def testEncodeDecodeSimpleBenchmarkProblem(self):
branin_problem = get_branin_simple_benchmark_problem()
sum_problem = get_sum_simple_benchmark_problem()
new_branin_problem = object_from_json(
object_to_json(
branin_problem,
encoder_registry=DEPRECATED_ENCODER_REGISTRY,
class_encoder_registry=DEPRECATED_CLASS_ENCODER_REGISTRY,
),
decoder_registry=DEPRECATED_DECODER_REGISTRY,
class_decoder_registry=DEPRECATED_CLASS_DECODER_REGISTRY,
)
new_sum_problem = object_from_json(
object_to_json(
sum_problem,
encoder_registry=DEPRECATED_ENCODER_REGISTRY,
class_encoder_registry=DEPRECATED_CLASS_ENCODER_REGISTRY,
),
decoder_registry=DEPRECATED_DECODER_REGISTRY,
class_decoder_registry=DEPRECATED_CLASS_DECODER_REGISTRY,
)
self.assertEqual(
branin_problem.f(1, 2), new_branin_problem.f(1, 2), branin(1, 2)
)
self.assertEqual(sum_problem.f([1, 2]), new_sum_problem.f([1, 2]), 3)
# Test using `from_botorch`.
ackley_problem = SimpleBenchmarkProblem(
f=from_botorch(Ackley()), noise_sd=0.0, minimize=True
)
new_ackley_problem = object_from_json(
object_to_json(
ackley_problem,
encoder_registry=DEPRECATED_ENCODER_REGISTRY,
class_encoder_registry=DEPRECATED_CLASS_ENCODER_REGISTRY,
),
decoder_registry=DEPRECATED_DECODER_REGISTRY,
class_decoder_registry=DEPRECATED_CLASS_DECODER_REGISTRY,
)
self.assertEqual(
ackley_problem.f(1, 2), new_ackley_problem.f(1, 2), ackley(1, 2)
)
def _extract_optimization_trace_from_synthetic_function(
experiment: Experiment, problem: SimpleBenchmarkProblem
) -> np.ndarray:
if any(isinstance(trial, BatchTrial) for trial in experiment.trials.values()):
raise NotImplementedError("Batched trials are not yet supported.")
true_values = []
for trial in experiment.trials.values():
parameters = not_none(checked_cast(Trial, trial).arm).parameters
# Expecting numerical parameters only.
value = problem.f(*[float(x) for x in parameters.values()]) # pyre-ignore[6]
true_values.append(value)
return best_feasible_objective(
optimization_config=experiment.optimization_config,
values={problem.name: true_values},
)
def simple_benchmark_problem_from_json(
object_json: Dict[str, Any]) -> SimpleBenchmarkProblem:
"""Load a benchmark problem from JSON."""
uses_synthetic_function = object_json.pop("uses_synthetic_function")
if uses_synthetic_function:
f = getattr(synthetic_functions, object_json.pop("function_name"))()
else:
f = pickle.loads(object_json.pop("f").encode())
domain = object_from_json(object_json.pop("domain"))
assert isinstance(domain, list) and all(
isinstance(x, tuple) for x in domain)
return SimpleBenchmarkProblem(
f=f,
name=object_json.pop("name"),
domain=cast(List[Tuple[float, float]], domain),
minimize=object_json.pop("minimize"),
)
def test_raise_all_exceptions(self):
"""Checks that an exception nested in the benchmarking stack is raised
when `raise_all_exceptions` is True.
"""
def broken_benchmark_replication(*args, **kwargs) -> Experiment:
raise ValueError("Oh, exception!")
with self.assertRaisesRegex(ValueError, "Oh, exception!"):
full_benchmark_run(
problems=[SimpleBenchmarkProblem(branin, noise_sd=0.4)],
methods=[
GenerationStrategy(steps=[
GenerationStep(model=Models.SOBOL, num_arms=-1)
])
],
num_replications=3,
num_trials=5,
raise_all_exceptions=True,
benchmark_replication=broken_benchmark_replication,
)
def simple_benchmark_problem_from_json(
object_json: Dict[str, Any]) -> SimpleBenchmarkProblem:
"""Load a benchmark problem from JSON."""
uses_synthetic_function = object_json.pop("uses_synthetic_function")
if uses_synthetic_function:
function_name = object_json.pop("function_name")
if function_name.startswith(synthetic_functions.FromBotorch.__name__):
raise NotImplementedError # TODO[Lena], pragma: no cover
else:
f = getattr(synthetic_functions, function_name)()
else:
f = pickle.loads(object_json.pop("f").encode())
domain = object_from_json(object_json.pop("domain"))
assert isinstance(domain, list) and all(
isinstance(x, (tuple, list)) for x in domain)
return SimpleBenchmarkProblem(
f=f,
name=object_json.pop("name"),
domain=cast(List[Tuple[float, float]], domain),
minimize=object_json.pop("minimize"),
)
# LICENSE file in the root directory of this source tree.
from ax.benchmark.benchmark_problem import BenchmarkProblem, SimpleBenchmarkProblem
from ax.utils.measurement.synthetic_functions import from_botorch
from ax.utils.testing.core_stubs import (
get_augmented_branin_optimization_config,
get_augmented_hartmann_optimization_config,
get_branin_search_space,
get_hartmann_search_space,
)
from botorch.test_functions.synthetic import Ackley, Branin
# Initialize the single-fidelity problems
ackley = SimpleBenchmarkProblem(f=from_botorch(Ackley()), noise_sd=0.0, minimize=True)
branin = SimpleBenchmarkProblem(f=from_botorch(Branin()), noise_sd=0.0, minimize=True)
single_fidelity_problem_group = [ackley, branin]
# Initialize the multi-fidelity problems
augmented_branin = BenchmarkProblem(
search_space=get_branin_search_space(with_fidelity_parameter=True),
optimization_config=get_augmented_branin_optimization_config(),
)
augmented_hartmann = BenchmarkProblem(
search_space=get_hartmann_search_space(with_fidelity_parameter=True),
optimization_config=get_augmented_hartmann_optimization_config(),
)
multi_fidelity_problem_group = [augmented_branin, augmented_hartmann]
# Gather all of the problems
def _benchmark_replication_Service_API(
problem: SimpleBenchmarkProblem,
method: GenerationStrategy,
num_trials: int,
experiment_name: str,
batch_size: int = 1,
raise_all_exceptions: bool = False,
benchmark_trial: FunctionType = benchmark_trial,
verbose_logging: bool = True,
# Number of trials that need to fail for a replication to be considered failed.
failed_trials_tolerated: int = 5,
async_benchmark_options: Optional[AsyncBenchmarkOptions] = None,
) -> Tuple[Experiment, List[Exception]]:
"""Run a benchmark replication via the Service API because the problem was
set up in a simplified way, without the use of Ax classes like `OptimizationConfig`
or `SearchSpace`.
"""
if async_benchmark_options is not None:
raise NonRetryableBenchmarkingError(
"`async_benchmark_options` not supported when using the Service API."
)
exceptions = []
if batch_size == 1:
ax_client = AxClient(generation_strategy=method,
verbose_logging=verbose_logging)
else: # pragma: no cover, TODO[T53975770]
assert batch_size > 1, "Batch size of 1 or greater is expected."
raise NotImplementedError(
"Batched benchmarking on `SimpleBenchmarkProblem`-s not yet implemented."
)
ax_client.create_experiment(
name=experiment_name,
parameters=problem.domain_as_ax_client_parameters(),
minimize=problem.minimize,
objective_name=problem.name,
)
parameter_names = list(ax_client.experiment.search_space.parameters.keys())
assert num_trials > 0
for _ in range(num_trials):
parameterization, idx = ax_client.get_next_trial()
param_values = np.array(
[parameterization.get(x) for x in parameter_names])
try:
mean, sem = benchmark_trial(parameterization=param_values,
evaluation_function=problem.f)
# If problem indicates a noise level and is using a synthetic callable,
# add normal noise to the measurement of the mean.
if problem.uses_synthetic_function and problem.noise_sd != 0.0:
noise = np.random.randn() * problem.noise_sd
sem = (sem or 0.0) + problem.noise_sd
logger.info(
f"Adding noise of {noise} to the measurement mean ({mean})."
f"Problem noise SD setting: {problem.noise_sd}.")
mean = mean + noise
ax_client.complete_trial(trial_index=idx, raw_data=(mean, sem))
except Exception as err: # TODO[T53975770]: test
if raise_all_exceptions:
raise
exceptions.append(err)
if len(exceptions) > failed_trials_tolerated:
raise RuntimeError( # TODO[T53975770]: test
f"More than {failed_trials_tolerated} failed for {experiment_name}."
)
return ax_client.experiment, exceptions
def get_mult_simple_benchmark_problem() -> SimpleBenchmarkProblem:
return SimpleBenchmarkProblem(
f=cast(FunctionType, sample_multiplication_fxn),
name="Sum",
domain=[(0.0, 1.0), (0.0, 1.0), (0.0, 1.0)],
)
def get_sum_simple_benchmark_problem() -> SimpleBenchmarkProblem:
return SimpleBenchmarkProblem(f=sum,
name="Sum",
domain=[(0.0, 1.0), (0.0, 1.0)])
def get_branin_simple_benchmark_problem() -> SimpleBenchmarkProblem:
return SimpleBenchmarkProblem(f=branin)
