macOS update:¬† universal binary now available

Bluebells near Brandon, Warks, 30 April

This post is from Matt Denwood:

A universal binary of JAGS 4.3.1 for macOS is now available from Sourceforge.  Although there are no changes in JAGS itself from version 4.3.0, there are some changes to the way in which JAGS is compiled and the package installer is built for macOS.  We therefore recommend that all macOS users install this version.`
 
The major change is that the binary now runs natively on both intel (i.e. x86_64) and M1 (i.e. arm64) macs.  This means that you don’t need to worry about which system you have when installing JAGS:  the same binary will work with any computer running macOS 10.13 (High Sierra) or later, and is fully compatible with both versions of the official R build for macOS (currently R-4.2.0.pkg and R-4.2.0-arm64.pkg).  The installer is also now signed and notarized using Apple’s Developer Program.
 
If you already have JAGS installed either at /usr/local or /opt/R/arm64 (as will be the case for some M1 users) then the installer will remove these before creating a single installation at /usr/local – this means that you may need to reinstall rjags and runjags after running the new JAGS installer. To do this, just by run install.packages(c(“rjags”,”runjags”)) from R as usual.

An alternative installation mechanism is also now provided that may be useful for automated installation on dedicated test/build machines. A new subdirectory is now provided at https://sourceforge.net/projects/mcmc-jags/files/JAGS/4.x/Mac%20OS%20X/extras/ that contains pre-compiled arm64/x86_64/universal binaries in .tar.gz format. The universal version is identical to the official binary (except for a small change in the startup message for JAGS when run from the command line), but can be more easily downloaded and installed using an automated script. The compilation script needed to build JAGS for full binary compatibility with R is also provided under this directory for reference: this will be updated whenever needed either due a change in JAGS or in the build system for R for macOS.

There is also a new dedicated thread on the JAGS help forum for installation on macOS, so please post any issues you may have here:  https://sourceforge.net/p/mcmc-jags/discussion/610037/thread/115edca29d/

Windows update: JAGS 4.3.1 is released

JAGS version 4.3.1 is now available from Sourceforge. This is a patch update with a singular purpose: to allow a new Windows binary to be built using the Rtools42 toolchain.

JAGS is built with the same compiler as R on Windows in order to maintain binary compatibility. Hence we need to recompile JAGS for Windows whenever the compiler used for R changes. The forthcoming release of R 4.2.0 on 22 April features a major change in the compiler used for the Windows version. This change has been well publicized and documented in a series of blog posts by Tomas Kalibera. The fundamental reason for the change is to move to the Microsoft Universal C Runtime (UCRT) which allows native support for UTF-8.

One of the biggest user-visible changes is that the compiler no longer supports 32-bit builds. R 4.2.0 for Windows is 64-bit only and the same is true of JAGS 4.3.1 on Windows. Tomas made the necessary changes to the JAGS installer to make it 64-bit only – as well as fixing a few installer bugs – and these changes are incorporated into JAGS 4.3.1. Many thanks to Tomas for this.

If you are not on Windows then there is absolutely no reason for you to upgrade. The JAGS library and modules are unchanged from the last release. The only changes are in the Windows installer and in the installation documentation.

If you are on Windows then you should delay upgrading JAGS until the release of R 4.2.0 because JAGS 4.3.1 is incompatible with the current release of R.

Apologies to everyone who was hoping for more substantial changes in the next release of JAGS. I promise that JAGS 5.0.0 is on its way.

Off to Warwick

university-of-warwick-campus

I apologise for the low frequency of updates to this blog. The big news is that I am moving back to the UK next month after 23 years in France working for the World Health Organization. I am very excited to be starting a new job in the Department of Statistics at the University of Warwick.

New manual

I have spent the last couple of months updating the JAGS manual for the release of 4.3.0. I have uploaded the manual to Sourceforge so you can now download it. This is mainly for the benefit of people coming to my JAGS tutorial next week at useR! 2017.

This was the final task before releasing JAGS 4.3.0 so you can expect some news on that front fairly soon.

 

Course: 1st Italian Astrostatistics School, 12-16 June 2017

sfondoAstroStatisticSchools

Stefano Andreon writes:

We are pleased to announce the 1st Italian Astrostatistics School that will be held at INAF-Osservatorio Astronomico di Brera, Milano from 12 to 16 June 2017.

The primary goal of the school is to train astronomers to the use of modern statistical techniques, specifically parameter estimation and model selection. The INTENSIVE course, lectured by Stefano Andreon and Roberto Trotta, is characterized by extended laboratory sessions (i.e. individual work at the computer), taking about two thirds of the school attendance. The capacity is limited to 35 participants (first come first served
basis). The school is open to PhD students in astronomy, who have priority  (they should register by May 15th), and if space allows, to post-docs and researchers. From May 15th, registration is open to all (subject to availability). Attendance to the course has pre-requirements. For school program, details, and registration see:

http://iaa.mi.oa-brera.inaf.it/IAA/AstroStatisticsSchool.html

Bayescomp 2018

intro-bg

Photo by Richard Aparicio

Bayescomp 2018 will be held in Barcelona, Spain, on 26-29 March. This is the successor to the MCMski conference series.

Bayes Comp 2018 is a biannual conference sponsored by the ISBA section of the same name. The conference and the section both aim to promote original research into computational methods for inference and decision making and to encourage the use of frontier computational tools among practitioners, the development of adapted software, languages, platforms, and dedicated machines, and to translate and disseminate methods developed in other disciplines among statisticians.

See the conference web site for more details. Note that I am not involved in the organisation of the conference, I am just disseminating the information.