MultiBUGS: A Parallel Implementation of the BUGS Modeling Framework for Faster Bayesian Inference
Dublin Core
Title
MultiBUGS: A Parallel Implementation of the BUGS Modeling Framework for Faster Bayesian Inference
Subject
BUGS, parallel computing, Markov chain Monte Carlo, Gibbs sampling, Bayesian
analysis, hierarchical models, directed acyclic graph.
analysis, hierarchical models, directed acyclic graph.
Description
MultiBUGS is a new version of the general-purpose Bayesian modeling software BUGS
that implements a generic algorithm for parallelizing Markov chain Monte Carlo (MCMC)
algorithms to speed up posterior inference of Bayesian models. The algorithm parallelizes evaluation of the product-form likelihoods formed when a parameter has many
children in the directed acyclic graph (DAG) representation; and parallelizes sampling
of conditionally-independent sets of parameters. A heuristic algorithm is used to decide
which approach to use for each parameter and to apportion computation across computational cores. This enables MultiBUGS to automatically parallelize the broad range of
statistical models that can be fitted using BUGS-language software, making the dramatic
speed-ups of modern multi-core computing accessible to applied statisticians, without
requiring any experience of parallel programming. We demonstrate the use of MultiBUGS on simulated data designed to mimic a hierarchical e-health linked-data study
of methadone prescriptions including 425,112 observations and 20,426 random effects.
Posterior inference for the e-health model takes several hours in existing software, but
MultiBUGS can perform inference in only 28 minutes using 48 computational cores.
that implements a generic algorithm for parallelizing Markov chain Monte Carlo (MCMC)
algorithms to speed up posterior inference of Bayesian models. The algorithm parallelizes evaluation of the product-form likelihoods formed when a parameter has many
children in the directed acyclic graph (DAG) representation; and parallelizes sampling
of conditionally-independent sets of parameters. A heuristic algorithm is used to decide
which approach to use for each parameter and to apportion computation across computational cores. This enables MultiBUGS to automatically parallelize the broad range of
statistical models that can be fitted using BUGS-language software, making the dramatic
speed-ups of modern multi-core computing accessible to applied statisticians, without
requiring any experience of parallel programming. We demonstrate the use of MultiBUGS on simulated data designed to mimic a hierarchical e-health linked-data study
of methadone prescriptions including 425,112 observations and 20,426 random effects.
Posterior inference for the e-health model takes several hours in existing software, but
MultiBUGS can perform inference in only 28 minutes using 48 computational cores.
Creator
Robert J. B. Goudie
Source
https://www.jstatsoft.org/article/view/v095i07
Publisher
University of Cambridge
Date
October 2020
Contributor
Fajar bagus W
Format
PDF
Language
Inggris
Type
Text
Files
Collection
Citation
Robert J. B. Goudie, “MultiBUGS: A Parallel Implementation of the BUGS Modeling Framework for Faster Bayesian Inference,” Repository Horizon University Indonesia, accessed April 5, 2025, https://repository.horizon.ac.id/items/show/8157.