GauPro_Gauss {GauPro}

Corr Gauss GP using inherited optim

Description

Corr Gauss GP using inherited optim

Corr Gauss GP using inherited optim

Format

R6Class object.

Value

Object of R6Class with methods for fitting GP model.

Super class

GauPro::GauPro -> GauPro_Gauss

Public fields

corr

Name of correlation

theta

Correlation parameters

theta_length

Length of theta

theta_map

Map for theta

theta_short

Short vector for theta

separable

Are the dimensions separable?

Methods

Public methods

• GauPro_Gauss$new()

• GauPro_Gauss$corr_func()

• GauPro_Gauss$deviance_theta()

• GauPro_Gauss$deviance_theta_log()

• GauPro_Gauss$deviance()

• GauPro_Gauss$deviance_grad()

• GauPro_Gauss$deviance_fngr()

• GauPro_Gauss$deviance_log()

• GauPro_Gauss$deviance_log2()

• GauPro_Gauss$deviance_log_grad()

• GauPro_Gauss$deviance_log2_grad()

• GauPro_Gauss$deviance_log2_fngr()

• GauPro_Gauss$get_optim_functions()

• GauPro_Gauss$param_optim_lower()

• GauPro_Gauss$param_optim_upper()

• GauPro_Gauss$param_optim_start()

• GauPro_Gauss$param_optim_start0()

• GauPro_Gauss$param_optim_jitter()

• GauPro_Gauss$update_params()

• GauPro_Gauss$grad()

• GauPro_Gauss$grad_dist()

• GauPro_Gauss$hessian()

• GauPro_Gauss$print()

• GauPro_Gauss$clone()

Inherited methods

• GauPro::GauPro$cool1Dplot()
• GauPro::GauPro$deviance_searchnug()
• GauPro::GauPro$fit()
• GauPro::GauPro$grad_norm()
• GauPro::GauPro$initialize_GauPr()
• GauPro::GauPro$loglikelihood()
• GauPro::GauPro$nugget_update()
• GauPro::GauPro$optim()
• GauPro::GauPro$optimRestart()
• GauPro::GauPro$plot()
• GauPro::GauPro$plot1D()
• GauPro::GauPro$plot2D()
• GauPro::GauPro$pred()
• GauPro::GauPro$pred_LOO()
• GauPro::GauPro$pred_mean()
• GauPro::GauPro$pred_meanC()
• GauPro::GauPro$pred_one_matrix()
• GauPro::GauPro$pred_var()
• GauPro::GauPro$predict()
• GauPro::GauPro$sample()
• GauPro::GauPro$update()
• GauPro::GauPro$update_K_and_estimates()
• GauPro::GauPro$update_corrparams()
• GauPro::GauPro$update_data()
• GauPro::GauPro$update_nugget()

Method new()

Create GauPro object

Usage

GauPro_Gauss$new(
  X,
  Z,
  verbose = 0,
  separable = T,
  useC = F,
  useGrad = T,
  parallel = FALSE,
  nug = 1e-06,
  nug.min = 1e-08,
  nug.est = T,
  param.est = T,
  theta = NULL,
  theta_short = NULL,
  theta_map = NULL,
  ...
)

Arguments

X

Matrix whose rows are the input points

Z

Output points corresponding to X

verbose

Amount of stuff to print. 0 is little, 2 is a lot.

separable

Are dimensions separable?

useC

Should C code be used when possible? Should be faster.

useGrad

Should the gradient be used?

parallel

Should code be run in parallel? Make optimization faster but uses more computer resources.

nug

Value for the nugget. The starting value if estimating it.

nug.min

Minimum allowable value for the nugget.

nug.est

Should the nugget be estimated?

param.est

Should the kernel parameters be estimated?

theta

Correlation parameters

theta_short

Correlation parameters, not recommended

theta_map

Correlation parameters, not recommended

...

Not used

Method corr_func()

Correlation function

Usage

GauPro_Gauss$corr_func(x, x2 = NULL, theta = self$theta)

Arguments

x

First point

x2

Second point

theta

Correlation parameter

Method deviance_theta()

Calculate deviance

Usage

GauPro_Gauss$deviance_theta(theta)

Arguments

theta

Correlation parameter

Method deviance_theta_log()

Calculate deviance

Usage

GauPro_Gauss$deviance_theta_log(beta)

Arguments

beta

Correlation parameter on log scale

Method deviance()

Calculate deviance

Usage

GauPro_Gauss$deviance(theta = self$theta, nug = self$nug)

Arguments

theta

Correlation parameter

nug

Nugget

Method deviance_grad()

Calculate deviance gradient

Usage

GauPro_Gauss$deviance_grad(
  theta = NULL,
  nug = self$nug,
  joint = NULL,
  overwhat = if (self$nug.est) "joint" else "theta"
)

Arguments

theta

Correlation parameter

nug

Nugget

joint

Calculate over theta and nug at same time?

overwhat

Calculate over theta and nug at same time?

Method deviance_fngr()

Calculate deviance and gradient at same time

Usage

GauPro_Gauss$deviance_fngr(
  theta = NULL,
  nug = NULL,
  overwhat = if (self$nug.est) "joint" else "theta"
)

Arguments

theta

Correlation parameter

nug

Nugget

overwhat

Calculate over theta and nug at same time?

joint

Calculate over theta and nug at same time?

Method deviance_log()

Calculate deviance gradient

Usage

GauPro_Gauss$deviance_log(beta = NULL, nug = self$nug, joint = NULL)

Arguments

beta

Correlation parameter on log scale

nug

Nugget

joint

Calculate over theta and nug at same time?

Method deviance_log2()

Calculate deviance on log scale

Usage

GauPro_Gauss$deviance_log2(beta = NULL, lognug = NULL, joint = NULL)

Arguments

beta

Correlation parameter on log scale

lognug

Log of nugget

joint

Calculate over theta and nug at same time?

Method deviance_log_grad()

Calculate deviance gradient on log scale

Usage

GauPro_Gauss$deviance_log_grad(
  beta = NULL,
  nug = self$nug,
  joint = NULL,
  overwhat = if (self$nug.est) "joint" else "theta"
)

Arguments

beta

Correlation parameter

nug

Nugget

joint

Calculate over theta and nug at same time?

overwhat

Calculate over theta and nug at same time?

Method deviance_log2_grad()

Calculate deviance gradient on log scale

Usage

GauPro_Gauss$deviance_log2_grad(
  beta = NULL,
  lognug = NULL,
  joint = NULL,
  overwhat = if (self$nug.est) "joint" else "theta"
)

Arguments

beta

Correlation parameter

lognug

Log of nugget

joint

Calculate over theta and nug at same time?

overwhat

Calculate over theta and nug at same time?

Method deviance_log2_fngr()

Calculate deviance and gradient on log scale

Usage

GauPro_Gauss$deviance_log2_fngr(
  beta = NULL,
  lognug = NULL,
  joint = NULL,
  overwhat = if (self$nug.est) "joint" else "theta"
)

Arguments

beta

Correlation parameter

lognug

Log of nugget

joint

Calculate over theta and nug at same time?

overwhat

Calculate over theta and nug at same time?

Method get_optim_functions()

Get optimization functions

Usage

GauPro_Gauss$get_optim_functions(param_update, nug.update)

Arguments

param_update

Should the parameters be updated?

nug.update

Should the nugget be updated?

Method param_optim_lower()

Lower bound of params

Usage

GauPro_Gauss$param_optim_lower()

Method param_optim_upper()

Upper bound of params

Usage

GauPro_Gauss$param_optim_upper()

Method param_optim_start()

Start value of params for optim

Usage

GauPro_Gauss$param_optim_start()

Method param_optim_start0()

Start value of params for optim

Usage

GauPro_Gauss$param_optim_start0()

Method param_optim_jitter()

Jitter value of params for optim

Usage

GauPro_Gauss$param_optim_jitter(param_value)

Arguments

param_value

param value to add jitter to

Method update_params()

Update value of params after optim

Usage

GauPro_Gauss$update_params(restarts, param_update, nug.update)

Arguments

restarts

Number of restarts

param_update

Are the params being updated?

nug.update

Is the nugget being updated?

Method grad()

Calculate the gradient

Usage

GauPro_Gauss$grad(XX)

Arguments

XX

Points to calculate grad at

Method grad_dist()

Calculate the gradient distribution

Usage

GauPro_Gauss$grad_dist(XX)

Arguments

XX

Points to calculate grad at

Method hessian()

Calculate the hessian

Usage

GauPro_Gauss$hessian(XX, useC = self$useC)

Arguments

XX

Points to calculate grad at

useC

Should C code be used to speed up?

Method print()

Print this object

Usage

GauPro_Gauss$print()

Method clone()

The objects of this class are cloneable with this method.

Usage

GauPro_Gauss$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

n <- 12
x <- matrix(seq(0,1,length.out = n), ncol=1)
y <- sin(2*pi*x) + rnorm(n,0,1e-1)
gp <- GauPro_Gauss$new(X=x, Z=y, parallel=FALSE)