def _manually_create_game():
"""Creates the game manually from the spiel building blocks."""
game_type = pyspiel.GameType(
"matching_pennies",
"Matching Pennies",
pyspiel.GameType.Dynamics.SIMULTANEOUS,
pyspiel.GameType.ChanceMode.DETERMINISTIC,
pyspiel.GameType.Information.ONE_SHOT,
pyspiel.GameType.Utility.ZERO_SUM,
pyspiel.GameType.RewardModel.TERMINAL,
2, # max num players
2, # min_num_players
True, # provides_information_state
True, # provides_information_state_tensor
False, # provides_observation
False, # provides_observation_tensor
dict() # parameter_specification
)
game = pyspiel.MatrixGame(
game_type,
{}, # game_parameters
["Heads", "Tails"], # row_action_names
["Heads", "Tails"], # col_action_names
[[-1, 1], [1, -1]], # row player utilities
[[1, -1], [-1, 1]] # col player utilities
)
return game
def test_game_type(self):
game_type = pyspiel.GameType(
"matrix_mp", "Matching Pennies", pyspiel.GameType.Dynamics.SIMULTANEOUS,
pyspiel.GameType.ChanceMode.DETERMINISTIC,
pyspiel.GameType.Information.PERFECT_INFORMATION,
pyspiel.GameType.Utility.ZERO_SUM,
pyspiel.GameType.RewardModel.TERMINAL, 2, 2, True, True, False, False,
dict())
self.assertEqual(game_type.chance_mode,
pyspiel.GameType.ChanceMode.DETERMINISTIC)
def get_type(self):
return pyspiel.GameType(
short_name="python_tic_tac_toe",
long_name="Python Tic-Tac-Toe",
dynamics=pyspiel.GameType.Dynamics.SEQUENTIAL,
chance_mode=pyspiel.GameType.ChanceMode.DETERMINISTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.ZERO_SUM,
reward_model=pyspiel.GameType.RewardModel.TERMINAL,
max_num_players=_NUM_PLAYERS,
min_num_players=_NUM_PLAYERS,
provides_information_state_string=True,
provides_information_state_tensor=False,
provides_observation_string=True,
provides_observation_tensor=True,
parameter_specification={},
)
def get_type(self):
return pyspiel.GameType(
"python_tic_tac_toe",
"Python Tic-Tac-Toe",
pyspiel.GameType.Dynamics.SEQUENTIAL,
pyspiel.GameType.ChanceMode.DETERMINISTIC,
pyspiel.GameType.Information.PERFECT_INFORMATION,
pyspiel.GameType.Utility.ZERO_SUM,
pyspiel.GameType.RewardModel.TERMINAL,
2, # max num players
2, # min_num_players
True, # provides_information_state
False, # provides_information_state_tensor
True, # provides_observation
False, # provides_observation_tensor
{} # parameter_specification
)
def get_type(self):
type = self._client.send({'type': 'game_type'})
# see https://github.com/deepmind/open_spiel/blob/master/open_spiel/spiel.h
# GameType for the values of these fields
return pyspiel.GameType(
short_name = type['short_name'],
long_name = type['long_name'],
dynamics = pyspiel.GameType.Dynamics(type['dynamics']),
chance_mode = pyspiel.GameType.ChanceMode(type['chance_mode']),
information = pyspiel.GameType.Information(type['information']),
utility = pyspiel.GameType.Utility(type['utility']),
reward_model = pyspiel.GameType.RewardModel(type['reward_model']),
max_num_players = type['max_num_players'],
min_num_players = type['min_num_players'],
provides_information_state_string = type['provides_information_state_string'],
provides_information_state_tensor = type['provides_information_state_tensor'],
provides_observation_string = type['provides_observation_string'],
provides_observation_tensor = type['provides_observation_tensor'],
parameter_specification=type['parameter_specification'])
import numpy as np
import pyspiel
_NUM_PLAYERS = 2
_DEFAULT_PARAMS = {"termination_probability": 0.125}
_PAYOFF = [[5, 0], [10, 1]]
_GAME_TYPE = pyspiel.GameType(
short_name="python_iterated_prisoners_dilemma",
long_name="Python Iterated Prisoner's Dilemma",
dynamics=pyspiel.GameType.Dynamics.SIMULTANEOUS,
chance_mode=pyspiel.GameType.ChanceMode.EXPLICIT_STOCHASTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.REWARDS,
max_num_players=_NUM_PLAYERS,
min_num_players=_NUM_PLAYERS,
provides_information_state_string=False,
provides_information_state_tensor=False,
provides_observation_string=False,
provides_observation_tensor=False,
provides_factored_observation_string=False,
parameter_specification=_DEFAULT_PARAMS)
_GAME_INFO = pyspiel.GameInfo(num_distinct_actions=2,
max_chance_outcomes=2,
num_players=2,
min_utility=0.,
max_utility=np.inf,
utility_sum=0.0,
max_game_length=9999)
# The reward matrix is represented as a string containing a
# space-separated list of values.
# Its size defines the number of populations in the mean field game.
"reward_matrix":
" ".join(str(v) for v in _DEFAULT_REWARD_MATRIX.flatten()),
"geometry": _DEFAULT_GEOMETRY
}
_GAME_TYPE = pyspiel.GameType(
short_name="python_mfg_predator_prey",
long_name="Python Mean Field Predator Prey",
dynamics=pyspiel.GameType.Dynamics.MEAN_FIELD,
chance_mode=pyspiel.GameType.ChanceMode.EXPLICIT_STOCHASTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.REWARDS,
# We cannot pass math.inf here, so we pass a very high integer value.
max_num_players=1000000000,
min_num_players=1,
provides_information_state_string=True,
provides_information_state_tensor=False,
provides_observation_string=True,
provides_observation_tensor=True,
parameter_specification=_DEFAULT_PARAMS)
def get_param(param_name, params):
return params.get(param_name, _DEFAULT_PARAMS[param_name])
class MFGPredatorPreyGame(pyspiel.Game):
"""Predator-prey multi-population MFG."""
import pyspiel
_DEFAULT_PARAMS = {
"max_num_time_step": 10,
"time_step_length": 0.5,
"players": -1
}
_GAME_TYPE = pyspiel.GameType(
short_name="python_dynamic_routing",
long_name="Python Dynamic Routing Game",
dynamics=pyspiel.GameType.Dynamics.SIMULTANEOUS,
chance_mode=pyspiel.GameType.ChanceMode.DETERMINISTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.REWARDS,
max_num_players=100,
min_num_players=0,
provides_information_state_string=True,
provides_information_state_tensor=True,
provides_observation_string=True,
provides_observation_tensor=True,
default_loadable=True,
provides_factored_observation_string=True,
parameter_specification=_DEFAULT_PARAMS)
class DynamicRoutingGame(pyspiel.Game):
"""Implementation of dynamic routing game.
At each simultaneous-move time, each vehicle/player with negative waiting time
chooses on which successor link they would like to go. When arriving on the
import pyspiel
_DEFAULT_PARAMS = {
"max_num_time_step": 10,
"time_step_length": 0.5,
"players": -1
}
_GAME_TYPE = pyspiel.GameType(
short_name="python_mfg_dynamic_routing",
long_name="Python Mean Field Routing Game",
dynamics=pyspiel.GameType.Dynamics.MEAN_FIELD,
chance_mode=pyspiel.GameType.ChanceMode.EXPLICIT_STOCHASTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.REWARDS,
max_num_players=1,
min_num_players=1,
provides_information_state_string=True,
provides_information_state_tensor=True,
provides_observation_string=True,
provides_observation_tensor=True,
default_loadable=True,
provides_factored_observation_string=True,
parameter_specification=_DEFAULT_PARAMS)
WAITING_TIME_NOT_ASSIGNED = -1
class MeanFieldRoutingGame(pyspiel.Game):
"""Implementation of mean field routing game.
relies on processing and updating states as it goes, e.g. MCTS.
"""
import numpy as np
import pyspiel
_NUM_PLAYERS = 2
_GAME_TYPE = pyspiel.GameType(
short_name="python_kuhn_poker",
long_name="Python Kuhn Poker",
dynamics=pyspiel.GameType.Dynamics.SEQUENTIAL,
chance_mode=pyspiel.GameType.ChanceMode.EXPLICIT_STOCHASTIC,
information=pyspiel.GameType.Information.IMPERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.ZERO_SUM,
reward_model=pyspiel.GameType.RewardModel.TERMINAL,
max_num_players=_NUM_PLAYERS,
min_num_players=_NUM_PLAYERS,
provides_information_state_string=True,
provides_information_state_tensor=True,
provides_observation_string=True,
provides_observation_tensor=True,
provides_factored_observation_string=True)
_GAME_INFO = pyspiel.GameInfo(
num_distinct_actions=2, # 0 = Pass, 1 = Bet
max_chance_outcomes=3, # The deck has 3 cards
num_players=2,
min_utility=-2.0,
max_utility=2.0,
utility_sum=0.0,
max_game_length=3) # e.g. Pass, Bet, Bet
_NUM_PLAYERS = 1
_SIZE = 10
_HORIZON = 10
_NUM_ACTIONS = 3
_NUM_CHANCE = 3
_EPSILON = 10**(-25)
_DEFAULT_PARAMS = {"size": _SIZE, "horizon": _HORIZON}
_GAME_TYPE = pyspiel.GameType(
short_name="python_mfg_crowd_modelling",
long_name="Python Mean Field Crowd Modelling",
dynamics=pyspiel.GameType.Dynamics.MEAN_FIELD,
chance_mode=pyspiel.GameType.ChanceMode.EXPLICIT_STOCHASTIC,
information=pyspiel.GameType.Information.PERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.REWARDS,
max_num_players=_NUM_PLAYERS,
min_num_players=_NUM_PLAYERS,
provides_information_state_string=True,
provides_information_state_tensor=False,
provides_observation_string=True,
provides_observation_tensor=True,
parameter_specification=_DEFAULT_PARAMS)
class MFGCrowdModellingGame(pyspiel.Game):
"""A Mean Field Crowd Modelling game.
A game starts by an initial chance node that select the initial state
of the MFG.
#_DEFAULT_REGION_REWARDS = [(5,4,1),(7,4,2),(6,4,2),(4,2,1),(4,2,0),(6,3,1),(5,3,1),(4,3,1),(5,3,2)]
#_DEFAULT_PLAYERS = 4
#_DEFAULT_BOARD = 'BCDEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA'
#_DEFAULT_GRANDES = [1,1,1,1]
#_DEFAULT_KING = 1
_DEFAULT_STATE = '{"players":4,"rewards":{{5,4,1},{7,4,2},{6,4,2},{4,2,1},{4,2,0},{6,3,1},{5,3,1},{4,3,1},{5,3,2}},"board":"BBBBAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA","grandes":{1,1,1,1},"king":1}'
_GAME_TYPE = pyspiel.GameType(
short_name="castillo_game",
long_name="El Grande Castillo Game",
dynamics=pyspiel.GameType.Dynamics.SEQUENTIAL,
chance_mode=pyspiel.GameType.ChanceMode.DETERMINISTIC,
information=pyspiel.GameType.Information.IMPERFECT_INFORMATION,
utility=pyspiel.GameType.Utility.GENERAL_SUM,
reward_model=pyspiel.GameType.RewardModel.TERMINAL,
max_num_players=_MAX_PLAYERS,
min_num_players=2,
provides_information_state_string=False,
provides_information_state_tensor=True,
provides_observation_string=False,
provides_observation_tensor=False,
provides_factored_observation_string=False)
_GAME_INFO = pyspiel.GameInfo(
num_distinct_actions=_NUM_REGIONS,
max_chance_outcomes=0,
num_players=5,
min_utility=-10.0,
max_utility=10.0,
utility_sum=0.0,
max_game_length=_MAX_PLAYERS)
