import specmine.tools.write_ttt_optimal
if __name__ == "__main__":
specmine.script(specmine.tools.write_ttt_optimal.main)
import cPickle as pickle
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(
out_path=("path to write policy pickle", ),
player=("player number", "option", None, int),
)
def main(out_path, player=-1):
"""Compute the optimal TTT policy for the specified player."""
# construct domain
opponent_domain = specmine.rl.TicTacToeDomain(player=-1 * player)
opponent_policy = specmine.rl.RandomPolicy(opponent_domain)
domain = specmine.rl.TicTacToeDomain(player=player,
opponent=opponent_policy)
# compute the optimal policy
policy = {}
for state in domain.states:
(board, player_to_move) = state
import specmine.experiments.ttt_analyze_evs
if __name__ == "__main__":
specmine.script(specmine.experiments.ttt_analyze_evs.main)
import csv
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(
out_path=("path to write output CSV",), number=("number of eigenvectors to compute", "option", None, int)
)
def main(out_path, number=9):
"""Analyze eigenvectors in the TTT domain."""
B = number
adict = specmine.tictac.load_adjacency_dict()
(gameplay_NN, index) = specmine.discovery.adjacency_dict_to_matrix(adict)
basis_NB = specmine.spectral.laplacian_basis(gameplay_NN, B)
start = specmine.tictac.BoardState()
rows = []
for i in xrange(3):
for j in xrange(3):
board = start.make_move(1, i, j)
n = index[(board, -1)]
for b in xrange(B):
import specmine.tools.write_ttt_values
if __name__ == "__main__":
specmine.script(specmine.tools.write_ttt_values.main)
import cPickle as pickle
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(
out_path=("path to write values pickle", ),
opponent_path=("path to opponent policy; random otherwise", ))
def main(out_path, opponent_path=None):
"""Compute the TTT value function using value iteration."""
# construct domain
opponent_domain = specmine.rl.TicTacToeDomain(player=-1)
if opponent_path is None:
opponent_policy = specmine.rl.RandomPolicy(opponent_domain)
else:
logger.info("loading opponent policy from %s", opponent_path)
with specmine.openz(opponent_path) as opponent_file:
opponent_policy = pickle.load(opponent_file)
domain = specmine.rl.TicTacToeDomain(player=1, opponent=opponent_policy)
import specmine.tools.write_ttt_values
if __name__ == "__main__":
specmine.script(specmine.tools.write_ttt_values.main)
import cPickle as pickle
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(
out_path = ("path to write values pickle",),
opponent_path = ("path to opponent policy; random otherwise",)
)
def main(out_path, opponent_path = None):
"""Compute the TTT value function using value iteration."""
# construct domain
opponent_domain = specmine.rl.TicTacToeDomain(player = -1)
if opponent_path is None:
opponent_policy = specmine.rl.RandomPolicy(opponent_domain)
else:
logger.info("loading opponent policy from %s", opponent_path)
with specmine.openz(opponent_path) as opponent_file:
opponent_policy = pickle.load(opponent_file)
domain = specmine.rl.TicTacToeDomain(player = 1, opponent = opponent_policy)
# compute the value function
import specmine.experiments.ttt_analyze_evs
if __name__ == "__main__":
specmine.script(specmine.experiments.ttt_analyze_evs.main)
import csv
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(
out_path=("path to write output CSV", ),
number=("number of eigenvectors to compute", "option", None, int),
)
def main(out_path, number=9):
"""Analyze eigenvectors in the TTT domain."""
B = number
adict = specmine.tictac.load_adjacency_dict()
(gameplay_NN, index) = specmine.discovery.adjacency_dict_to_matrix(adict)
basis_NB = specmine.spectral.laplacian_basis(gameplay_NN, B)
start = specmine.tictac.BoardState()
rows = []
for i in xrange(3):
for j in xrange(3):
board = start.make_move(1, i, j)
n = index[(board, -1)]
import specmine.tools.write_ttt_optimal
if __name__ == "__main__":
specmine.script(specmine.tools.write_ttt_optimal.main)
import cPickle as pickle
import specmine
logger = specmine.get_logger(__name__)
specmine.annotations(out_path=("path to write policy pickle",), player=("player number", "option", None, int))
def main(out_path, player=-1):
"""Compute the optimal TTT policy for the specified player."""
# construct domain
opponent_domain = specmine.rl.TicTacToeDomain(player=-1 * player)
opponent_policy = specmine.rl.RandomPolicy(opponent_domain)
domain = specmine.rl.TicTacToeDomain(player=player, opponent=opponent_policy)
# compute the optimal policy
policy = {}
for state in domain.states:
(board, player_to_move) = state
if player == player_to_move:
(move_i, move_j, _) = specmine.tictac.ab_optimal_move(board, player)
