def masked_branin(x):
x0 = np.abs(x[:, 0] - 0.5) < 0.25
x1 = np.abs(x[:, 1] - 0.5) < 0.40
mask_nan = np.logical_and(x0, x1)
y = np.array(objectives.branin(x))
y[mask_nan] = np.nan
return tf.convert_to_tensor(y.reshape(-1, 1), x.dtype)
def _dim_two_gp(mean_shift: Tuple[float, float] = (0.0, 0.0)) -> GaussianProcess:
matern52 = tfp.math.psd_kernels.MaternFiveHalves(
amplitude=tf.cast(2.3, tf.float64), length_scale=tf.cast(0.5, tf.float64)
)
return GaussianProcess(
[lambda x: mean_shift[0] + branin(x), lambda x: mean_shift[1] + quadratic(x)],
[matern52, rbf()],
)
def __call__(self, x):
if random.random() < 0.05:
self._is_broken = True
if self._is_broken:
raise Exception("Observer is broken")
return {OBJECTIVE: trieste.data.Dataset(x, branin(x))}
def test_batch_monte_carlo_expected_improvement_raises_for_model_with_wrong_event_shape() -> None:
builder = BatchMonteCarloExpectedImprovement(100)
data = mk_dataset([(0.0, 0.0)], [(0.0, 0.0)])
matern52 = tfp.math.psd_kernels.MaternFiveHalves(
amplitude=tf.cast(2.3, tf.float64), length_scale=tf.cast(0.5, tf.float64)
)
model = GaussianProcess([lambda x: branin(x), lambda x: quadratic(x)], [matern52, rbf()])
with pytest.raises(TF_DEBUGGING_ERROR_TYPES):
builder.prepare_acquisition_function(data, model)
def masked_branin(x):
mask_nan = np.sqrt((x[:, 0] - 0.5) ** 2 + (x[:, 1] - .4) ** 2) < 0.3
y = np.array(objectives.branin(x))
y[mask_nan] = np.nan
return tf.convert_to_tensor(y.reshape(-1, 1), x.dtype)
def test_branin_maps_argmin_values_to_global_minima() -> None:
expected = tf.broadcast_to(BRANIN_GLOBAL_MINIMUM, [3, 1])
npt.assert_allclose(branin(BRANIN_GLOBAL_ARGMIN), expected, atol=1e-6)
def test_branin_no_points_are_less_than_global_minimum() -> None:
search_values_1d = tf.range(1001.0) / 1000
x0, x1 = (tf.reshape(t, [-1, 1])
for t in tf.meshgrid(search_values_1d, search_values_1d))
x = tf.squeeze(tf.stack([x0, x1], axis=-1))
assert tf.reduce_all(branin(x) > BRANIN_GLOBAL_MINIMUM)
def test_branin_no_function_values_are_less_than_global_minimum() -> None:
samples = Box([0.0, 0.0], [1.0, 1.0]).sample(1000)
npt.assert_array_less(tf.squeeze(BRANIN_MINIMUM) - 1e-6, branin(samples))
def predict(self, query_points: TensorType) -> Tuple[TensorType, TensorType]:
return branin(query_points), self.kernel(query_points, query_points)[:, None]
def test_branin_no_function_values_are_less_than_global_minimum() -> None:
npt.assert_array_less(
tf.squeeze(BRANIN_MINIMUM) - 1e-6, branin(_unit_grid_2d()))
