sampler: Produce samples from the approximated joint posterior for the...

In inbo/INLA: Full Bayesian Analysis of Latent Gaussian Models using Integrated Nested Laplace Approximations

Description

Produce samples from the approximated joint posterior for the hyperparameters

Usage

inla.hyperpar.sample(n, result, intern=FALSE, improve.marginals = FALSE)

Arguments

n	Integer. Number of samples required.
result	An inla-object, f.ex the output from an inla-call.
intern	Logical. If TRUE then produce samples in the internal scale for the hyperparmater, if FALSE then produce samples in the user-scale. (For example log-precision (intern) and precision (user-scale))
improve.marginals	Logical. If TRUE, then improve the samples taking into account possible better marginal estimates for the hyperparameters in result.

Value

A matrix where each sample is a row. The contents of the column is described in the rownames.

Author(s)

Havard Rue [e-mail]

Examples

n = 100
r = inla(y ~ 1 + f(idx), data = data.frame(y=rnorm(n), idx = 1:n))
ns = 500
x = inla.hyperpar.sample(ns, r)

rr = inla.hyperpar(r)
xx = inla.hyperpar.sample(ns, rr, improve.marginals=TRUE)

inbo/INLA documentation built on Dec. 6, 2019, 9:51 a.m.