Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
After an extended beta period, JAGS 5.0.0 is now available. The release is more of a process than a single event, but we have started by uploading the source tarball and the Windows installer to SourceForge. The macOS installer will be along hopefully next week.
Thanks to Jeroen Ooms, Windows binary installers are available from the r-windows project on GitHub. This includes binaries for aarch64/Arm64.
Changes since the beta release
- Better multi-core support on macOS (More on this in a future post).
- Fixed a delayed rejection bug in the
MNormalsampler. - Added a new elliptical slice sampler contributed by Marcel Jonker.
- The
dic.samplesfunction could throw a logic error (See ticket #66). I am unable to reproduce this but JAGS 5.0.0 includes a workaround that should avoid triggering this bug. - Fixed a reproducibility bug in
BaseRNGFactory.
I would not normally add a new sampler at this late stage, but the elliptical slice sampler was ready to go and performs surprisingly well for a non-adaptive sampler. If you experience any trouble with this new sampler then you can turn it of with
set.factory("bugs::EllipticalSlice", type="sampler", state=FALSE)
JAGS will then fall back on the adaptive random walk Metropolis-Hastings sampler MNormal.
The rjags package and CRAN migration
The current rjags package on CRAN (version 4-17) will not work with JAGS 5.0.0. You need version 5-x but for the time being this will only be available from SourceForge.
- Source tarball
- Windows binary (x86_64)
- macOS binary
The reason for not uploading the rjags package to CRAN is that it currently breaks a lot of reverse dependencies. I have been quite proactive in contacting the maintainers of these package and providing patches, but we still need to give them more time.
We are monitoring the reverse dependency situation and will be liaising with the CRAN maintainers.
Preview of upcoming posts
In order to help users and developers with the transition to JAGS version 5, a series of “What’s new in JAGS 5.0.0” blog posts are planned covering the following topics.
- Initial values
- Reproducibility
- Multicore support
- ABI compatibility
- New diagnostics
These will include guidance for package maintainers to make the transition to JAGS 5.0.0.
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.