Learn R Programming

⚠️There's a newer version (1.08) of this package.Take me there.

gptk (version 1.02)

Gaussian Processes Tool-Kit

Description

The gptk package implements a general-purpose toolkit for Gaussian process regression with an RBF covariance function. Based on a MATLAB implementation by Neil D. Lawrence. See inst/doc/index.html for more details.

Copy Link

Version

Install

install.packages('gptk')

Monthly Downloads

51

Version

1.02

License

AGPL-3

Maintainer

Alfredo A Kalaitzis

Last Published

June 17th, 2011

Functions in gptk (1.02)

Extract a parameter vector from a GP model.

optimiDefaultConstraint

Returns function for parameter constraint.

whiteKernParamInit

WHITE kernel parameter initialisation.

Constrains a parameter.

Return default options for GP model.

A function to move the axes crossing point to the origin.

Sample from a Gaussian with a given covariance.

gpPosteriorMeanVar

Mean and variances of the posterior at points given by X.

Plot a contour of the 2D Gaussian distribution with covariance matrix K.

gaussianNoiseOut

Compute the output of the GAUSSIAN noise given the input mean and variance.

demInterpolation

Gaussian Process Interpolation Demo

Wrapper function for GP objective.

kernDiagGradient

Compute the gradient of the kernel's parameters for the diagonal.

Gradient wrapper for a GP model.

Update the vector `alpha' for computing posterior mean quickly.

gpLogLikeGradients

Compute the gradients for the parameters and X.

modelExtractParam

Extract the parameters of a model.

Optimise the given function using (scaled) conjugate gradients.

Sparse objective function gradients wrt Covariance functions for inducing variables.

Gaussian Process Regression Demo

Initialise a kernel structure.

modelGradientCheck

Check gradients of given model.

multiKernParamInit

MULTI kernel parameter initialisation.

Gaussian Process Sampling Demo

Return indices of present data.

rbfKernParamInit

RBF kernel parameter initialisation.

Compute the gradient of the kernel wrt X.

Give the output of a model for given X.

Initialise a noise structure.

Gaussian Process Plotter

Compute the matrix m given the model.

Create a GP model with inducing variables/pseudo-inputs.

gpScaleBiasGradient

Compute the log likelihood gradient wrt the scales.

cmpndNoiseParamInit

CMPND noise parameter initialisation.

gpUpdateKernels

Update the kernels that are needed.

Test the gradients of the likelihood wrt the covariance.

whiteKernDiagGradX

Gradient of WHITE kernel's diagonal with respect to X.

Gradient of RBF kernel with respect to input locations.

Return indices of given block.

modelOutputGrad

Compute derivatives with respect to params of model outputs.

Noise model's parameter initialisation.

Plot Samples from a GP.

Expand a parameter vector into a GP model.

Optimise the inducing variable based kernel.

Gaussian Process Optimisation Demo

Display a model.

Update the representations of A and D associated with the model.

Evaluate the output of an Gaussian process model.

Give the output of the noise model given the mean and variance.

gpLogLikelihood

Compute the log likelihood of a GP.

Gaussian Process 2D Covariance Demo

Compute the gradient wrt the kernel parameters.

Compute the kernel given the parameters and X.

gaussianNoiseParamInit

GAUSSIAN noise parameter initialisation.

Kernel parameter initialisation.

gpMeanFunctionGradient

Compute the log likelihood gradient wrt the scales.

rbfKernDiagGradX

Gradient of RBF kernel's diagonal with respect to X.

Model log-likelihood/objective error function and its gradient.

Run some tests on the specified kernel.

Gradient of WHITE kernel with respect to input locations.

modelExpandParam

Update a model structure with new parameters or update the posterior processes.

cmpndKernParamInit

CMPND kernel parameter initialisation.

Test the gradients of the gpCovGrads function and the gp models.

gpPosteriorSample

Plot Samples from a GP Posterior.