blackboxopt.optimizers.testing

Tests that can be imported and used to test optimizer implementations against this packages blackbox optimizer interface.

handles_conditional_space(optimizer_class, optimizer_kwargs, seed=None, n_max_evaluations=10)

Check if optimizer handles conditional i.e. hierarchical search spaces.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None
n_max_evaluations	int	Maximum number of evaluation to try	10

Source code in blackboxopt/optimizers/testing.py

def handles_conditional_space(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
    n_max_evaluations: int = 10,
):
    """Check if optimizer handles conditional i.e. hierarchical search spaces.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
        n_max_evaluations: Maximum number of evaluation to try
    """
    space = ps.ParameterSpace()
    space.add(ps.CategoricalParameter("optimizer", ("adam", "sgd")))
    space.add(ps.ContinuousParameter("lr", (0.0001, 0.1), transformation="log"))
    space.add(
        ps.ContinuousParameter("momentum", (0.0, 1.0)),
        lambda optimizer: optimizer == "sgd",
    )
    space.seed(seed)

    opt = _initialize_optimizer(
        optimizer_class, optimizer_kwargs, space=space, seed=seed
    )

    for _ in range(n_max_evaluations):
        es = opt.generate_evaluation_specification()
        objectives = {
            "loss": es.configuration.get("momentum", 1.0) * es.configuration["lr"] ** 2
        }
        if isinstance(opt, MultiObjectiveOptimizer):
            objectives["score"] = -1.0 * es.configuration["lr"] ** 2
        opt.report(
            es.create_evaluation(
                objectives=objectives, constraints={"constraint": 10.0}
            )
        )

handles_reporting_evaluations_list(optimizer_class, optimizer_kwargs, seed=None)

Check if optimizer's report method can process an iterable of evaluations.

All optimizers should be able to allow reporting batches of evaluations. It's up to the optimizer's implementation, if evaluations in a batch are processed one by one like if they were reported individually, or if a batch is handled differently.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None

Source code in blackboxopt/optimizers/testing.py

def handles_reporting_evaluations_list(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
):
    """Check if optimizer's report method can process an iterable of evaluations.

    All optimizers should be able to allow reporting batches of evaluations. It's up to
    the optimizer's implementation, if evaluations in a batch are processed
    one by one like if they were reported individually, or if a batch is handled
    differently.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
    """
    opt = _initialize_optimizer(optimizer_class, optimizer_kwargs, seed=seed)
    evaluations = []
    for i in range(3):
        es = opt.generate_evaluation_specification()

        objectives = {"loss": 0.42 * i}
        if isinstance(opt, MultiObjectiveOptimizer):
            objectives["score"] = float(i)

        evaluation = es.create_evaluation(
            objectives=objectives, constraints={"constraint": 10.0 * i}
        )
        evaluations.append(evaluation)

    opt.report(evaluations)

is_deterministic_when_reporting_shuffled_evaluations(optimizer_class, optimizer_kwargs, seed=None)

Check if determinism isn't affected by the order of initially reported data.

Repeatedly initialize the optimizer with the same parameter space and a fixed seed. Report a set of initial evaluations in randomized order as initial data. Start optimizing and check if the generated configurations for all optimizers are equal.

By doing multiple evaluations, this tests covers effects that become visible after a while, e.g. only after stages got completed in staged iteration samplers.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None

Source code in blackboxopt/optimizers/testing.py

def is_deterministic_when_reporting_shuffled_evaluations(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
):
    """Check if determinism isn't affected by the order of initially reported data.

    Repeatedly initialize the optimizer with the same parameter space and a fixed seed.
    Report a set of initial evaluations in randomized order as initial data. Start
    optimizing and check if the generated configurations for all optimizers are equal.

    By doing multiple evaluations, this tests covers effects that become visible after
    a while, e.g. only after stages got completed in staged iteration samplers.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
    """
    if seed is None:
        seed = 0

    space = ps.ParameterSpace()
    space.add(ps.ContinuousParameter("p1", (0, 1)))
    space.seed(seed)

    def _run_experiment_1d(es):
        x = es.configuration["p1"]
        _x = np.copy(np.atleast_2d(x))
        params = np.array([0.75, 0.0, -10.0, 0.0, 0.0])
        y = np.polyval(params, _x)
        return float(np.squeeze(y))

    runs: Dict[int, Dict] = {0: {}, 1: {}}
    for run_idx, run in runs.items():
        run["evaluations"] = []
        opt = _initialize_optimizer(
            optimizer_class, optimizer_kwargs, space=space, seed=seed
        )

        # Report initial data in different order
        eval_specs = [opt.generate_evaluation_specification() for _ in range(5)]
        run["initial_evaluations"] = [
            es.create_evaluation(
                objectives={"loss": _run_experiment_1d(es)},
                constraints={"constraint": 10.0},
            )
            for es in eval_specs
        ]
        if isinstance(opt, MultiObjectiveOptimizer):
            for e in run["initial_evaluations"]:
                e.objectives["score"] = -1.0 * e.objectives["loss"] ** 2

        shuffle_rng = random.Random(run_idx)
        shuffle_rng.shuffle(run["initial_evaluations"])
        opt.report(run["initial_evaluations"])

        # Start optimizing
        for _ in range(5):
            es = opt.generate_evaluation_specification()

            objectives = {"loss": _run_experiment_1d(es)}
            if isinstance(opt, MultiObjectiveOptimizer):
                objectives["score"] = -1.0 * objectives["loss"] ** 2
            evaluation = es.create_evaluation(
                objectives=objectives, constraints={"constraint": 10.0}
            )

            opt.report(evaluation)
            run["evaluations"].append(evaluation)

    initial_configs_run_0 = [e.configuration for e in runs[0]["initial_evaluations"]]
    initial_configs_run_1 = [e.configuration for e in runs[1]["initial_evaluations"]]

    configs_run_0_as_floats = [e.configuration["p1"] for e in runs[0]["evaluations"]]
    configs_run_1_as_floats = [e.configuration["p1"] for e in runs[1]["evaluations"]]

    assert initial_configs_run_0 != initial_configs_run_1
    np.testing.assert_almost_equal(
        configs_run_0_as_floats, configs_run_1_as_floats, decimal=3
    )

is_deterministic_with_fixed_seed_and_larger_space(optimizer_class, optimizer_kwargs, seed=None)

Check if optimizer is deterministic.

Initialize the optimizer twice with the same parameter space and a fixed seed. For each optimizer run optimization loop n_evaluations times, namely, get an evaluation specification and report a placeholder result back. The list of configurations should be equal for both optimizers.

This tests covers multiple parameter types by using a mixed search space.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None

Source code in blackboxopt/optimizers/testing.py

def is_deterministic_with_fixed_seed_and_larger_space(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
):
    """Check if optimizer is deterministic.

    Initialize the optimizer twice with the same parameter space and a fixed seed.
    For each optimizer run optimization loop `n_evaluations` times, namely,
    get an evaluation specification and report a placeholder result back.
    The list of configurations should be equal for both optimizers.

    This tests covers multiple parameter types by using a mixed search space.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
    """
    if seed is None:
        seed = 42

    n_evaluations = 5
    losses = [0.1, 0.2, 0.3, 0.4, 0.5]

    run_0_configs: List[Evaluation] = []
    run_1_configs: List[Evaluation] = []

    for run_configs in [run_0_configs, run_1_configs]:
        opt = _initialize_optimizer(optimizer_class, optimizer_kwargs, seed=seed)

        for i in range(n_evaluations):
            es = opt.generate_evaluation_specification()

            objectives = {"loss": losses[i]}
            if isinstance(opt, MultiObjectiveOptimizer):
                objectives["score"] = -1.0 * losses[i] ** 2
            evaluation = es.create_evaluation(
                objectives=objectives, constraints={"constraint": 10.0}
            )

            opt.report(evaluation)

            run_configs.append(evaluation.configuration)

    assert len(run_0_configs) == n_evaluations
    assert run_0_configs == run_1_configs

optimize_single_parameter_sequentially_for_n_max_evaluations(optimizer_class, optimizer_kwargs, seed=None, n_max_evaluations=20)

[summary]

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	[description]	required
optimizer_kwargs	dict	[description]	required
n_max_evaluations	int	[description]	20
seed	Optional[int]	(optional) custom seed	None

Returns:

Type	Description
	[description]

Source code in blackboxopt/optimizers/testing.py

def optimize_single_parameter_sequentially_for_n_max_evaluations(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
    n_max_evaluations: int = 20,
):
    """[summary]

    Args:
        optimizer_class: [description]
        optimizer_kwargs: [description]
        n_max_evaluations: [description]
        seed: (optional) custom seed

    Returns:
        [description]
    """

    def quadratic_function(p1):
        return p1**2

    assert issubclass(optimizer_class, Optimizer), (
        "The default test suite is only applicable for implementations of "
        "blackboxopt.base.Optimizer"
    )

    optimizer = _initialize_optimizer(optimizer_class, optimizer_kwargs, seed=seed)

    eval_spec = optimizer.generate_evaluation_specification()

    if issubclass(optimizer_class, MultiObjectiveOptimizer):
        evaluation = eval_spec.create_evaluation(
            objectives={"loss": None, "score": None},
            constraints={"constraint": 10.0},
        )
    else:
        evaluation = eval_spec.create_evaluation(
            objectives={"loss": None}, constraints={"constraint": 10.0}
        )
    optimizer.report(evaluation)

    for _ in range(n_max_evaluations):

        try:
            eval_spec = optimizer.generate_evaluation_specification()
        except OptimizationComplete:
            break

        loss = quadratic_function(p1=eval_spec.configuration["p1"])
        if issubclass(optimizer_class, MultiObjectiveOptimizer):
            evaluation_result = {"loss": loss, "score": -loss}
        else:
            evaluation_result = {"loss": loss}

        evaluation = eval_spec.create_evaluation(
            objectives=evaluation_result, constraints={"constraint": 10.0}
        )
        optimizer.report(evaluation)

raises_evaluation_error_when_reporting_unknown_objective(optimizer_class, optimizer_kwargs, seed=None)

Check if optimizer's report method raises exception in case objective is unknown.

Also make sure that the faulty evaluations (and only those) are included in the exception.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None

Source code in blackboxopt/optimizers/testing.py

def raises_evaluation_error_when_reporting_unknown_objective(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
):
    """Check if optimizer's report method raises exception in case objective is unknown.

    Also make sure that the faulty evaluations (and only those) are included in the
    exception.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
    """
    opt = _initialize_optimizer(optimizer_class, optimizer_kwargs, seed=seed)
    es_1 = opt.generate_evaluation_specification()
    es_2 = opt.generate_evaluation_specification()
    es_3 = opt.generate_evaluation_specification()

    # NOTE: The following is not using pytest.raises because this would add pytest as
    #       a regular dependency to blackboxopt.
    try:
        evaluation_1 = es_1.create_evaluation(
            objectives={"loss": 1}, constraints={"constraint": 10.0}
        )
        evaluation_2 = es_2.create_evaluation(
            objectives={"unknown_objective": 2}, constraints={"constraint": 10.0}
        )
        evaluation_3 = es_3.create_evaluation(
            objectives={"loss": 4}, constraints={"constraint": 10.0}
        )
        evaluations = [evaluation_1, evaluation_2, evaluation_3]
        if isinstance(opt, MultiObjectiveOptimizer):
            for e in evaluations:
                e.objectives["score"] = 0.0
        opt.report(evaluations)

        raise AssertionError(
            f"Optimizer {optimizer_class} did not raise an ObjectivesError when a "
            + "result including an unknown objective name was reported."
        )

    except EvaluationsError as exception:
        invalid_evaluations = [e for e, _ in exception.evaluations_with_errors]
        assert len(invalid_evaluations) == 1
        assert evaluation_2 in invalid_evaluations

respects_fixed_parameter(optimizer_class, optimizer_kwargs, seed=None)

Check if optimizer's generated evaluation specifications contain the values a parameter in the search space was fixed to.

Parameters:

Name	Type	Description	Default
optimizer_class	Union[Type[blackboxopt.base.SingleObjectiveOptimizer], Type[blackboxopt.base.MultiObjectiveOptimizer]]	Optimizer to test.	required
optimizer_kwargs	dict	Expected to contain additional arguments for initializing the optimizer. (search_space and objective(s) are set automatically by the test.)	required
seed	Optional[int]	(optional) custom seed	None

Source code in blackboxopt/optimizers/testing.py

def respects_fixed_parameter(
    optimizer_class: Union[
        Type[SingleObjectiveOptimizer], Type[MultiObjectiveOptimizer]
    ],
    optimizer_kwargs: dict,
    seed: Optional[int] = None,
):
    """Check if optimizer's generated evaluation specifications contain the values
    a parameter in the search space was fixed to.

    Args:
        optimizer_class: Optimizer to test.
        optimizer_kwargs: Expected to contain additional arguments for initializing
            the optimizer. (`search_space` and `objective(s)` are set automatically
            by the test.)
        seed: (optional) custom seed
    """
    space = ps.ParameterSpace()
    space.add(ps.ContinuousParameter("my_fixed_param", (-10.0, 200.0)))
    space.add(ps.ContinuousParameter("x", (-2.0, 2.0)))
    space.seed(seed)

    fixed_value = 1.0
    space.fix(my_fixed_param=fixed_value)
    opt = _initialize_optimizer(
        optimizer_class, optimizer_kwargs, space=space, seed=seed
    )
    for _ in range(5):
        es = opt.generate_evaluation_specification()
        assert es.configuration["my_fixed_param"] == fixed_value

        objectives = {"loss": es.configuration["x"] ** 2}
        if isinstance(opt, MultiObjectiveOptimizer):
            objectives["score"] = -objectives["loss"]

        opt.report(
            es.create_evaluation(
                objectives=objectives,
                constraints={"constraint": 10.0},
            )
        )