optuna.samplers.QMCSampler

class optuna.samplers.QMCSampler(*, qmc_type='sobol', scramble=False, seed=None, independent_sampler=None, warn_asynchronous_seeding=True, warn_independent_sampling=True)[source]

A Quasi Monte Carlo Sampler that generates low-discrepancy sequences.

Quasi Monte Carlo (QMC) sequences are designed to have lower discrepancies than standard random sequences. They are known to perform better than the standard random sequences in hyperparameter optimization.

For further information about the use of QMC sequences for hyperparameter optimization, please refer to the following paper:

We use the QMC implementations in Scipy. For the details of the QMC algorithm, see the Scipy API references on scipy.stats.qmc.

Note

The search space of the sampler is determined by either previous trials in the study or the first trial that this sampler samples.

If there are previous trials in the study, QMCSampler infers its search space using the trial which was created first in the study.

Otherwise (if the study has no previous trials), QMCSampler samples the first trial using its independent_sampler and then infers the search space in the second trial.

As mentioned above, the search space of the QMCSampler is determined by the first trial of the study. Once the search space is determined, it cannot be changed afterwards.

Parameters:

  • qmc_type (str) –

    The type of QMC sequence to be sampled. This must be one of “halton” and “sobol”. Default is “sobol”.

    Note

    Sobol’ sequence is designed to have low-discrepancy property when the number of samples is \(n=2^m\) for each positive integer \(m\). When it is possible to pre-specify the number of trials suggested by QMCSampler, it is recommended that the number of trials should be set as power of two.

  • scramble (bool) – If this option is True, scrambling (randomization) is applied to the QMC sequences.

  • seed (int | None) –

    A seed for QMCSampler. This argument is used only when scramble is True. If this is None, the seed is initialized randomly. Default is None.

    Note

    When using multiple QMCSampler’s in parallel and/or distributed optimization, all the samplers must share the same seed when the scrambling is enabled. Otherwise, the low-discrepancy property of the samples will be degraded.

  • independent_sampler (BaseSampler | None) –

    A BaseSampler instance that is used for independent sampling. The first trial of the study and the parameters not contained in the relative search space are sampled by this sampler.

    If None is specified, RandomSampler is used as the default.

    See also

    samplers module provides built-in independent samplers such as RandomSampler and TPESampler.

  • warn_independent_sampling (bool) –

    If this is True, a warning message is emitted when the value of a parameter is sampled by using an independent sampler.

    Note that the parameters of the first trial in a study are sampled via an independent sampler in most cases, so no warning messages are emitted in such cases.

  • warn_asynchronous_seeding (bool) –

    If this is True, a warning message is emitted when the scrambling (randomization) is applied to the QMC sequence and the random seed of the sampler is not set manually.

    Note

    When using parallel and/or distributed optimization without manually setting the seed, the seed is set randomly for each instances of QMCSampler for different workers, which ends up asynchronous seeding for multiple samplers used in the optimization.

    See also

    See parameter seed in QMCSampler.

Example

Optimize a simple quadratic function by using QMCSampler.

import optuna


def objective(trial):
    x = trial.suggest_float("x", -1, 1)
    y = trial.suggest_int("y", -1, 1)
    return x**2 + y


sampler = optuna.samplers.QMCSampler()
study = optuna.create_study(sampler=sampler)
study.optimize(objective, n_trials=8)

Note

Added in v3.0.0 as an experimental feature. The interface may change in newer versions without prior notice. See https://github.com/optuna/optuna/releases/tag/v3.0.0.

Methods

after_trial(study, trial, state, values)

Trial post-processing.

before_trial(study, trial)

Trial pre-processing.

infer_relative_search_space(study, trial)

Infer the search space that will be used by relative sampling in the target trial.

reseed_rng()

Reseed sampler's random number generator.

sample_independent(study, trial, param_name, ...)

Sample a parameter for a given distribution.

sample_relative(study, trial, search_space)

Sample parameters in a given search space.
after_trial(study, trial, state, values)[source]

Trial post-processing.

This method is called after the objective function returns and right before the trial is finished and its state is stored.

Note

Added in v2.4.0 as an experimental feature. The interface may change in newer versions without prior notice. See https://github.com/optuna/optuna/releases/tag/v2.4.0.

Parameters:

  • study (Study) – Target study object.

  • trial (optuna.trial.FrozenTrial) – Target trial object. Take a copy before modifying this object.

  • state (TrialState) – Resulting trial state.

  • values (Sequence[float] | None) – Resulting trial values. Guaranteed to not be None if trial succeeded.

Return type:

None

before_trial(study, trial)[source]

Trial pre-processing.

This method is called before the objective function is called and right after the trial is instantiated. More precisely, this method is called during trial initialization, just before the infer_relative_search_space() call. In other words, it is responsible for pre-processing that should be done before inferring the search space.

Note

Added in v3.3.0 as an experimental feature. The interface may change in newer versions without prior notice. See https://github.com/optuna/optuna/releases/tag/v3.3.0.

Parameters:

  • study (Study) – Target study object.

  • trial (FrozenTrial) – Target trial object.

Return type:

None

infer_relative_search_space(study, trial)[source]

Infer the search space that will be used by relative sampling in the target trial.

This method is called right before sample_relative() method, and the search space returned by this method is passed to it. The parameters not contained in the search space will be sampled by using sample_independent() method.

Parameters:

  • study (Study) – Target study object.

  • trial (FrozenTrial) – Target trial object. Take a copy before modifying this object.

Returns:

A dictionary containing the parameter names and parameter’s distributions.

Return type:

dict[str, BaseDistribution]

See also

Please refer to intersection_search_space() as an implementation of infer_relative_search_space().

reseed_rng()[source]

Reseed sampler’s random number generator.

This method is called by the Study instance if trials are executed in parallel with the option n_jobs>1. In that case, the sampler instance will be replicated including the state of the random number generator, and they may suggest the same values. To prevent this issue, this method assigns a different seed to each random number generator.

Return type:

None

sample_independent(study, trial, param_name, param_distribution)[source]

Sample a parameter for a given distribution.

This method is called only for the parameters not contained in the search space returned by sample_relative() method. This method is suitable for sampling algorithms that do not use relationship between parameters such as random sampling and TPE.

Note

The failed trials are ignored by any build-in samplers when they sample new parameters. Thus, failed trials are regarded as deleted in the samplers’ perspective.

Parameters:

  • study (Study) – Target study object.

  • trial (FrozenTrial) – Target trial object. Take a copy before modifying this object.

  • param_name (str) – Name of the sampled parameter.

  • param_distribution (BaseDistribution) – Distribution object that specifies a prior and/or scale of the sampling algorithm.

Returns:

A parameter value.

Return type:

Any

sample_relative(study, trial, search_space)[source]

Sample parameters in a given search space.

This method is called once at the beginning of each trial, i.e., right before the evaluation of the objective function. This method is suitable for sampling algorithms that use relationship between parameters such as Gaussian Process and CMA-ES.

Note

The failed trials are ignored by any build-in samplers when they sample new parameters. Thus, failed trials are regarded as deleted in the samplers’ perspective.

Parameters:
Returns:

A dictionary containing the parameter names and the values.

Return type:

dict[str, Any]