[R] Current state of support for BUGS access for Linux users?

Paul Johnson pauljohn32 at gmail.com
Mon Jan 16 23:13:44 CET 2006


I'm going to encourage some students to try Bayesian ideas for
hierarchical models.
I want to run the WinBUGS and R examples in Tony Lancaster's An
Introduction to Modern Bayesian Econometrics.  That features MS
Windows and "bugs" from R2WinBUGS.

Today, I want to ask how people are doing this in Linux? I have found
a plethora of possibilities, some of which are not quite ready, some
of which work only under MS Windows.  Right now I just want to know
"what actually works".

Here's where I stand now in Fedora Core 4 Linux.
1. OpenBUGS-2.1.1 runs in Linux.  I can run "linbugs" (the console
version similar to the old BUGS) and also I can run--under wine--the
newest version of "winbugs.exe" that is circulated with OpenBUGS.  As
far as I can tell, the graphical interface in wine/winbugs works in
almost all elements.  A few things seem not quite right in the GUI
(can't initialize more than one chain, difficult to specify variables
for monitoring), but it does work.

It is easier to install and work with OpenBUGS's version of
winbugs.exe than with Winbugs-1.4 because the Open version does not
have that annoying license registration and "winbugs.exe" is not
wrapped inside an installation script.   I'm a little confused about
WinBUGS versions because the BRugs documents
refer to WinBUGS-1.5, which refers to
http://www.biostat.umn.edu/~brad/software/BRugs/WinBUGS15.zip, which
can be downloaded without any of the registration steps, but WinBUGS15
is not mentioned in the WinBUGS site (where 1.4.1 appears to be the

Supposing I get the winbugs.exe question settled:

2. How to most dependably send jobs from R to "linbugs" or "winbugs.exe"?

The BRugs package is preferred?

For a long time, R2WinBUGS was Windows-only, but toward the end of
last fall I noticed that R2WinBUGS now does compile and install under
R in Linux.

however, its help still says:
SystemRequirements:   WinBUGS 1.4 on Windows

I'd appreciate any advice.

Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas

More information about the R-help mailing list