# Copyright (c) Meta Platforms, Inc. and affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
# pyre-strict
from contextlib import contextmanager, ExitStack
from functools import wraps
from typing import Any, Callable, Dict, Generator, Optional
from unittest import mock
from botorch.fit import fit_fully_bayesian_model_nuts
from botorch.generation.gen import minimize_with_timeout
from botorch.optim.initializers import (
gen_batch_initial_conditions,
gen_one_shot_kg_initial_conditions,
)
from scipy.optimize import OptimizeResult
from torch import Tensor
[docs]@contextmanager
def fast_botorch_optimize_context_manager(
force: bool = False,
) -> Generator[None, None, None]:
"""A context manager to force botorch to speed up optimization. Currently, the
primary tactic is to force the underlying scipy methods to stop after just one
iteration.
force: If True will not raise an AssertionError if no mocks are called.
USE RESPONSIBLY.
"""
def one_iteration_minimize(
*args: Any, **kwargs: Any
) -> OptimizeResult: # pyre-ignore[11]
if kwargs["options"] is None:
kwargs["options"] = {}
kwargs["options"]["maxiter"] = 1
return minimize_with_timeout(*args, **kwargs)
def minimal_gen_ics(*args: Any, **kwargs: Any) -> Tensor:
kwargs["num_restarts"] = 2
kwargs["raw_samples"] = 4
return gen_batch_initial_conditions(*args, **kwargs)
def minimal_gen_os_ics(*args: Any, **kwargs: Any) -> Optional[Tensor]:
kwargs["num_restarts"] = 2
kwargs["raw_samples"] = 4
return gen_one_shot_kg_initial_conditions(*args, **kwargs)
def minimal_fit_fully_bayesian(*args: Any, **kwargs: Any) -> None:
fit_fully_bayesian_model_nuts(*args, **_get_minimal_mcmc_kwargs(**kwargs))
with ExitStack() as es:
mock_generation = es.enter_context(
mock.patch(
"botorch.generation.gen.minimize_with_timeout",
wraps=one_iteration_minimize,
)
)
mock_fit = es.enter_context(
mock.patch(
"botorch.optim.core.minimize_with_timeout",
wraps=one_iteration_minimize,
)
)
mock_gen_ics = es.enter_context(
mock.patch(
"botorch.optim.optimize.gen_batch_initial_conditions",
wraps=minimal_gen_ics,
)
)
mock_gen_os_ics = es.enter_context(
mock.patch(
"botorch.optim.optimize.gen_one_shot_kg_initial_conditions",
wraps=minimal_gen_os_ics,
)
)
mock_mcmc_mbm = es.enter_context(
mock.patch(
"ax.models.torch.botorch_modular.utils.fit_fully_bayesian_model_nuts",
wraps=minimal_fit_fully_bayesian,
)
)
yield
if (not force) and all(
mock_.call_count < 1
for mock_ in [
mock_generation,
mock_fit,
mock_gen_ics,
mock_gen_os_ics,
mock_mcmc_mbm,
]
):
raise AssertionError(
"No mocks were called in the context manager. Please remove unused "
"fast_botorch_optimize_context_manager()."
)
# pyre-fixme[24]: Generic type `Callable` expects 2 type parameters.
[docs]def fast_botorch_optimize(f: Callable) -> Callable:
"""Wraps f in the fast_botorch_optimize_context_manager for use as a decorator."""
@wraps(f)
# pyre-fixme[3]: Return type must be annotated.
def inner(*args: Any, **kwargs: Any):
with fast_botorch_optimize_context_manager():
return f(*args, **kwargs)
return inner
[docs]@contextmanager
def skip_fit_gpytorch_mll_context_manager() -> Generator[None, None, None]:
"""A context manager that makes `fit_gpytorch_mll` a no-op.
This should only be used to speed up slow tests.
"""
with mock.patch(
"botorch.fit.FitGPyTorchMLL", side_effect=lambda *args, **kwargs: args[0]
) as mock_fit:
yield
if mock_fit.call_count < 1:
raise AssertionError(
"No mocks were called in the context manager. Please remove unused "
"skip_fit_gpytorch_mll_context_manager()."
)
# pyre-fixme[24]: Generic type `Callable` expects 2 type parameters.
[docs]def skip_fit_gpytorch_mll(f: Callable) -> Callable:
"""Wraps f in the skip_fit_gpytorch_mll_context_manager for use as a decorator."""
@wraps(f)
# pyre-fixme[3]: Return type must be annotated.
def inner(*args: Any, **kwargs: Any):
with skip_fit_gpytorch_mll_context_manager():
return f(*args, **kwargs)
return inner
def _get_minimal_mcmc_kwargs(**kwargs: Any) -> Dict[str, Any]:
kwargs["warmup_steps"] = 0
# Just get as many samples as otherwise expected.
kwargs["num_samples"] = kwargs.get("num_samples", 256) // kwargs.get("thinning", 16)
kwargs["thinning"] = 1
return kwargs
