emukit.multi_fidelity.kernels package

Submodules

Contains kernel for use with Linear Multi Fidelity model

class emukit.multi_fidelity.kernels.linear_multi_fidelity_kernel.LinearMultiFidelityKernel(*args, **kw)

Bases: CombinationKernel

K(X, X2=None)

Covariance matrix

See section 2.5 of Kennedy & O’Hagan paper for details

Kdiag(X)

Diagonal of covariance matrix

See section 2.5 of Kennedy & O’Hagan paper for details

gradients_X(dL_dK, X, X2=None): Gradients of likelihood wrt X. Gradient wrt fidelity index are set to 0

gradients_X_diag(dL_dKdiag, X): Gradients of likelihood wrt X. Gradient wrt fidelity index are set to 0

update_gradients_diag(dL_dKdiag, X): Gradients of likelihood wrt model hyper-parameters

update_gradients_full(dL_dK, X, X2=None): Gradients of likelihood wrt model hyper-parameters given dL_dK.