sampling

class gen_classes.sampling.UniformSample(VOCS, random_seed=1, *args, **kwargs)

Samples over the domain specified in the VOCS.

Implements suggest() and ingest() identically to the other generators.

Parameters:

VOCS (VOCS)
random_seed (int)

suggest(num_points=0)

Request the next set of points to evaluate.

Parameters:: num_points (int | None)
Return type:: List[dict]

ingest(results)

Send the results of evaluations to the generator.

Parameters:: results (List[dict])
Return type:: None

class gen_classes.external.sampling.UniformSample(VOCS)

This sampler adheres to the gest-api VOCS interface and data structures (no numpy).

Each variable is a scalar.

Parameters:: VOCS (VOCS)

suggest(n_trials)

Request the next set of points to evaluate.

Returns a list of points in the input space, to be evaluated next. Each element of the list is a separate point. Keys of the dictionary include the name of each input variable specified in VOCS (variables + constants). Values of the dictionaries are scalars unless the variable was declared with an array dtype attribute.

Parameters:

num_points (int | None): Optional number of points to generate. When not provided, the generator decides how many points to return.

Returns:

list[dict]: A list of dictionaries, where each dictionary represents a point to evaluate. Each dictionary contains keys for all variables and constants from the VOCS, and optionally an "_id" key if the generator supports IDs.

Raises:

ValueError: If num_points is specified but the generator cannot produce that exact number of points.

Notes:

When num_points is passed, the generator should return exactly that number of points or raise a ValueError.
When num_points is not passed, the generator decides how many points to return. Different generators will return different numbers of points. For instance, the simplex would return 1 or 3 points. A genetic algorithm could return the whole population. Batched Bayesian optimization would return the batch size (i.e., number of points that can be processed in parallel), which would be specified in the constructor.
Some generators can assign a unique identifier to each generated point (indicated by the returns_id class attribute). If implemented, this identifier should appear in the dictionary under the key "_id".

>>> points = generator.suggest(2)
>>> print(points)
[{"x": 1.2, "y": 0.8}, {"x": -0.2, "y": 0.4}]

>>> points = generator.suggest()  # Generator decides
>>> print(points)
[{"x": 1.2, "y": 0.8}, {"x": -0.2, "y": 0.4}, {"x": 4.3, "y": -0.1}]

ingest(calc_in)

Send the results of evaluations to the generator.

Feeds data (past evaluations) to the generator. Each element of the list is a separate point. Keys of the dictionary must include each named field specified in the VOCS provided to the generator on instantiation (variables, constants, objectives, constraints, and observables).

Parameters:

results (list[dict]): A list of dictionaries, where each dictionary represents an evaluated point. Each dictionary must contain keys for all variables, constants, objectives, constraints, and observables from the VOCS.

Raises:

ValueError: If points are given to ingest with an _id value that is not known internally (for generators that support IDs).

Notes:

Any points provided to the generator via ingest that were not created by the current generator instance should omit the _id field.
If points are given to ingest with an _id value that is not known internally, a ValueError error should be raised.

>>> point = generator.suggest(1)
>>> point
[{"x": 1, "y": 1}]
>>> point[0]["f"] = objective(point[0])
>>> point
[{"x": 1, "y": 1, "f": 2}]
>>> generator.ingest(point)