def alphabeta(self, rnd: PlayerRoundCheating, last_card: int,
alpha: (int, int), beta: (int, int),
maximizingPlayer: bool) -> Move:
"""
Returns the best card to play, based on the minimax algorithm
Args:
rnd: For the first call this is the current round. For further calls inside the recursion this ist the round that is being simulated
last_card: the last Card that has been played
maximizingPlayer: if you need to maximize or minimize
Returns:
the best card to play, int encoded
:param maximizingPlayer:
:param last_card:
:param rnd:
:param beta:
:param alpha:
"""
if rnd.nr_cards_in_trick > 3:
rule = RuleSchieber()
points = rule.calc_points(rnd.current_trick,
rnd.nr_played_cards == 36, rnd.trump)
winner = rule.calc_winner(rnd.current_trick,
rnd.trick_first_player[rnd.nr_tricks],
rnd.trump)
if winner == NORTH or winner == SOUTH:
bestCard = Move(last_card, points)
else:
bestCard = Move(last_card, -points)
return bestCard
elif maximizingPlayer:
cardsToPlay = np.flatnonzero(rnd.get_valid_cards())
for card in cardsToPlay:
newRound = self.fake_move(rnd, card)
alpha = max(alpha,
self.alphabeta(newRound, card, alpha, beta, False))
if beta <= alpha:
break
return alpha
else:
cardsToPlay = np.flatnonzero(rnd.get_valid_cards())
for card in cardsToPlay:
newRound = self.fake_move(rnd, card)
beta = min(beta,
self.alphabeta(newRound, card, alpha, beta, True))
if beta <= alpha:
break
return beta
return bestCard
def expandNode(self, node: Node, rnd: PlayerRoundCheating):
validCards = np.flatnonzero(rnd.get_valid_cards())
for card in validCards:
newNode = Node()
newNode.setParent(node)
newNode.getAction().setRound(rnd)
newNode.getAction().setPlayerNr(node.getAction().getRound().player)
newNode.getAction().setCard(card)
node.addChild(newNode)
def play_card(self, rnd: PlayerRoundCheating) -> int:
"""
Player returns a card to play based on the given round information.
Args:
rnd: current round
Returns:
card to play, int encoded
"""
playerRound = PlayerRoundCheating()
playerRound.set_from_round(rnd)
cardToPlay = self.alphabeta(playerRound, 0, Move(0, -sys.maxsize),
Move(0, sys.maxsize), True)
return cardToPlay.get_card()
def __expand_node(self, node: Node, round: PlayerRoundCheating):
valid_cards = np.flatnonzero(round.get_valid_cards())
for card in valid_cards:
newNode = Node()
newNode.set_parent(node)
newNode.getNodeMCTSInformation().setRound(round)
newNode.getNodeMCTSInformation().setPlayerNr(node.getNodeMCTSInformation().getRound().player)
newNode.getNodeMCTSInformation().setCard(card)
node.addChild(newNode)
def fake_move(self, rnd: PlayerRoundCheating,
cardToPlay: int) -> PlayerRoundCheating:
nextRound = PlayerRoundCheating()
nextRound.set_from_round(rnd)
nextRound.current_trick[rnd.nr_cards_in_trick] = cardToPlay
nextRound.hands[nextRound.player, cardToPlay] = 0
nextRound.nr_cards_in_trick = nextRound.nr_cards_in_trick + 1
nextRound.nr_played_cards = nextRound.nr_played_cards + 1
nextRound.player = (nextRound.player - 1) % 4
nextRound.hand = nextRound.hands[nextRound.player]
return nextRound
def simulateRound(self, node: Node):
rnd = get_round_from_player_round(node.getAction().getRound(),
node.getAction().getRound().hands)
rnd.action_play_card(node.getAction().getCard())
cards = rnd.nr_played_cards
randomPlayer = RandomPlayerSchieber()
while cards < 36:
player_rnd = PlayerRoundCheating()
player_rnd.set_from_round(rnd)
card_action = randomPlayer.play_card(player_rnd)
rnd.action_play_card(card_action)
cards += 1
myPoints = rnd.points_team_0
pointsEnemy = rnd.points_team_1
maxPoints = myPoints + pointsEnemy
if myPoints > pointsEnemy:
return (myPoints - 0) / (maxPoints - 0)
else:
return 0
def __simulate_round(self, node: Node): # Random walk
round = get_round_from_player_round(node.getNodeMCTSInformation().getRound(),
node.getNodeMCTSInformation().getRound().hands)
round.action_play_card(node.getNodeMCTSInformation().getCard())
cards = round.nr_played_cards
random_player = MyPlayer()
while cards < 36:
player_round = PlayerRoundCheating()
player_round.set_from_round(round)
card_action = random_player.play_card(player_round)
round.action_play_card(card_action)
cards += 1
myPoints = round.points_team_0
pointsEnemy = round.points_team_1
maxPoints = myPoints + pointsEnemy
if myPoints > pointsEnemy:
return (myPoints - 0) / (maxPoints - 0)
else:
return 0