This is wrapper to the NMixMCMC main simulation which allows vectorized evaluation and possibly parallel computation.
THIS FUNCTION IS NOT TO BE CALLED BY ORDINARY USERS.
NMixMCMCwrapper(chain = 1,
scale, prior, inits, Cpar, RJMCMC, CRJMCMC,
actionAll, nMCMC, keep.chains, PED,
dens.zero, lx_w)A list having almost the same components as object returned by
NMixMCMC function.
identification of the chain sampled in a particular call of this function, usually number like 1, 2, ...
a list with the following components
\(n\times p\) matrix with shifted and scaled main limits of observed intervals.
\(n\times p\) matrix with shifted and scaled upper limits of observed intervals.
\(n\times p\) matrix with censoring indicators.
dimension of the response.
number of observations.
a numeric vector with integer prior parameters.
a numeric vector with double precission prior parameters.
a character vector with levels of an optional factor covariate on the mixture weights.
a list specifying how to scale the data before running
MCMC. See argument scale in NMixMCMC
a list specifying prior hyperparameters. See argument
prior in NMixMCMC.
a list of length at least chain. Its
chain-th component is used. Each component of the list should
have the structure of init argument of function
NMixMCMC.
a list specifying parameters for RJ-MCMC.
See argument RJMCMC in NMixMCMC
a numeric vector with parameters for RJ-MCMC.
argument for underlying C++ function.
vector giving the length of MCMC etc.
logical. If FALSE, only summary statistics
are returned in the resulting object. This might be useful in the
model searching step to save some memory.
a logical value which indicates whether the penalized
expected deviance (see Plummer, 2008 for more details)
will be computed (which requires two parallel
chains). Even if keep.chains is FALSE, it is necessary
to keep (for a while) at least some chains to compute PED.
small number (1e-300) to determine whether the contribution to the deviance (\(-\log\) density) is equal to infinity. Such values are trimmed when computing expected deviance.
Arnošt Komárek arnost.komarek@mff.cuni.cz
NMixMCMC.