mixglasso_init: mixglasso_init

Description

mixglasso_init (initialization and lambda set by user)

Usage

mixglasso_init(x, n.comp, lambda, u.init, mix.prob.init, gamma = 0.5, pen = "glasso.parcor", penalize.diagonal = FALSE, term = 10^{     -3 }, miniter = 5, maxiter = 1000, min.compsize = 5, show.trace = FALSE)

Arguments

Input data matrix

n.comp

Number of mixture components

lambda

Regularization parameter

u.init

Initial responsibilities

mix.prob.init

Initial component probablities

gamma

Determines form of penalty

pen

Determines form of penalty: glasso.parcor (default), glasso.invcov, glasso.invcor

penalize.diagonal

Should the diagonal of the inverse covariance matrix be penalized ? Default=FALSE (recommended)

term

Termination criterion of EM algorithm. Default=10^-3

miniter

Minimal number of EM iteration before 'stop EM if any(compsize)

maxiter

Maximal number of EM iteration. Default=1000

min.compsize

Stop EM if any(compsize)

show.trace

Should information during execution be printed ? Default=FALSE

Value

mix.prob: Component probabilities
Mu: Component specific mean vectors
Sig: Component specific covariance matrices
SigInv: Component specific inverse covariance matrices
iter: Number of EM iterations
loglik: Log-likelihood
bic: -loglik+log(n)*DF/2
mmdl: -loglik+penmmdl/2
u: Component responsibilities
comp: Component assignments
compsize: Size of components
pi.comps: Component probabilities
warn: Warnings during EM algorithm

Details

This function runs mixglasso; requires initialization (u.init,mix.prob.init)