mcmcsae (version 0.7.4)

compute_GMRF_matrices: Compute (I)GMRF incidence, precision and restriction matrices corresponding to a generic model component

Description

This function computes incidence, precision and restriction matrices, or a subset thereof, for a Gaussian Markov Random Field (GMRF). A GMRF is specified by a formula passed to the factor argument, in the same way as for the factor argument of gen.

Usage

compute_GMRF_matrices(
  factor,
  data,
  D = TRUE,
  Q = TRUE,
  R = TRUE,
  cols2remove = NULL,
  remove.redundant.R.cols = TRUE,
  enclos = .GlobalEnv,
  n.parent = 1L,
  ...
)

Value

A list containing some or all of the components D (incidence matrix),

Q (precision matrix) and R (restriction matrix).

Arguments

factor

factor formula of a generic model component, see gen.

data

data frame to be used in deriving the matrices.

D

if TRUE compute the incidence matrix.

Q

if TRUE compute the precision matrix.

R

if TRUE compute the restriction matrix.

cols2remove

if an integer vector is passed, the dimensions (columns of D, rows and columns of Q and rows of R) that are removed. This can be useful in the case of empty domains.

remove.redundant.R.cols

whether to test for and remove redundant restrictions from restriction matrix R

enclos

enclosure to look for objects not found in data.

n.parent

for internal use; in case of custom factor, the number of frames up the calling stack in which to evaluate any custom matrices

...

further arguments passed to economizeMatrix.

Examples

n <- 1000
dat <- data.frame(
  x = rnorm(n),
  f1 = factor(sample(1:50, n, replace=TRUE)),
  f2 = factor(sample(1:10, n, replace=TRUE))
)
mats <- compute_GMRF_matrices(~ f1 * RW1(f2), dat)
str(mats)