def test_sequential_optimize(self, mock_joint_optimize, cuda=False):
q = 3
num_restarts = 2
raw_samples = 10
options = {}
tkwargs = {
"device": torch.device("cuda") if cuda else torch.device("cpu")
}
for dtype in (torch.float, torch.double):
mock_acq_function = MockAcquisitionFunction()
tkwargs["dtype"] = dtype
joint_optimize_return_values = [
torch.tensor([[[1.1, 2.1, 3.1]]], **tkwargs) for _ in range(q)
]
mock_joint_optimize.side_effect = joint_optimize_return_values
expected_candidates = torch.cat(joint_optimize_return_values,
dim=-2).round()
bounds = torch.stack(
[torch.zeros(3, **tkwargs), 4 * torch.ones(3, **tkwargs)])
inequality_constraints = [(torch.tensor([3]), torch.tensor([4]),
torch.tensor(5))]
candidates = sequential_optimize(
acq_function=mock_acq_function,
bounds=bounds,
q=q,
num_restarts=num_restarts,
raw_samples=raw_samples,
options=options,
inequality_constraints=inequality_constraints,
post_processing_func=rounding_func,
)
self.assertTrue(torch.equal(candidates, expected_candidates))
expected_call_kwargs = {
"acq_function": mock_acq_function,
"bounds": bounds,
"q": 1,
"num_restarts": num_restarts,
"raw_samples": raw_samples,
"options": options,
"inequality_constraints": inequality_constraints,
"equality_constraints": None,
"fixed_features": None,
}
call_args_list = mock_joint_optimize.call_args_list[-q:]
for i in range(q):
self.assertEqual(call_args_list[i][1], expected_call_kwargs)
# test that error is raised for acquisition functions without X_baseline
mock_acq_function = MockAcquisitionFunction(
has_X_baseline_attr=False)
with self.assertRaises(UnsupportedError):
sequential_optimize(
acq_function=mock_acq_function,
bounds=bounds,
q=q,
num_restarts=num_restarts,
raw_samples=raw_samples,
)
def test_sequential_optimize(self, mock_joint_optimize, cuda=False):
q = 3
num_restarts = 2
raw_samples = 10
options = {}
tkwargs = {"device": torch.device("cuda") if cuda else torch.device("cpu")}
for dtype in (torch.float, torch.double):
mock_acq_function = MockAcquisitionFunction()
tkwargs["dtype"] = dtype
joint_optimize_return_values = [
torch.tensor([[[1.1, 2.1, 3.1]]], **tkwargs) for _ in range(q)
]
mock_joint_optimize.side_effect = joint_optimize_return_values
expected_candidates = torch.cat(
joint_optimize_return_values, dim=-2
).round()
bounds = torch.stack(
[torch.zeros(3, **tkwargs), 4 * torch.ones(3, **tkwargs)]
)
inequality_constraints = [
(torch.tensor([3]), torch.tensor([4]), torch.tensor(5))
]
candidates = sequential_optimize(
acq_function=mock_acq_function,
bounds=bounds,
q=q,
num_restarts=num_restarts,
raw_samples=raw_samples,
options=options,
inequality_constraints=inequality_constraints,
post_processing_func=rounding_func,
)
self.assertTrue(torch.equal(candidates, expected_candidates))
expected_call_kwargs = {
"acq_function": mock_acq_function,
"bounds": bounds,
"q": 1,
"num_restarts": num_restarts,
"raw_samples": raw_samples,
"options": options,
"inequality_constraints": inequality_constraints,
"equality_constraints": None,
"fixed_features": None,
}
call_args_list = mock_joint_optimize.call_args_list[-q:]
for i in range(q):
self.assertEqual(call_args_list[i][1], expected_call_kwargs)
# test that error is raised for acquisition functions without X_baseline
mock_acq_function = MockAcquisitionFunction(has_X_baseline_attr=False)
with self.assertRaises(UnsupportedError):
sequential_optimize(
acq_function=mock_acq_function,
bounds=bounds,
q=q,
num_restarts=num_restarts,
raw_samples=raw_samples,
)
def test_sequential_optimize(self, mock_optimize_acqf):
with warnings.catch_warnings(record=True) as ws:
candidates, acq_values = sequential_optimize(**self.shared_kwargs)
self.assertTrue(any(issubclass(w.category, DeprecationWarning) for w in ws))
self.assertTrue(
any("sequential_optimize is deprecated" in str(w.message) for w in ws)
)
mock_optimize_acqf.assert_called_once_with(
**self.shared_kwargs,
return_best_only=True,
sequential=True,
batch_initial_conditions=None,
)
self.assertIsNone(candidates)
self.assertIsNone(acq_values)