mh: Metropolis-Hastings MCMC sampling

In MarcoDVisser/MCMCR: Markov-chain monte-carlo methods and rejection sampling in R

Description

Simple protocol for a MCMC sampling algorithm. Given a set of priors, and a likelihood function, the algorithm samples new values for N parameters from a standard normal with a jump variance that is automatically adjusted. The algoritm accepts new proposal proportional to a random fraction.

Usage

mh(dat, start, LL, priors, N, giveup = 100)

Arguments

dat	dataset to be based to the likelihood function
start	a vector of starting values
LL	a likelihood function that accepts the parameters and data
priors	functions that give the prior probability for the parameter
N	total number of sampling rounds (one rounds goes through all parameters once).
giveup	after how many rounds of no new accepted values does the algorithm give up and assume convergence?

Author(s)

Marco D. Visser

MarcoDVisser/MCMCR documentation built on May 7, 2019, 2:49 p.m.