#13 Building a Probabilistic Programming Framework in Julia, with Chad Scherrer

Apr 8, 2020

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Julia's Modularity

Julia's modularity allows mixing and matching packages without performance penalties.
This is unlike Python, where large packages are preferred to avoid Python's overhead.

INSIGHT

Julia's Performance

Julia's JIT compilation and multiple dispatch enable high performance.
The compiler specializes code for specific argument types, avoiding runtime overhead.

ANECDOTE

Chad Scherrer's Bayesian Journey

Chad Scherrer's introduction to Bayesian methods involved a Haskell-based probabilistic programming language called Passage.
Passage generated C code with OpenMP for Gibbs sampling.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

How is Julia doing? I’m talking about the programming language, of course! What does the probabilistic programming landscape in Julia look like? What are Julia’s distinctive features, and when would it be interesting to use it?

To talk about that, I invited Chad Scherrer. Chad is a Senior Research Scientist at RelationalAI, a company that uses Artificial Intelligence technologies to solve business problems.

Coming from a mathematics background, Chad did his PhD at Indiana University of Bloomington and has been working in statistics and data science for a decade now. Through this experience, he’s been using and developing probabilistic programming languages – so he’s familiar with python, R, PyMC, Stan and all the blockbusters of the field.

But since 2018, he’s particularly interested in Julia and developed Soss, an open-source lightweight probabilistic programming package for Julia. In this episode, he’ll tell us why he decided to create this package, and which choices he made that made Soss what it is today. But we’ll also talk about other projects in Julia, like Turing or Gen for instance.

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

Links from the show:

Chad's Website: https://cscherrer.github.io/
Chad on Twitter: https://twitter.com/ChadScherrer
Soss Package: https://github.com/cscherrer/Soss.jl
Soss Presentation at 2019 Strata NYC: https://slides.com/cscherrer/2019-09-26-strata#/
Passage -- A Parallel Sampler Generator for Hierarchical Bayesian Modeling: https://bit.ly/2UTmaYB
Dynamic HMC in Julia: https://github.com/tpapp/DynamicHMC.jl
Advanced HMC in Julia: https://github.com/TuringLang/AdvancedHMC.jl
Monte Carlo Measurements in Julia: https://github.com/baggepinnen/MonteCarloMeasurements.jl
Turing.jl -- Bayesian inference with probabilistic programming: https://turing.ml/dev/
Gen.jl -- Probabilistic modeling and inference in Julia: https://www.gen.dev/
Etalumis -- Bringing Probabilistic Programming to Scientific Simulators at Scale: https://arxiv.org/abs/1907.03382
Omega.jl -- A programming language for causal and probabilistic reasoning: http://www.zenna.org/Omega.jl/latest/
JuliaLang -- The Ingredients for a Composable Programming Language: https://white.ucc.asn.au/2020/02/09/whycompositionaljulia.html
Simpy -- Discrete event simulation for Python: https://simpy.readthedocs.io/en/latest/