Thompson Sampling

gpjax.decision_making.utility_functions.thompson_sampling

ThompsonSampling dataclass

Bases: AbstractSinglePointUtilityFunctionBuilder

Form a utility function by drawing an approximate sample from the posterior, using decoupled sampling as introduced in Wilson et. al. (2020). Note that we return the negative of the sample as the utility function, as utility functions are maximised.

Note that this is a single batch utility function, as it doesn't support classical batching. However, Thompson sampling can be used in a batched setting by drawing a batch of different samples from the GP posterior. This can be done by calling build_utility_function with different keys, an example of which can be seen in the ask method of the UtilityDrivenDecisionMaker class. The samples can then be optimised sequentially.

Attributes:

Name	Type	Description
num_features	int	The number of random Fourier features to use when drawing the approximate sample from the posterior. Defaults to 100.

num_features: int = 100 class-attribute instance-attribute

check_objective_present(posteriors: Mapping[str, AbstractPosterior], datasets: Mapping[str, Dataset]) -> None

Check that the objective posterior and dataset are present in the posteriors and datasets.

Parameters:

Name	Type	Description	Default
posteriors	Mapping[str, AbstractPosterior]	Dictionary of posteriors to be	required
datasets	Mapping[str, Dataset]	Dictionary of datasets which may be used	required

Raises:

Type	Description
ValueError	If the objective posterior or dataset are not present in the

__init__(num_features: int = 100) -> None

__post_init__()

build_utility_function(posteriors: Mapping[str, ConjugatePosterior], datasets: Mapping[str, Dataset], key: KeyArray) -> SinglePointUtilityFunction

Draw an approximate sample from the posterior of the objective model and return the negative of this sample as a utility function, as utility functions are maximised.

Parameters:

Name	Type	Description	Default
posteriors	Mapping[str, AbstractPosterior]	Dictionary of posteriors to be	required
datasets	Mapping[str, Dataset]	Dictionary of datasets which may be used	required
key	KeyArray	JAX PRNG key used for random number generation. This can be	required

Returns:

Name	Type	Description
SinglePointUtilityFunction	SinglePointUtilityFunction	An appproximate sample from the objective model
	SinglePointUtilityFunction	posterior to to be maximised in order to decide which point to query
	SinglePointUtilityFunction	next.