def test_transitionMatrix(self):
P = BairdCounterexample.constructTransitionMatrix(targetPolicy)
E = np.zeros((7, 7))
E[:, 6] = 1
self.assertTrue(np.allclose(P, E))
P = BairdCounterexample.constructTransitionMatrix(behaviorPolicy)
E = np.zeros((7, 7))
E[:, :] = 1 / 7
self.assertTrue(np.allclose(P, E))
def test_stateful(self):
env = BairdCounterexample(0)
s = env.start()
self.assertEqual(s, 6)
r, sp, t = env.step(0)
self.assertEqual(sp, 1)
self.assertEqual(r, 0)
self.assertFalse(t)
r, sp, t = env.step(1)
self.assertEqual(sp, 6)
self.assertEqual(r, 0)
self.assertFalse(t)
def test_actions(self):
actions = BairdCounterexample.actions(0)
self.assertEqual(list(actions), [0, 1])
def test_rewardVector(self):
R = BairdCounterexample.constructRewardVector(targetPolicy)
E = np.zeros(7)
self.assertTrue(np.allclose(R, E))
def test_terminal(self):
t = BairdCounterexample.terminal(0, 0, 1)
self.assertFalse(t)
def test_reward(self):
r = BairdCounterexample.reward(0, 0, 1)
self.assertEqual(r, 0)
def test_nextStates(self):
sp = BairdCounterexample.nextStates(0, 0)
self.assertEqual(list(sp), [0, 1, 2, 3, 4, 5])
sp = BairdCounterexample.nextStates(2, 1)
self.assertEqual(list(sp), [6])