BTGM: Performs Bayesian Tensor Graphical Model Estimation

In Rene-Gutierrez/BayTenGraMod: Bayesian Tensor Graphical Model

Description

This function generates nmcmc MCMC samples for a Bayesian Tensor Graphical Model. Each precision matrix is modeled according to a GWishart distribution. The output, returns the samples for each presion matrix as well as the adjacency matrix samples.

Usage

BTGM(
  t,
  b = NULL,
  D = NULL,
  C = NULL,
  beta = NULL,
  burnin = 0,
  nmcmc = 1,
  method = "DMH"
)

Arguments

t	List of Sample Tensors
b	A vector with the Prior degrees of Freedom for each precision matrix. If NULL it sets the prior for each presicion at 3.
D	List of Prior Scale matrices for each precision matrix. If NULL it defaults to the corresponding Identity matrices for each presicion matrix.
C	List of Initial Values of the Precicion Matrices to start the MC markov chain. If NULL each presicion matrix is initialized as the corresponding identity matrix.
beta	A vector containing the prior probability of having and edge between to vertices for each precision. If NULL it sets the value at 0.5 for every edge matrix.
burnin	Number of sample to burn in in the MCMC.
nmcmc	Number of MCMC samples desired as output. That is without considering the burn-in period.
method	Either 'E' for Exchange or 'DMH' for Double Metropolis Hastings. By default is set to 'DMH'.

Value

List containg two other lists. One for the Precison matrices and another one for the adjacency matrices.

samC

A list of arrays for Precicion matrices, the list goes through every precision matrix.

samE

A list of arrays for Adjacency matrices, the list goes through every adjacency matrix.

Author(s)

Rene Gutierrez Marquez

Rene-Gutierrez/BayTenGraMod documentation built on Dec. 12, 2020, 11:24 a.m.