JAGS 2.2.0 is released

JAGS 2.2.0 is now available .  This is a patch release, so should present no major changes to the user interface.

You will find the source as well as binary packages for Windows and Mac OS X from Sourceforge .  For 64-bit Mac OS X a new optimized binary package is available called  JAGSdist-2.2.0 fast.dmg which offers substantial speed improvements over the standard Mac OS X binary. For other platforms, please consult the JAGS home page to see if there is a binary distribution available.

Thanks to the team who produce the binary packages: Bill Northcott (Mac OS X), Lars Vilhuber (Fedora, RHEL, SLES, OpenSUSE), Dirk Eddelbuettel (Debian, Ubuntu) and Michael Rutter (Ubuntu).

Advertisements

8 thoughts on “JAGS 2.2.0 is released

  1. Hi all,

    Today I compared the performance of HBC (Hierarchical Bayesian Compiler, v0.7) vs. JAGS (v2.1.0) using the Mixture-of-Gaussians example from HBC as the benchmarking problem. I used 100,000 iterations for both runs. It took 68sec for HBC and 32sec for JAGS. In other words, JAGS was twice as fast. I don’t care about absolute numbers, but about relative performance. Since HBC is compiled code while JAGS is interpreted, I expected to see significantly better results on HBC. Clearly, this didn’t happen.

    While I can think of several possible explanations, I wanted to put this out to everyone else who might have done similar benchmarking work and get their feedback.

    Regards,
    Ali

  2. That is certainly not what I would expect! I have no experience with HBC. Maybe there is some hand-tuning step that you have not done.

    Remember as well that from a statistical point of view what counts is the run time divided by the effective sample size, not the number of iterations.

  3. Did something happen to the dgen.gamma distribution. I tried to migrate from Winbugs/OpenBugs to JAGS. But it seems the manual includes the dgen.gamma distribution while it is not really implemented. JAGS complains about an unknown distribution. I tried all spellings (gen.gamma, dgen.gamma, dggamma etc.) to no avail. Is this a bug?

  4. It’s a bug. The generalized gamma distribution was included in the bugs module but never loaded into the distribution table. I have fixed this in the source.

  5. Thats an awfully long time. Are there any beta builds? Or what is the easiest way to compile it (what tools do i need?, what compiler?, is their a step-by-step guide)?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s