def test_suggested_value():
# type: () -> None
independent_sampler = FirstTrialOnlyRandomSampler()
sampler = optuna.integration.SkoptSampler(
independent_sampler=independent_sampler,
skopt_kwargs={'n_initial_points': 5})
# direction='minimize'
study = optuna.create_study(sampler=sampler, direction='minimize')
study.optimize(_objective, n_trials=10, catch=())
for trial in study.trials:
for param_name, param_value in trial.params_in_internal_repr.items():
assert trial.distributions[param_name]._contains(param_value)
# direction='maximize'
study = optuna.create_study(sampler=sampler, direction='maximize')
study.optimize(_objective, n_trials=10, catch=())
for trial in study.trials:
for param_name, param_value in trial.params_in_internal_repr.items():
assert trial.distributions[param_name]._contains(param_value)
from optuna.integration import SkoptSampler
from optuna.testing.sampler import FirstTrialOnlyRandomSampler
if optuna.type_checking.TYPE_CHECKING:
from typing import Callable # NOQA
from optuna.samplers import BaseSampler # NOQA
parametrize_sampler = pytest.mark.parametrize('sampler_class', [
optuna.integration.SkoptSampler,
optuna.integration.CmaEsSampler,
])
@pytest.mark.parametrize('sampler_class', [
lambda: SkoptSampler(independent_sampler=FirstTrialOnlyRandomSampler(),
skopt_kwargs={'n_initial_points': 5}),
lambda: CmaEsSampler(independent_sampler=FirstTrialOnlyRandomSampler()),
])
def test_suggested_value(sampler_class):
# type: (Callable[[], BaseSampler]) -> None
sampler = sampler_class()
# direction='minimize'
study = optuna.create_study(sampler=sampler, direction='minimize')
study.optimize(_objective, n_trials=10, catch=())
for trial in study.trials:
for param_name, param_value in trial.params.items():
distribution = trial.distributions[param_name]
param_value_in_internal_repr = distribution.to_internal_repr(
param_value)
import optuna
from optuna.integration import PyCmaSampler
from optuna.integration import SkoptSampler
from optuna.samplers import BaseSampler
from optuna.testing.sampler import FirstTrialOnlyRandomSampler
parametrize_sampler = pytest.mark.parametrize(
"sampler_class",
[optuna.integration.SkoptSampler, optuna.integration.PyCmaSampler])
@pytest.mark.parametrize(
"sampler_class",
[
lambda: SkoptSampler(independent_sampler=FirstTrialOnlyRandomSampler()
),
lambda: PyCmaSampler(independent_sampler=FirstTrialOnlyRandomSampler()
),
],
)
def test_suggested_value(sampler_class: Callable[[], BaseSampler]) -> None:
sampler = sampler_class()
# direction='minimize'
study = optuna.create_study(sampler=sampler, direction="minimize")
study.optimize(_objective, n_trials=10, catch=())
for trial in study.trials:
for param_name, param_value in trial.params.items():
distribution = trial.distributions[param_name]
param_value_in_internal_repr = distribution.to_internal_repr(
from optuna.testing.sampler import FirstTrialOnlyRandomSampler
if optuna.types.TYPE_CHECKING:
from typing import Callable # NOQA
from optuna.samplers import BaseSampler # NOQA
parametrize_sampler = pytest.mark.parametrize('sampler_class', [
optuna.integration.SkoptSampler,
optuna.integration.CmaEsSampler,
])
@pytest.mark.parametrize('sampler_class', [
lambda: SkoptSampler(
independent_sampler=FirstTrialOnlyRandomSampler(), skopt_kwargs={'n_initial_points': 5}),
lambda: CmaEsSampler(independent_sampler=FirstTrialOnlyRandomSampler()),
])
def test_suggested_value(sampler_class):
# type: (Callable[[], BaseSampler]) -> None
sampler = sampler_class()
# direction='minimize'
study = optuna.create_study(sampler=sampler, direction='minimize')
study.optimize(_objective, n_trials=10, catch=())
for trial in study.trials:
for param_name, param_value in trial.params.items():
distribution = trial.distributions[param_name]
param_value_in_internal_repr = distribution.to_internal_repr(param_value)
assert distribution._contains(param_value_in_internal_repr)