Say Hello to PyMC 3.5

PyMC Developers
2 min readJul 21, 2018

--

Posted by Chris Fonnesbeck on behalf of the PyMC development team

The PyMC development team is proud to announce the release of version 3.5 today. This version features several usability enhancements, so we recommend this update to all users. The implementation of parallel sampling has been refactored to be more robust with large traces, show progress bars for multiple chains, and return partial results for interrupted sampling. A few new variable transformations have been added, most notably an Ordered transformation to ensure that mode switching does not occur during sampling, and a Chain transformation for chaining together multiple transforms on the same variable. Additionally, a sample_prior_predictive function has been added to allow for sampling from the prior predictive distribution.

Version 3.5 also heralds the return of directional acyclic graph (DAG) plots, which have not been seen since PyMC2. Calling model_to_graphviz on any model will generate a beautifully rendered plot of the DAG, which features the use of plates, to keep larger models looking compact.

The full list of changes are available in the release notes. You can grab a copy right now via pip:

pip install -U pymc3

Anaconda users may have to wait a few hours for the conda package to be made available on conda-forge.

As always, we’d like to remind users of our Discourse page where you can ask questions and receive advice on PyMC3 usage and best practices from the wider user and developer community. If you find any bugs in the new release, please report them to our issue tracker.

Thanks to the entire development team for all their hard work in making this release possible.

Happy sampling!

--

--

PyMC Developers
PyMC Developers

Written by PyMC Developers

Probabilistic Programming in Python

No responses yet