ExpImp: Import and export of hetGP objects

In hetGP: Heteroskedastic Gaussian Process Modeling and Design under Replication

Import and export of hetGP objects

Description

Functions to make hetGP objects lighter before exporting them, and to reverse this after import. The rebuild function may also be used to obtain more robust inverse of covariance matrices using ginv.

Usage

rebuild(object, robust)

## S3 method for class 'homGP'
rebuild(object, robust = FALSE)

strip(object)

## S3 method for class 'hetGP'
rebuild(object, robust = FALSE)

## S3 method for class 'homTP'
rebuild(object, robust = FALSE)

## S3 method for class 'hetTP'
rebuild(object, robust = FALSE)

Arguments

object	homGP or homTP model without slot Ki (inverse covariance matrix), or hetGP or hetTP model without slot Ki or Kgi
robust	if TRUE ginv is used for matrix inversion, otherwise it is done via Cholesky.

Value

object with additional or removed slots.

Examples

set.seed(32)
## motorcycle data
library(MASS)
X <- matrix(mcycle$times, ncol = 1)
Z <- mcycle$accel
## Model fitting
model <- mleHetGP(X = X, Z = Z, lower = 0.1, upper = 50)

# Check size
object.size(model)

# Remove internal elements, e.g., to save it
model <- strip(model)

# Check new size
object.size(model)

# Now rebuild model, and use ginv instead
model <- rebuild(model, robust = TRUE)
object.size(model)

hetGP documentation built on Oct. 3, 2023, 1:07 a.m.