def test_player_move():
board = g.Board()
player = g.Player(board, 1)
player.epsilon = 1
board.set_state([2, 1, 2, 1, 2, 1, 2, 0, 2])
assert player.move() == 7
player = g.Player(board, 2)
player.epsilon = 1
board.set_state([2, 1, 2, 1, 2, 1, 2, 0, 2])
assert player.move() == 7
def test_player_move():
board = g.Board()
player = g.Player(board, 1)
board.set_state([1, 0, 1, 0, 2, 0, 0, 0, 0])
player.set_Q(1, 10)
player.epsilon = 0
assert player.move() == 1
def test_player_rewards():
board = g.Board()
player = g.Player(board, 1)
assert player.reward(3) == 0
board.set_state([1, 1, 1, 0, 0, 0, 0, 0, 0])
assert player.reward(3) == 100
board.set_state([2, 2, 2, 0, 0, 0, 0, 0, 0])
assert player.reward(3) == -100
def test_player_bestmove():
board = g.Board()
player = g.Player(board, 1)
board.set_state([2, 1, 2, 1, 2, 1, 2, 0, 2])
player.set_Q(7, 10)
assert np.array_equal(player.eval_board(),
[0., 0., 0., 0., 0., 0., 0., 10., 0.])
board.set_state([0, 1, 2, 1, 2, 1, 2, 0, 2])
player.set_Q(0, 20)
player.set_Q(7, 10)
assert np.array_equal(player.eval_board(),
[20., 0., 0., 0., 0., 0., 0., 10., 0.])
def test_player_Q():
board = g.Board()
player = g.Player(board, 1)
player.set_Q(1, 56)
assert player.get_Q(1) == 56
import deeptactoe
from deeptactoe import game as g
from pprint import pprint as pp
import numpy as np
board = g.Board()
player1 = g.Player(board, 1)
player2 = g.Player(board, 2, beh='human')
player1.epsilon = 1
try:
player1.load_Q()
except:
print("Not possible to load Q matrix. Computer will play at random")
def sanitise(board, move):
if (np.any(np.equal(board.get_empty(), int(move)))):
return False
else:
print("Illegal move!")
return True
def resolution(board):
case = board.get_condition()
if (case == 1):
print("The Computer wins")
elif (case == 2):
print("Human wins")
else:
print("Tie")
import deeptactoe
from deeptactoe import game as g
from pprint import pprint as pp
import matplotlib
matplotlib.use('GTK3Cairo')
import matplotlib.pyplot as plt
import numpy as np
board = g.Board()
player1 = g.Player(board, 1)
player2 = g.Player(board, 2, beh='fixed')
player1.epsilon = 1
board.start_game()
wins = []
n_ep = 10000
for episode in range(n_ep):
board.clear()
for i in range(6):
action = player1.move()
player1.update_Q(action)
player1.update_epsilon(episode)
board.update(action, 1)
if (board.get_condition() < 4): break
action = player2.move()
board.update(action, 2)