rc.gpf.likelihoods

Contains extensions to gpflow.likelihoods.

Classes

MOGaussian

A non-diagonal, multivariate likelihood, extending gpflow. The code is the multivariate version of gf.likelihoods.Gaussian.

Module Contents

class MOGaussian(variance, **kwargs)

Bases: gpflow.likelihoods.QuadratureLikelihood

A non-diagonal, multivariate likelihood, extending gpflow. The code is the multivariate version of gf.likelihoods.Gaussian.

The Gaussian likelihood is appropriate where uncertainties associated with the data are believed to follow a normal distribution, with constant variance.

Very small uncertainties can lead to numerical instability during the optimization process. A lower bound of 1e-3 is therefore imposed on the likelihood Variance.cholesky_diagonal elements by default.

N(data)

The number of samples in data, assuming the last 2 dimensions have been concatenated to LN.

Return type:

int

split_axis_shape(data)

Split the final data axis length LN into the pair (L,N).

Return type:

Tuple[int, int]