def __init__(self,
env,
endeavours_bias=0.1,
longterm_satisfaction_bias=0.9):
self.endeavours_bias = endeavours_bias
self.longterm_satisfaction_bias = longterm_satisfaction_bias
self.env = env
num_states = env.COLS * env.ROWS
num_actions = len(Action.all())
self.qvalues = np.zeros((num_states, num_actions))
def choose_action(self):
if np.random.random() < self.endeavours_bias:
return np.random.choice(Action.all())
else:
return Action.all()[np.argmax(self.qvalues[self.get_state()])]
def test_actions(self):
actions = Action.all()
self.assertEqual(4, len(actions))