Posterior

class gpjax.conditioning.Posterior

Bases: Module

A conditioned Gaussian process, \(p(f \mid \mathcal{D})\).

The result of conditioning a joint model on data. Immutable: the factorisation of the training covariance is computed once at condition time and cached on this object; every query is a view of it. Query the process at test inputs by calling it:

posterior = model.condition(train_data)   # or: model | train_data
predictive = posterior(test_inputs)

predict(test_inputs, train_data=None, *, covariance='dense')

Sugar for calling the posterior: predict(t) == self(t).

Retained for signature compatibility with the pre-v1.0 API; train_data is accepted and ignored — this process is already conditioned on its training set.

Parameters:

• test_inputs (Num[jaxlib._jax.Array, 'N D'] | Num[ndarray, 'N D'])

• train_data (Dataset | None)

• covariance (Literal['dense', 'diagonal'])

Return type:

GaussianDistribution