factory-choleskyFactory: Function Factory for Generating chol Function with unified...

In GeneralizedWendland: Fully Parameterized Generalized Wendland Covariance Function

Function Factory for Generating chol Function with unified arguments

Description

A function factory which returns a function of the form function(Sigma) which performs a cholesky decomposition using an approach tailored to the type of input Sigma. Currently works for vectors, matrices, spam objects, and dgCMatrix objects from the Matrix package.

Usage

choleskyFactory(chol.args = list(), Rstruct = NULL)

Arguments

chol.args	A list of optional settings for a cholesky function.
Rstruct	A spam.chol.NgPeyton object which represents the sparsity structure of covariance matrix Sigma.

Details

The output of choleskyFactory is intended to replace calls to chol.default or chol.spam. The object type is determined during runtime, after which the appropriate function is called to obtain the cholesky decomposition. For spam objects, the function attempts to use update.spam.chol.NgPeyton if Rstruct is specified, and upon failure defaults to chol.spam. The result is then assigned in the execution environment of choleskyFactory, so that Rstruct will be defined in the next call.

Value

A function of the form function(Sigma).

Author(s)

Thomas Caspar Fischer

References

Hadley Wickham (2015) Advanced R, CRC Press. Reinhard Furrer and Roman Flury and Florian Gerber (2022) spam: SPArse Matrix, R package version 2.8-0.

See Also

chol, chol.spam, update.spam.chol.NgPeyton

Examples

set.seed(1234)
locations <- data.frame(x = runif(10), y = runif(10))
theta <- c(0.5,1,1,0,0)
dmat <- as.matrix(dist(locations))
Sigma <- cov.wendland(dmat, theta)

cholFun <- choleskyFactory(chol.args = list())
cholD <- cholFun(Sigma)

cholFun <- choleskyFactory(chol.args = list(pivot = TRUE))
cholD_pivot <- cholFun(Sigma)

cholFun <- choleskyFactory(chol.args = list(pivot = "RCM"))
cholS_RCM <- cholFun(spam::as.spam(Sigma))

GeneralizedWendland documentation built on June 22, 2022, 9:06 a.m.