def pazaak():
b = PazaakBoard()
player_side_deck = [
Card(x) for x in [choice(range(1, 7)) for _ in range(4)]
]
opponent_side_deck = [
Card(x) for x in [choice(range(1, 7)) for _ in range(4)]
]
players = [
PazaakPlayer(player=1, side_deck=player_side_deck),
PazaakPlayer(player=2, side_deck=opponent_side_deck)
]
state = PazaakState(board=b, players=players, player=players[0])
while b.status(players=state.players) == -1:
if not state.player.stand:
state = state.random_card()
node = Node(state=state)
tree = Mcts(root=node)
print(">>>>> CURR PLAYER: <<<<<<<", state.player.player)
b = tree.find_next_move(100)
state = PazaakState(board=b,
player=state.player,
players=state.players,
player_index=state.player_index)
print("TURN\n")
b.print()
def get_next_board():
global app_state
if app_state.board.status(players=app_state.players) == -1:
if not app_state.player.stand:
app_state = app_state.random_card()
node = Node(state=app_state)
tree = Mcts(root=node)
app_state.board = tree.find_next_move(100)
app_state = PazaakState(board=app_state.board,
player=app_state.player,
players=app_state.players,
player_index=app_state.player_index)
return app_state.board
return app_state.board
def tic_tac_toe():
b = TicTacBoard()
players = [TicTacPlayer(1), TicTacPlayer(2)]
state = TicTacState(board=b, player=players[0], players=players)
while b.status() == -1:
node = Node(state=state)
tree = Mcts(root=node)
print(">>>>> CURR PLAYER: <<<<<<<", state.player.player)
b = tree.find_next_move(100)
state = TicTacState(board=b,
player=state.player,
players=state.players,
player_index=state.player_index)
print("TURN\n")
b.print()
4: LazyAgent
}
print("Welcome in Pong")
selected_opponent, = input(
"Select opponent for MCTS (1 - Random, 2 - Safe, 3 - Aggressive, 4 - Lazy): "
).split()
game = PongGame()
game = PongMonitor(game, ".", force=True)
game.reset()
opponent = possible_opponents[int(selected_opponent)]()
mcts_agent = GreedyAgent()
tree = Mcts(game, simulation_agent=mcts_agent)
# tree = Mcts(game)
count = 0
while not game.done:
count = count + 1
start = time()
tree.run(30, verbose=True)
stop = time()
ob = game._get_obs()
# if ob is not None:
# game.ale.saveScreenPNG('images/' + str(count) + '-state.png')
# print(count, end=" ")
# for i, val in enumerate(ob):
# print(val, end=" ")
filename = './logs-defence/' + playout['method'] + '-' + \
('no-skip' if playout['exploration_parameter'] is False else 'with-skip') + \
'-' + str(playout['exploration_parameter']) + \
'/pong-' + playout['method'] + '-' + str(playout['runs']) + '-against-' + opponent_names[
playout['agent']] + '_' + datetime.now().strftime("%Y%m%d-%H%M%S")
print(filename)
game = PongMonitor(game, filename, force=False)
game.reset()
pong_logger = PDLogger(filename)
opponent = possible_opponents[agent]()
mcts_agent = GreedyAgent()
tree = None
if playout['method'] == 'greedy':
tree = Mcts(game, simulation_agent=mcts_agent, logger=pong_logger, skip_actions=playout['skip_actions'],
exploration_parameter=playout['exploration_parameter'])
if playout['method'] == 'random':
tree = Mcts(game, logger=pong_logger, skip_actions=playout['skip_actions'],
exploration_parameter=playout['exploration_parameter'])
count = 0
while not game.done:
count = count + 1
start = time()
tree.run(playout['runs'], verbose=True)
stop = time()
ob = game._get_obs()
# if ob is not None:
# game.ale.saveScreenPNG('images/' + str(count) + '-state.png')
file_path = './logs-nim/' + str(playout['runs']) + \
'-' + str(playout['opponent']) + \
'-' + str(playout['exploration_parameter'])
filename = file_path + '/nim-' + str(playout['piles']) + '-' + str(
playout['objects']) + '-' + str(playout['runs']) + '_vs_' + str(
playout['opponent']) + '_' + datetime.now().strftime(
"%Y%m%d-%H%M%S")
print(filename)
output_dir = Path(file_path)
output_dir.mkdir(parents=True, exist_ok=True)
game = Nim(int(piles), int(objects))
nim_logger = PDLogger(filename)
tree = Mcts(game,
logger=nim_logger,
exploration_parameter=playout['exploration_parameter'])
tree.run(1)
count = 0
winner = 0
while not game.done:
count = count + 1
tree.run(playout['runs'])
action = tree.predict()
game.act(action)
tree.move_root(action)
if game.done:
print("You won!")
winner = 1
"Set game settings (`number of piles` `number of objects`): ").split()
game = Nim(int(piles), int(objects))
QL = QLearn(game)
state_copy = game.get_state()
QL.train(ExpertAgent())
print_rewards(QL.reward_all_ep, 100000)
print()
count = 0
for i in range(10):
print('Try yourself against QL :)')
game.set_state(state_copy, False, 0)
tree = Mcts(game, exploration_parameter=1.41)
tree.run(1)
winner = 0
while not game.done:
print(game.piles)
action = QL.select_move(game)
print('CPU 0 move: %s' % str(action))
game.act(action)
tree.move_root(action)
if game.done:
winner = 1
break
from mcts.mcts import Mcts
from pong.pong_game import PongGame
from time import time
from pong.gym_agents import *
from pong.monitor import PongMonitor
from ddqn.ddqn_agent import DdqnAgent, dqn_heuristic
game = PongGame()
game = PongMonitor(game, ".", force=True)
game.reset()
mcts_agent = GreedyAgent()
tree = Mcts(game, simulation_agent=mcts_agent, heuristic=dqn_heuristic)
ddqn_agent = DdqnAgent()
while not game.done:
ob = game._get_obs()
action1 = ddqn_agent.act(ob)
game.act(action1)
tree.move_root(action1)
tree.run(5, verbose=True)
action2 = tree.predict()
game.act(action2)
tree.move_root(action2)
game.render()
from mcts.mcts import Mcts
from nim.nim import Nim
print("Hello in Nim")
piles, objects = input(
"Set game settings (`number of piles` `number of objects`): ").split()
game = Nim(int(piles), int(objects))
tree = Mcts(game)
tree.run(1)
while not game.done:
print(game.piles)
# move = input("Your move (`pile` `objects`): ").split()
# action = tuple(int(x) for x in move)
tree.run(1200)
action = tree.predict()
print('CPU 0 move: %s' % str(action))
game.act(action)
tree.move_root(action)
if game.done:
print("You won!")
exit()
print(game.piles)
tree.run(1200)
action = tree.predict()