player_state.do_action(action)
next_player_state = new_player_state.do_move()
if next_player_state.verify_move(board):
valid_player_state_tuples.append(
(new_player_state, next_player_state))
possible_action_count = len(valid_player_state_tuples)
if possible_action_count > 0:
local_probability_factor = (
1 / possible_action_count) * global_probability_factor
for new_player_state, next_player_state in valid_player_state_tuples:
affected_cells = next_player_state.steps_to_this_point
for cell_x, cell_y in affected_cells:
if board.point_is_on_board(cell_x, cell_y):
probabilities[cell_y, cell_x] += local_probability_factor
min_player_steps[cell_y, cell_x] = step_offset
# print(f"{local_probability_factor}\t{probability_cutoff}")
if depth > 1 and local_probability_factor > probability_cutoff:
recursion_probabilities, recursion_min_player_steps = \
calculate_probabilities_for_player(board, next_player_state,
depth=depth - 1,
step_offset=step_offset + 1,
probability_cutoff=probability_cutoff,
global_probability_factor=local_probability_factor)
probabilities += recursion_probabilities
min_player_steps = np.minimum(min_player_steps,
recursion_min_player_steps)
return probabilities, min_player_steps
def verify_state(self, board: Board) -> bool:
# check for speed conditions and if the new position is on the board
return self.verify_speed() and board.point_is_on_board(
self.position_x, self.position_y)