def test_generate_action_from_policy(self):
monte = mc.MonteCarlo()
state = (10, 10)
action = monte._generate_action_on_state(state)
self.assertTrue(action is 0 or action is 1)
def test_generate_episode_checkList(self):
monte = mc.MonteCarlo()
action_list = monte._generate_episode()
(state, action, reward) = action_list[0]
self.assertIn(reward, [-1, 0, 1])
self.assertGreaterEqual(state[0], 0)
self.assertGreaterEqual(state[1], 0)
self.assertTrue(action is 0 or action is 1)
def test_monteCarloEval(selfs):
monte = mc.MonteCarlo()
monte.monte_carlo_control(10000)
monte.test_strategy(400)
def test_update_from_episode(self):
monte = mc.MonteCarlo()
episode = monte._generate_episode()
monte._update_from_episode(episode)
def test_generate_episode_List_notEmpty(self):
monte = mc.MonteCarlo()
action_list = monte._generate_episode()
self.assertTrue(action_list)
def test_q_init(self):
monte = mc.MonteCarlo()
self.assertEquals(np.shape(monte._counter), (2, 10, 21))
self.assertEquals(np.shape(monte._value_action_state), (2, 10, 21))