def random_card(self):
""" Returns a card uniformly at random.
(Called `randomCard` in the C++ version.)
"""
return util.choice([oJack, oQueen, oKing])
def perform_action(self, action):
""" Receives the agent's action and calculates the new environment percept.
(Called `performAction` in the C++ version.)
"""
assert self.is_valid_action(action)
# Save the action.
self.action = action
# Opponent plays rock if it won the last round by playing rock, otherwise
# it plays randomly.
if (self.observation == aRock) and (self.reward == rLose):
self.observation = aRock
else:
self.observation = util.choice(self.valid_actions)
# end if
# Determine reward.
if action == self.observation:
# If both the agent and the opponent made the same move, it's a draw.
self.reward = rDraw
elif action == aRock:
# If the opponent made a scissors move, then the agent wins if they played rock.
self.reward = rWin if self.observation == oScissors else rLose
elif action == aScissors:
# If the opponent made a paper move, then the agent wins if they played scissors.
self.reward = rWin if self.observation == oPaper else rLose
elif action == aPaper:
# If the opponent made a rock move, then the agent wins if they played paper.
self.reward = rWin if self.observation == oRock else rLose
# end if
# Return the resulting observation and reward.
return (self.observation, self.reward)
def random_card(self):
""" Returns a card uniformly at random.
(Called `randomCard` in the C++ version.)
"""
return util.choice([oJack, oQueen, oKing])
def perform_action(self, action):
""" Receives the agent's action and calculates the new environment percept.
(Called `performAction` in the C++ version.)
"""
assert self.is_valid_action(action)
# Save the action.
self.action = action
# Opponent plays rock if it won the last round by playing rock, otherwise
# it plays randomly.
if (self.observation == aRock) and (self.reward == rLose):
self.observation = aRock
else:
self.observation = util.choice(self.valid_actions)
# end if
# Determine reward.
if action == self.observation:
# If both the agent and the opponent made the same move, it's a draw.
self.reward = rDraw
elif action == aRock:
# If the opponent made a scissors move, then the agent wins if they played rock.
self.reward = rWin if self.observation == oScissors else rLose
elif action == aScissors:
# If the opponent made a paper move, then the agent wins if they played scissors.
self.reward = rWin if self.observation == oPaper else rLose
elif action == aPaper:
# If the opponent made a rock move, then the agent wins if they played paper.
self.reward = rWin if self.observation == oRock else rLose
# end if
# Return the resulting observation and reward.
return (self.observation, self.reward)
def teleport_agent(self):
""" Randomly places the agent at any `cTeleportTo` location.
(Called `teleportAgent` in the C++ version.)
"""
self.teleported = True
# This is altered from the C++ version to be far more efficient.
# (e.g. instead of random search of the maze, use a random choice
# over a pre-computed list of possible destinations.)
self.row, self.col = util.choice(self.teleport_to_locations)
def teleport_agent(self):
""" Randomly places the agent at any `cTeleportTo` location.
(Called `teleportAgent` in the C++ version.)
"""
self.teleported = True
# This is altered from the C++ version to be far more efficient.
# (e.g. instead of random search of the maze, use a random choice
# over a pre-computed list of possible destinations.)
self.row, self.col = util.choice(self.teleport_to_locations)
# end def
# end class
def generate_random_action(self):
""" Returns an action generated uniformly at random.
(Called `genRandomAction` in the C++ version.)
"""
return util.choice(self.environment.valid_actions)
def generate_random_action(self):
""" Returns an action generated uniformly at random.
"""
return util.choice(self.environment.valid_actions)
def generate_random_action(self):
""" Returns an action generated uniformly at random.
"""
return util.choice(self.generate_all_actions())
def generate_random_action(self):
""" Returns an action generated uniformly at random.
(Called `genRandomAction` in the C++ version.)
"""
return util.choice(self.environment.valid_actions)
