def _move_selected(state, button, move):
"""Interface callback for proceeding through game play with user"""
# Guard clause for when already-filled place is attempted
if button.get_label() in [X, O]:
return interface.set_invalid_move_state(state)
# Update board state based on move
human_id = board.not_id(state.agent_state.identifier)
state.board_state = board.place(state.board_state, human_id, move)
# Update UI to reflect model
interface.update_board(state)
# Check for game finished
if board.is_finished(state.board_state):
# Display game results
interface.game_finished(state)
else:
# Process agent's decision
(move, agent_state) = agent.move(state.agent_state, state.board_state)
# Update board and agent state
state.board_state = board.place(state.board_state,
state.agent_state.identifier, move)
state.agent_state = agent_state
# Update UI to reflect model
interface.update_board(state)
# Another check for game finish to prevent inadvertent user events
if board.is_finished(state.board_state):
# Display game results
interface.game_finished(state)
def _reward_for_move(state, board_state, move):
"""Helper function for determining reward from suggested move"""
return Pipe(
# Create new board for looking up reward
board.place(board_state, state.identifier, move),
# Lookup reward (or default)
Then2(rewards.lookupR, state.rewards_state, state.identifier))
def _play_train_move(player, board_state):
"""Helper for handling single agent's move"""
# Get move and updated player state
(move, player) = agent.move(player, board_state)
# Update board state
board_state = board.place(board_state, player.identifier, move)
return (board_state, player)
def reward_matrix(state, board_state):
"""Matrix of the reward value of all available moves"""
return reduce(
# Accumulate board with reward for each move
lambda acc, m: board.place(
acc, "{0:.3f}".format(_reward_for_move(state, board_state, m)), m),
# Get available moves
board_move.available(board_state),
# Use board as initial state
board_state).value
def place_pieces(self, board: board):
#gets dictionary of worker(key) and where it is being placed(value)
worker_psns = self.strategy.place_init_workers(board)
for worker, psn in worker_psns:
board.place(worker, psn)
return board
def _move_and_agent(state, board_state, move):
"""Helper function for returning move and updated agent state"""
last_board_state = board.place(board_state, state.identifier, move)
return (move, state.set(last_board_state=last_board_state))