runlmc.lmc.likelihood module

class runlmc.lmc.likelihood.ApproxLMCLikelihood(functional_kernel, grid_kern, grid_dists, interpolants, Ys, deriv)

Bases: runlmc.lmc.likelihood.LMCLikelihood

alpha()

class runlmc.lmc.likelihood.ExactLMCLikelihood(functional_kernel, Xs, Ys)

Bases: runlmc.lmc.likelihood.LMCLikelihood

alpha()

static kernel_from_indices(Xs, Zs, functional_kernel)

Computes the dense, exact kernel matrix for an LMC kernel specified by functional_kernel. The kernel matrix that is computed is relative to the kernel application to pairs from the Cartesian product Xs and Zs.

class runlmc.lmc.likelihood.LMCLikelihood(functional_kernel, Ys)

Bases: object

Separate hyperparameter-based likelihood differentiation from the model class for separation of concerns. Different sub-classes may implement the below methods differently, with different asymptotic performance properties.

alpha()

coreg_diags_gradients()

coreg_vec_gradients()

kernel_gradients()

noise_gradient()