def test_reward_lose(self):
m.I.combo = 5
m.I.combo_max = 99
m.I.points = 5
m.reward()
self.assertEqual(m.I.combo, 0)
self.assertEqual(m.I.combo_max, 99)
self.assertEqual(m.I.points, 5)
def play(model):
# parameters
gamma = 0.9
board = np.zeros((N, N), dtype=np.int8)
turn = True
while True:
# change black and white
if not turn:
board = -board
actions = np.array(np.where(board == 0))
r = input()
r = map(int, r.split(' '))
board[next(r), next(r)] = 1
board = -board
main.dispBoard(board)
if main.winning(-board):
return
turn = not turn
# can move
actions = np.array(np.where(board == 0))
# as feature vectors
# features = main.getFeatures(board, actions)
# set algorithm here.
# r = np.argmax(model.get(features)[:, 0])
r = main.getMove(board, model, True, 2)
action = actions[:, r]
Reward = main.reward(board, action)
# put
board[action[0], action[1]] = 1
# restore black and white
if not turn:
board = -board
main.dispBoard(board)
# all masses are filled, win
if Reward != 0:
return
# end of this turn
turn = not turn
def com_turn(image, board, points):
actions = np.array(np.where(board == 0))
# r = np.argmax(model.get(features)[:, 0])
r = main.getMove(board, model, True, depth=2)
action = actions[:, r]
Reward = main.reward(board, action)
# put
board[action[0], action[1]] = 1
# all masses are filled, win
p = action * 70 + 35
p = p.tolist()
p.reverse()
cv2.circle(image, center=tuple(p), radius=20, color=255, thickness=-1)
return (board, Reward)
def test_reward_gain_vlv6(self):
m.I.lvl = 6
m.reward(1)
self.assertEqual(m.I.points, 15)
self.assertEqual(m.I.time_left, 15)
def test_reward_gain_vlv5(self):
m.I.lvl = 5
m.reward(1)
self.assertEqual(m.I.points, 10)
self.assertEqual(m.I.time_left, 13)
def test_reward_gain_vlv4(self):
m.I.lvl = 4
m.reward(1)
self.assertEqual(m.I.points, 5)
self.assertEqual(m.I.time_left, 12)
def test_reward_gain_vlv0(self):
m.reward(1)
self.assertEqual(m.I.points, 1)
self.assertEqual(m.I.time_left, 11)
def test_reward_gain_combo_max(self):
m.I.combo = 40
m.reward(1)
self.assertEqual(m.I.combo, 40)
self.assertEqual(m.I.lvl, 8)
def test_reward_gain_vlv8(self):
m.I.lvl = 8
m.reward(1)
self.assertEqual(m.I.points, 30)
self.assertEqual(m.I.time_left, 18)
