sample.gibbs: Sampling method using single-site Gibbs sampler

In CRF: Conditional Random Fields

Description

Generating samples from the distribution

Usage

sample.gibbs(
  crf,
  size,
  burn.in = 1000,
  start = apply(crf$node.pot, 1, which.max)
)

Arguments

crf	The CRF
size	The sample size
burn.in	The number of samples at the beginning that will be discarded
start	An initial configuration

Details

Approximate sampling using a single-site Gibbs sampler

Value

This function will return a matrix with size rows and crf$n.nodes columns, in which each row is a sampled configuration.

Examples

library(CRF)
data(Small)
s <- sample.gibbs(Small$crf, 100)

CRF documentation built on Dec. 2, 2019, 1:11 a.m.