Gaussian

class gpjax.likelihoods.Gaussian(obs_stddev=1.0, integrator=<gpjax.integrators.AnalyticalGaussianIntegrator object>)

Bases: AbstractLikelihood

Gaussian likelihood object.

Expand for references to gpjax.likelihoods.Gaussian

GPJax / “Hello, GP!”

Migrations / 0.18.x → 1.0.0 / The conditioning API

fit

fit_natgrads

ConjugateModel

conjugate_loocv

StateSpaceConjugateModel

fit

fit_lbfgs

fit_scipy

state_space_mll

Parameters:

• obs_stddev (Any)

• integrator (AbstractIntegrator)

link_function(f)

The link function of the Gaussian likelihood.

Parameters:

f (Float[Array, "..."]) – Function values.

Returns:

The likelihood function.

Return type:

npd.Normal

noise_vector(n)

Per-observation noise variance vector (scalar broadcast for single-output).

Parameters:

n (int)

Return type:

Float[jaxlib._jax.Array, ‘N’] | Float[ndarray, ‘N’]

predict(dist)

Evaluate the Gaussian likelihood at a predictive distribution.

Preserves diagonal scale when the input carries a lineax.DiagonalLinearOperator (including when wrapped in lx.TaggedLinearOperator as emitted by DiagonalKernelComputation / ConstantDiagonalKernelComputation). Always returns GaussianDistribution. This widens the previous return type from numpyro.distributions.MultivariateNormal — see CHANGELOG v0.15.

Parameters:

dist (MultivariateNormal | GaussianDistribution) – The Gaussian process posterior at a finite set of test points.

Returns:

The predictive distribution with observation noise added to the diagonal of the covariance.

Return type:

GaussianDistribution

prepare_targets(y, mx)

Return targets and mean in the format expected by the unified predict/MLL path.

Parameters:

• y (Float[jaxlib._jax.Array, 'N 1'] | Float[ndarray, 'N 1'])

• mx (Float[jaxlib._jax.Array, 'N 1'] | Float[ndarray, 'N 1'])

Return type:

tuple[Float[jaxlib._jax.Array, ‘N 1’] | Float[ndarray, ‘N 1’], Float[jaxlib._jax.Array, ‘N 1’] | Float[ndarray, ‘N 1’]]