def test__update_hyperrectangles(make_random_dataset):
"""Testing if the updating of the hyperrectangles works as expected.
As above, the core functionality is tested in for the function with the logic.
Here, we're more interested in seeing how it works with the class object
"""
X, _ = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 4, beta_scale=1)
palinstance._means = np.array([[0, 0, 1, 0], [0, 0, 0, 1]])
palinstance.std = np.array([[0, 0, 0, 0], [1, 0, 0, 0]])
with pytest.raises(TypeError):
# Beta is not defined
palinstance._update_hyperrectangles()
palinstance._update_beta()
palinstance._update_hyperrectangles()
assert palinstance.rectangle_lows is not None
assert palinstance.rectangle_ups is not None
assert palinstance.rectangle_lows[0][0] == 0
assert palinstance.rectangle_ups[0][0] == 0
assert palinstance.rectangle_lows[0][2] == 1
assert palinstance.rectangle_ups[0][2] == 1
assert palinstance.rectangle_lows[1][0] < -1
assert palinstance.rectangle_ups[1][0] > 1
assert len(palinstance.hyperrectangle_sizes) == len(palinstance._means)
def test_orchestration_run_one_step(make_random_dataset):
"""Test if the orchestration works.
In the base class it should raise an error as without
prediction function we cannot do anything
"""
X, y = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 3, beta_scale=1)
sample_idx = np.array([1, 2, 3, 4])
palinstance.update_train_set(sample_idx, y[sample_idx])
with pytest.raises(NotImplementedError):
_ = palinstance.run_one_step()
def test_update_train_set(make_random_dataset):
"""Check if the update of the training set works"""
X, y = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 3)
assert not palinstance._has_train_set
assert palinstance.sampled.sum() == 0
palinstance.update_train_set(np.array([0]), y[0, :].reshape(-1, 3))
assert palinstance.sampled_indices == np.array([0])
assert palinstance.number_sampled_points == 1
assert (palinstance.y[0] == y[0, :]).all()
def test_beta_update(make_random_dataset):
"""testing that the beta update works"""
X, _ = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 3)
assert palinstance.beta is None
palinstance._update_beta()
assert palinstance.beta is not None
assert palinstance.beta == 1 / 9 * 2 * np.log(
3 * 100 * np.square(np.pi) * np.square(2) / (6 * 0.05))
def test_pal_base(make_random_dataset):
"""Testing basic functionality of the PAL base class"""
palinstance = PALBase(make_random_dataset[0], ["model"], 3)
assert palinstance.number_discarded_points == 0
assert palinstance.number_pareto_optimal_points == 0
assert palinstance.number_unclassified_points == 100
assert palinstance.number_sampled_points == 0
assert palinstance.number_design_points == 100
assert palinstance.should_cross_validate()
assert len(palinstance.discarded_points) == 0
assert len(palinstance.pareto_optimal_points) == 0
assert len(palinstance.unclassified_points) == 100
with pytest.raises(ValueError):
palinstance.hyperrectangle_sizes # pylint:disable=pointless-statement
assert (str(palinstance) == "pyepal at iteration 1. \
0 Pareto optimal points, \
0 discarded points, \
100 unclassified points.")
assert palinstance._should_optimize_hyperparameters()
assert not palinstance._has_train_set
with pytest.raises(ValueError):
palinstance.sample()
assert palinstance.y.shape == (100, 3)
assert not palinstance.uses_fixed_epsilon
palinstance = PALBase(make_random_dataset[0], ["model"],
3,
ranges=[1, 1, 1])
assert palinstance.uses_fixed_epsilon
def test__replace_by_measurements(make_random_dataset):
"""Test that replacing the mean/std by the measured ones works"""
X, y = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 3, beta_scale=1)
assert palinstance.measurement_uncertainty.sum() == 0
sample_idx = np.array([1, 2, 3, 4])
palinstance.update_train_set(sample_idx, y[sample_idx], y[sample_idx])
palinstance._means = palinstance.measurement_uncertainty
palinstance.std = palinstance.measurement_uncertainty
palinstance._replace_by_measurements()
assert (palinstance.y == palinstance.std).all()
def test_turn_to_maximization(make_random_dataset):
"""Test that flipping the sign for minimization problems works"""
X, y = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 3)
palinstance.update_train_set(np.array([0]), y[0, :].reshape(-1, 3))
assert (palinstance.y[0] == y[0, :]).all()
assert (palinstance._y[0] == y[0, :]).all()
palinstance = PALBase(X, ["model"], 3, goals=[1, 1, -1])
palinstance.update_train_set(np.array([0]), y[0, :].reshape(-1, 3))
assert (palinstance.y[0] == y[0, :] * np.array([1, 1, -1])).all()
assert (palinstance._y[0] == y[0, :]).all()
def test__update_coef_var_mask(make_random_dataset):
"""Test that the coefficient of variation mask works as expected"""
X, _ = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X[:2], ["model"], 3, beta_scale=1)
palinstance._means = np.array([[1, 1, 1, 1], [1, 1, 1, 1]])
palinstance.std = np.array([[0, 0, 0, 0], [3, 1, 1, 1]])
assert (palinstance.coef_var_mask == np.array([True, True])).all()
palinstance._update_coef_var_mask()
assert (palinstance.coef_var_mask == np.array([True, False])).all()
palinstance._means = np.array([[0, 0, 0, 0], [0, 0, 0, 0]])
palinstance.std = np.array([[0, 0, 0, 0], [3, 1, 1, 1]])
assert (palinstance.coef_var_mask == np.array([True, False])).all()
def test_reset_classification(make_random_dataset):
"""Make sure the reset of the reclassification status works as expected."""
X, _ = make_random_dataset # pylint: disable=invalid-name
palinstance = PALBase(X, ["model"], 3)
lows = np.zeros((100, 3))
highs = np.zeros((100, 3))
means = np.full((100, 3), 1)
palinstance._means = means
palinstance.std = np.full((100, 3), 0.1)
pareto_optimal = np.array([False] * 98 + [True, True])
sampled = np.array([[False] * 3, [False] * 3, [False] * 3, [False] * 3])
unclassified = np.array([True] * 98 + [False, False])
palinstance.rectangle_lows = lows
palinstance.rectangle_ups = highs
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
palinstance._reset_classification()
assert palinstance.number_unclassified_points == len(X)
assert palinstance.number_pareto_optimal_points == 0
assert palinstance.number_discarded_points == 0
def test_sample(make_random_dataset):
"""Test the sampling"""
X, _ = make_random_dataset # pylint:disable=invalid-name
palinstance = PALBase(X, ["model"], 4)
lows = np.array([
[0.0, 0.0, 0.0, 0.0],
[-1.0, -1.0, -1.0, -1.0],
[-2.0, -2.0, -2.0, -2.0],
[2.0, 2.0, 2.0, 2.0],
])
highs = np.array([
[1.0, 1.0, 1.0, 1.0],
[1.0, 1.0, 1.0, 1.0],
[1.0, 1.0, 1.0, 1.0],
[2.0, 2.0, 2.0, 2.0],
])
means = np.array([[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]])
palinstance._means = means
pareto_optimal = np.array([False, False, True, True])
sampled = np.array([[False] * 4, [False] * 4, [False] * 4, [False] * 4])
unclassified = np.array([True, True, False, False])
palinstance.rectangle_lows = lows
palinstance.rectangle_ups = highs
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
sampled_idx = palinstance.sample()
assert sampled_idx == 2
pareto_optimal = np.array([False, False, True, True])
sampled = np.array([[False] * 4, [False] * 4, [False] * 4, [False] * 4])
unclassified = np.array([True, True, False, False])
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
sampled_idx = palinstance.sample()
assert sampled_idx == 2
pareto_optimal = np.array([False, False, True, True])
sampled = np.array([[False] * 4, [False] * 4, [True] * 4, [False] * 4])
unclassified = np.array([True, True, False, False])
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
sampled_idx = palinstance.sample()
assert sampled_idx == 1
pareto_optimal = np.array([False, False, False, True])
sampled = np.array([[False] * 4, [False] * 4, [True] * 4, [False] * 4])
unclassified = np.array([True, True, False, False])
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
sampled_idx = palinstance.sample()
assert sampled_idx == 1
def test_augment_design_space(make_random_dataset):
"""Testing the basic functionality of the augmentation method
Does NOT test the re-classification step, which needs a model"""
X, _ = make_random_dataset # pylint: disable=invalid-name
X_augmented = np.vstack([X, X]) # pylint: disable=invalid-name
palinstance = PALBase(X, ["model"], 3)
# Iteration count to low
with pytest.raises(ValueError):
palinstance.augment_design_space(X_augmented)
palinstance.iteration = 2
# Incorrect shape
with pytest.raises(AssertionError):
palinstance.augment_design_space(X_augmented[:, 2])
with pytest.raises(ValueError):
palinstance.augment_design_space(X_augmented[:, :2])
# Mock that we already ran that
lows = np.zeros((100, 3))
highs = np.zeros((100, 3))
means = np.full((100, 3), 1)
palinstance._means = means
palinstance.std = np.full((100, 3), 0.1)
pareto_optimal = np.array([False] * 98 + [True, True])
sampled = np.array([[False] * 3, [False] * 3, [False] * 3, [False] * 3])
unclassified = np.array([True] * 98 + [False, False])
palinstance.rectangle_lows = lows
palinstance.rectangle_ups = highs
palinstance.sampled = sampled
palinstance.pareto_optimal = pareto_optimal
palinstance.unclassified = unclassified
# As we do not have a model, we cannot test the classification
palinstance.augment_design_space(X_augmented, clean_classify=False)
assert palinstance.number_discarded_points == 0
assert palinstance.number_pareto_optimal_points == 2
assert palinstance.number_unclassified_points == 298
assert palinstance.number_sampled_points == 0
assert palinstance.number_design_points == 300
assert len(palinstance.means) == 300
assert len(palinstance.std) == 300