package gpr

You can search for identifiers within the package.

in-package search v0.2.0

Module `Deriv.Inputs`Source

Derivatives of the (cross-) covariance matrix of inputs.

Sourcetype diag

Representation of precomputed data for calculating the derivative of the diagonal of the covariance matrix of inputs.

Sourcetype cross

Representation of precomputed data for calculating the derivative of the cross-covariance matrix between inputs and inducing inputs.

Sourceval calc_shared_diag : Eval.Kernel.t -> Eval.Inputs.t -> Lacaml.D.vec * diag

calc_shared_diag kernel inputs

returns
the pair (eval, diag), where eval is the diagonal of the covariance matrix of inputs for kernel, and diag is the precomputed data needed for taking derivatives.

val calc_shared_cross : 
  Eval.Kernel.t ->
  inputs:Eval.Inputs.t ->
  inducing:Eval.Inducing.t ->
  Lacaml.D.mat * cross

calc_shared_cross kernel ~inputs ~inducing

returns
the pair (eval, cross), where eval is the cross-covariance matrix of inputs and inducing inputs for kernel, and diag is the precomputed data needed for taking derivatives.

Sourceval calc_deriv_diag : diag -> Hyper.t -> Interfaces.diag_deriv

calc_deriv_diag diag hyper

returns
the derivative of the diagonal of the covariance matrix of inputs given precomputed data diag and the hyper-variable.

Sourceval calc_deriv_cross : cross -> Hyper.t -> Interfaces.mat_deriv

calc_deriv_cross cross hyper

returns
the derivative of the cross-covariance matrix of the inputs and inducing inputs given precomputed data cross and the hyper-variable.