def test_update_array_roulette_vacancy_picker_01(self):
array = np.array([[0, 0, 0], [0, 1, 1], [1, 2, 2]])
utility = get_utility_for_array(create_flat_utility(0.5), array, True)
states_for_picked = [
(0.1, np.array([[0, 0, 2], [0, 1, 1], [1, 0, 2]])),
(0.2, np.array([[2, 0, 0], [0, 1, 1], [1, 0, 2]])),
(1.2, np.array([[2, 0, 0], [0, 1, 1], [1, 0, 2]])),
(2.3, np.array([[0, 2, 0], [0, 1, 1], [1, 0, 2]])),
(3.4, np.array([[0, 0, 0], [2, 1, 1], [1, 0, 2]])),
(3.5, np.array([[0, 0, 0], [2, 1, 1], [1, 0, 2]])),
]
result = SimulationResult()
for picked_value, expected_state in states_for_picked:
array = np.array([[0, 0, 0], [0, 1, 1], [1, 2, 2]])
vacancy_picker = _create_roulette_picker(0.1, utility, False,
lambda *a: picked_value)
agent_picker = _create_roulette_picker(0.0, utility, True,
lambda *a: 0.0)
update_array(array, utility, result, agent_picker, vacancy_picker,
True, ['entropy_average'], True)
with self.subTest(v=picked_value, out=array, exp=expected_state):
self.assertTrue(np.array_equal(array, expected_state))
def test_roulette_picker_weight01(self):
agent_indices = [
(1, 2),
(7, 8),
(4, 5),
(8, 9),
(3, 4),
]
agent_indices_sorted = [
(8, 9),
(7, 8),
(4, 5),
(1, 2),
(3, 4),
]
agent_utilities = [
0.75,
0.25,
0.5,
0.1,
1.0,
]
def utility(agent_index, agent_type=None):
i = agent_indices.index(tuple(agent_index))
return agent_utilities[i]
agent_cum_weights = [1.0, 1.85, 2.45, 2.8, 2.9]
picked_values = np.arange(0, 3.0, 0.1)
for picked_value in picked_values:
def uniform_dist(hibound):
return picked_value
roulette_picker = _create_roulette_picker(
0.1, utility, for_agents=True, uniform_dist=uniform_dist)
index = -1
for i in range(len(agent_cum_weights)):
if picked_value < agent_cum_weights[i] or math.isclose(
picked_value, agent_cum_weights[i]):
index = i
break
expected_output = agent_indices_sorted[index]
output = roulette_picker(np.array(agent_indices))
with self.subTest(picked_value=picked_value):
self.assertEqual(output, expected_output)