fabMix_parallelModels: Function for model-level parallelization

In fabMix: Overfitting Bayesian Mixtures of Factor Analyzers with Parsimonious Covariance and Unknown Number of Components

Description

This function runs multiple copies of the fabMix function in parallel.

Usage

fabMix_parallelModels(model, nChains, dirPriorAlphas, rawData, outDir, Kmax, mCycles, 
	burnCycles, g, h, alpha_sigma, beta_sigma, q, normalize,  
	thinning, zStart, nIterPerCycle, gibbs_z, 
	warm_up_overfitting, warm_up, overfittingInitialization, 
	progressGraphs, gwar, rmDir, parallelModels, lowerTriangular)

Arguments

model	Any subset of "UUU" "CUU" "UCU" "CCU" "UCC" "UUC" "CUC", "CCC" indicating the fitted models.
nChains	The number of parallel heated chains. When 'dirPriorAlphas' is supplied, 'nChains' can be ignored.
dirPriorAlphas	vector of length nChains in the form of an increasing sequence of positive scalars. Each entry contains the (common) prior Dirichlet parameter for the corresponding chain. Default: dirPriorAlphas = c(1, 1 + dN*(2:nChains - 1))/Kmax, where dN = 1, for nChains > 1. Otherwise, dirPriorAlphas = 1/Kmax.
rawData	The observed data as an n\times p matrix. Clustering is performed on the rows of the matrix.
outDir	Name of the output folder. An error is thrown if this directory already exists.
Kmax	Number of components in the overfitted mixture. Default: 20.
mCycles	Number of MCMC cycles. Each cycle consists of nIterPerCycle MCMC iterations. At the end of each cycle a swap of the state of two randomly chosen adjacent chains is attempted.
burnCycles	Number of cycles that will be discarded as burn-in period.
g	Prior parameter g. Default value: g = 0.5.
h	Prior parameter h. Default value: h = 0.5.
alpha_sigma	Prior parameter α. Default value: α = 0.5.
beta_sigma	Prior parameter β. Default value: β = 0.5.
q	A vector containing the number of factors to be fitted.
normalize	Should the observed data be normalized? Default value: TRUE. (Recommended)
thinning	Optional integer denoting the thinning of the keeped MCMC cycles.
zStart	Optional starting value for the allocation vector.
nIterPerCycle	Number of iteration per MCMC cycle. Default value: 10.
gibbs_z	Select the gibbs sampling scheme for updating latent allocations of mixture model. Default value: 1.
warm_up_overfitting	Number of iterations for the overfitting initialization scheme. Default value: 500.
warm_up	Number of iterations that will be used to initialize the models before starting proposing switchings. Default value: 5000.
overfittingInitialization	Logical value indicating whether the chains are initialized via the overfitting initialization scheme. Default: TRUE.
progressGraphs	Logical value indicating whether to plot successive states of the chains while the sampler runs. Default: FALSE.
gwar	Initialization parameter. Default: 0.05.
rmDir	Logical value indicating whether to delete the outDir directory. Default: TRUE.
parallelModels	Model-level parallelization: An optional integer specifying the number of cores that will be used in order to fit in parallel each member of model.
lowerTriangular	logical value indicating whether a lower triangular parameterization should be imposed on the matrix of factor loadings (if TRUE) or not. Default: TRUE.

Value

An object of class fabMix.object (see the fabMix function).

Note

See the fabMix function for examples.

Author(s)

Panagiotis Papastamoulis

fabMix documentation built on Feb. 20, 2020, 1:09 a.m.