#30 Symbolic Computation & Dynamic Linear Models, with Brandon Willard

Learning Bayesian Statistics

It’s funny how powerful symbols are, right? The Eiffel Tower makes you think of Paris, the Statue of Liberty is New-York, and the Trevi Fountain… is Rome of course! Just with one symbol, you can invoke multiple concepts and ideas.

You probably know that symbols are omnipresent in mathematics — but did you know that they are also very important in statistics, especially probabilistic programming?

Rest assured, I didn’t really know either… until I talked with Brandon Willard! Brandon is indeed a big proponent of relational programming and symbolic computation, and he often promotes their use in research and industry. Actually, a few weeks after our recording, Brandon started spearheading the revival of Theano through the JAX backend that we’re currently working on for the future version of PyMC3!

As you guessed it, Brandon is a core developer of PyMC, and also a contributor to Airflow and IPython, just to name a few. His interests revolve around the means and methods of mathematical modeling and its automation. In a nutshell, he’s a Bayesian statistician: he likes to use the language and logic of probability to quantify uncertainty and frame problems.

After a Bachelor’s in physics and mathematics, Brandon got a Master’s degree in statistics from the University of Chicago. He’s worked in different areas in his career – from finance, transportation and energy to start-ups, gov-tech and academia. Brandon particularly loves projects where popular statistical libraries are inadequate, where sophisticated models must be combined in non-trivial ways, or when you have to deal with high-dimensional and discrete processes.

Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work at https://bababrinkman.com/ !

Thank you to my Patrons for making this episode possible!

Yusuke Saito, Avi Bryant, Ero Carrera, Brian Huey, Giuliano Cruz, Tim Gasser, James Wade, Tradd Salvo, Adam Bartonicek, William Benton, Alan O'Donnell, Mark Ormsby, Demetri Pananos, James Ahloy, Jon Berezowski, Robin Taylor, Thomas Wiecki, Chad Scherrer, Vincent Arel-Bundock, Nathaniel Neitzke, Zwelithini Tunyiswa, Elea McDonnell Feit, Bertrand Wilden, James Thompson, Stephen Oates, Gian Luca Di Tanna, Jack Wells, Matthew Maldonado, Ian Costley, Ally Salim, Larry Gill, Joshua Duncan, Ian Moran, Paul Oreto, Colin Caprani, George Ho and Colin Carroll.

Visit https://www.patreon.com/learnbayesstats to unlock exclusive Bayesian swag ;)

Links from the show:

• Brandon's website: https://brandonwillard.github.io/
• Brandon on GitHub: https://github.com/brandonwillard
• The Future of PyMC3, or "Theano is Dead, Long Live Theano": https://pymc-devs.medium.com/the-future-of-pymc3-or-theano-is-dead-long-live-theano-d8005f8a0e9b
• New Theano-PyMC library: https://github.com/pymc-devs/Theano-PyMC
• Symbolic PyMC: https://pymc-devs.github.io/symbolic-pymc/
• A Role for Symbolic Computation in the General Estimation of Statistical Models: https://brandonwillard.github.io/a-role-for-symbolic-computation-in-the-general-estimation-of-statistical-models.html
• Symbolic Math in PyMC3: https://brandonwillard.github.io/symbolic-math-in-pymc3.html
• Dynamic Linear Models in Theano: https://brandonwillard.github.io/dynamic-linear-models-in-theano.html
• Symbolic PyMC Radon Example in PyMC4: https://brandonwillard.github.io/symbolic-pymc-radon-example-in-pymc4.html
• What I Wish Someone Had Told Me About Tensor Computation Libraries: https://eigenfoo.xyz/tensor-computation-libraries/