mcmc: Constructor for 'mcmc' class

In simonsays1980/finmix: An R package for Bayesian estimation of finite mixture distributions

Description

Calling mcmc() constructs an object of class mcmc that specifies the hyper-parameters for the MCMC procedure. Each MCMC sampling needs an mcmc object that specifies the way, how MCMC sampling should be performed and what kind and how much of data should be stored.

Usage

mcmc(
  burnin = 0,
  M = 5000,
  startpar = TRUE,
  storeS = 1000,
  storepost = TRUE,
  ranperm = TRUE,
  storeinv = TRUE
)

Arguments

burnin	An integer defining the number of steps in the burn-in phase of Gibbs-sampling.
M	An integer defining the number of steps in Gibbs-sampling to be stored.
startpar	A logical indicating, if starting by sampling the parameters. If FALSE sampling starts by sampling the indicators S.
storeS	An integer specifying how many of the last sampled indicators should be stored in the output.
storepost	A logical indicating if the posterior probabilities should be stored. This becomes for example important for specific relabeling algorithms, but also for analysis.
ranperm	A logical indicating, if random permutation should be used. If TRUE the parameters are permutated randomly between the number of components after each sampling step in MCMC.
storeinv	A logical indicating if the inverse variance-covariance matrices for multivariate normal or Student-t mixtures should be stored.

Value

An object of class mcmc containing all hyper-parameters for MCMC sampling.

See Also

• mcmc for the definition of the mcmc class

• mcmcstart() for setting up all objects for MCMC sampling

• mixturemcmc() for running MCMC sampling for finite mixture models

Examples

f_mcmc <- mcmc()

simonsays1980/finmix documentation built on Dec. 23, 2021, 2:25 a.m.