sammo.search_op

This module contains a variety of search operators that can be used to define a discrete search space for GridSearch or define a set of initial candidates for other search algorithms.

Module Contents

Functions

one_of	Search operator for selecting one of the given candidates.
many_of	Search operator for n choose k.
permutate	Search operator for permutating a list of components.
optional	Search operator for making a component optional.
get_points_from_search_space	Materialize a number of points from a search space.
get_first_point	Return the first value of the enumerated search space.

API

sammo.search_op.one_of(candidates: beartype.typing.Iterable, name: str | None = None) → beartype.typing.Any

Search operator for selecting one of the given candidates.

Parameters:

• candidates – The list of candidates to choose from.

• name – The name of the operator for later reference.

sammo.search_op.many_of(num_choices: int, candidates: beartype.typing.Iterable, name: str | None = None) → beartype.typing.Any

Search operator for n choose k.

Parameters:

• num_choices – The number of candidates to choose.

• candidates – The list of candidates to choose from.

• name – The name of the operator for later reference.

sammo.search_op.permutate(candidates: beartype.typing.Iterable, name: str | None = None) → beartype.typing.Any

Search operator for permutating a list of components.

Parameters:

• candidates – The list of components to permute.

• name – The name of the operator for later reference.

sammo.search_op.optional(candidate, name=None) → beartype.typing.Any

Search operator for making a component optional.

Parameters:

• val – The value to include or exclude.

• name – The name of the operator for later reference.

sammo.search_op.get_points_from_search_space(search_space: beartype.typing.Callable | sammo.base.Component | list | dict, n_points: int, sample: bool = False, seed: int = 42, return_names: bool = False) → list[sammo.base.Component]

Materialize a number of points from a search space.

Parameters:

• search_space – Search space, either represented as function or a single Output class.

• n_points – Number of points to materialize.

• sample – Whether to sample from the search space or enumerate and return first n_points.

• seed – Random seed for sampling.

• return_names – Whether to return the names of the points.

sammo.search_op.get_first_point(search_space: beartype.typing.Callable | sammo.base.Component | list | dict) → sammo.base.Component

Return the first value of the enumerated search space.

Parameters:

search_space – Search space, either represented as function or a single Output class.