Module bayesian 

Gaussian-process Bayesian optimizers for expensive objectives. Bayesian optimisation for expensive bounded continuous objectives.

The implementation is intentionally domain-agnostic. It models objective values with a Gaussian-process surrogate over normalised [0, 1]^n coordinates, uses a Matérn-5/2 kernel with ARD lengthscales, and proposes one or more candidates per iteration through EI or Monte-Carlo batch q-EI/qEHVI acquisition.

Structs

BayesOptConfig

Configuration for bayesian_optimization.

BayesOptIntermediate

Per-iteration callback payload for bayesian_optimization.

BayesOptParetoReport

Result of a bayesian_multi_objective run.

BayesOptReport

Result of a bayesian_optimization run.

BayesParetoSolution

Pareto solution returned by BayesOptParetoReport.

Enums

BayesAcquisition

Acquisition strategy used to select candidates from the surrogate.

Functions

bayesian_multi_objective

Minimise a vector objective with Monte-Carlo EHVI.

bayesian_optimization

Minimise f with Gaussian-process Bayesian optimisation.

Type Aliases

BayesOptCallback

Callback type used by BayesOptConfig.