def test_get_children():
hyp_space = utils.create_graph_hyp_space()
dag_one = hyp_space["common_cause_1"]
assert np.array_equal(dag_one.get_children(
0, dag_one.graph), np.array([1, 2]))
assert np.array_equal(dag_one.get_children(1, dag_one.graph), np.array([]))
assert np.array_equal(dag_one.get_children(2, dag_one.graph), np.array([]))
dag_two = hyp_space["common_effect_1"]
assert np.array_equal(dag_two.get_children(
0, dag_two.graph), np.array([2]))
assert np.array_equal(dag_two.get_children(
1, dag_two.graph), np.array([2]))
assert np.array_equal(dag_two.get_children(2, dag_two.graph), np.array([]))
dag_three = hyp_space["causal_chain_1"]
assert np.array_equal(dag_three.get_children(
0, dag_three.graph), np.array([1]))
assert np.array_equal(dag_three.get_children(
1, dag_three.graph), np.array([2]))
assert np.array_equal(dag_three.get_children(
2, dag_three.graph), np.array([]))
def test_graph_active_learner_two():
t = 0.8 # transmission rate
b = 0.0 # background rate
hyp_space = utils.create_graph_hyp_space(t=t, b=b)
example_two_graph_names = ['causal_chain_1', 'causal_chain_2']
example_two_graphs = [hyp_space[graph_name]
for graph_name in example_two_graph_names]
gal_two = GraphActiveLearner(example_two_graphs)
gal_two.update_posterior()
true_eig = np.array([0.11594429, 0.44202786, 0.44202786])
assert np.allclose(true_eig, gal_two.expected_information_gain())
def test_graph_active_learner_three():
t = 0.8 # transmission rate
b = 0.0 # background rate
hyp_space = utils.create_graph_hyp_space(t=t, b=b)
example_three_graph_names = ['common_effect_2', 'single_link_1']
example_three_graphs = [hyp_space[graph_name]
for graph_name in example_three_graph_names]
gal_three = GraphActiveLearner(example_three_graphs)
gal_three.update_posterior()
true_eig = np.array([0.0, 0.0, 1.0])
assert np.allclose(true_eig, gal_three.expected_information_gain())
def test_graph_active_learner_one():
t = 0.8 # transmission rate
b = 0.0 # background rate
# example one
hyp_space = utils.create_graph_hyp_space(t=t, b=b)
example_one_graph_names = ['common_cause_1', 'common_cause_2']
example_one_graphs = [hyp_space[graph_name]
for graph_name in example_one_graph_names]
gal_one = GraphActiveLearner(example_one_graphs)
gal_one.update_posterior()
true_eig = np.array([0.5, 0.5, 0])
assert np.array_equal(true_eig, gal_one.expected_information_gain())
def test_intervene():
t = 0.8
b = 0.01
hyp_space = utils.create_graph_hyp_space(t, b)
common_effect_1 = hyp_space["common_effect_1"]
common_effect_intervene_0 = np.array([[0, 0, 1], [0, 0, 1], [0, 0, 0]])
common_effect_intervene_1 = np.array([[0, 0, 1], [0, 0, 1], [0, 0, 0]])
common_effect_intervene_2 = np.array([[0, 0, 0], [0, 0, 0], [0, 0, 0]])
common_effect_1.intervene(0)
assert np.array_equal(
common_effect_1.intervened_graph, common_effect_intervene_0)
common_effect_1.intervene(1)
assert np.array_equal(
common_effect_1.intervened_graph, common_effect_intervene_1)
common_effect_1.intervene(2)
assert np.array_equal(
common_effect_1.intervened_graph, common_effect_intervene_2)
causal_chain_1 = hyp_space["causal_chain_1"]
causal_chain_intervene_0 = np.array([[0, 1, 0], [0, 0, 1], [0, 0, 0]])
causal_chain_intervene_1 = np.array([[0, 0, 0], [0, 0, 1], [0, 0, 0]])
causal_chain_intervene_2 = np.array([[0, 1, 0], [0, 0, 0], [0, 0, 0]])
causal_chain_1.intervene(0)
assert np.array_equal(
causal_chain_1.intervened_graph, causal_chain_intervene_0)
causal_chain_1.intervene(1)
assert np.array_equal(
causal_chain_1.intervened_graph, causal_chain_intervene_1)
causal_chain_1.intervene(2)
assert np.array_equal(
causal_chain_1.intervened_graph, causal_chain_intervene_2)