stephens1997a_poisson_cc: Stephens (1997a) relabeling algorithm for Poisson mixtures

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

Description

For internal usage only. This function runs the re-labeling algorithm from Stephens (1997a) for MCMC samples of a Poisson mixture distribution. For optimization a Nelder-Mead-Algorithm from the NLopt library is used. This is also the reason why the package depends on the nloptr package which provides a header file for direct access to the C routines.

Usage

stephens1997a_poisson_cc(values1, values2, pars, perm)

Arguments

values1	A matrix containing the sampled component parameters lambda.
values2	A matrix containing the sampled weights.
pars	A vector containing the parameters of the prior distributions of the component parameters and weights. More specifically, the parameters of the Dirichlet distribution for the weights and the shape and rate parameters for the Gamma distributions of the component parameters.
perm	A matrix with all possible permutations of the labels.

Value

A matrix of dimension MxK that holding the optimal labeling.

References

• Stephens, Matthew (1997a), Discussion of "On Bayesian Analysis of Mixtures with an Unknown Number of Components", Journal of the Royal Statistical Society: Series B (Statistical Methodology), 59: 731-792.

See Also

• mcmcpermute() for the calling function

• stephens1997b_poisson_cc() for the re-labeling algorithm from Stephens (1997b)

• stephens1997a_binomial_cc() for the equivalent implementation for mixtures of Binomial distributions

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