GHS: Graphical Horseshoe MCMC Sampler Using Data Augmented Block Gibbs Sampler

Draw posterior samples to estimate the precision matrix for multivariate Gaussian data. Posterior means of the samples is the graphical horseshoe estimate by Li, Bhadra and Craig(2017) <arXiv:1707.06661>. The function uses matrix decomposition and variable change from the Bayesian graphical lasso by Wang(2012) <doi:10.1214/12-BA729>, and the variable augmentation for sampling under the horseshoe prior by Makalic and Schmidt(2016) <arXiv:1508.03884>. Structure of the graphical horseshoe function was inspired by the Bayesian graphical lasso function using blocked sampling, authored by Wang(2012) <doi:10.1214/12-BA729>.

Getting started

Package details

AuthorAshutosh Srivastava<[email protected]>, Anindya Bhadra<[email protected]>
MaintainerAshutosh Srivastava<[email protected]>
LicenseGPL-2
Version0.1
Package repositoryView on CRAN
Installation Install the latest version of this package by entering the following in R:
install.packages("GHS")

Try the GHS package in your browser

Any scripts or data that you put into this service are public.

GHS documentation built on Oct. 31, 2018, 1:04 a.m.