Stan General Thursday Meeting, June 20, 2019 11 am EDT

Hangouts Link:

Instructions: Ask to attend and someone should let you in in the first 10 minutes of the meeting. Email if you have problems or want to attend the physical meeting in New York City.

Meeting Agenda:

2 min round robin of what you have been doing, introduction if you are new and what your interests in Stan are.

General Meeting:

Item 1: What is the scope of case studies? Someone’s trying to submit a case study based on the doc for their R package extending Stan. Is that in scope?

Please add agenda items in responses.



Edward R: c++ implementations of new R-hat and ESS diagnostics.

Simon: Working on adaptive leapfrog, more work on inference.

Andrew: Making some progress getting hier models faster.

Ben G: Rstan release still held up, just one package holding us back.
Reviewing some books. Working on papers.

Charles: In Finland, focusing on nested Laplace approx, doing good speed increases. Working on algebraic solver. Working on tutorial for StanCon.

Mitzi: User interface, cmdStanPy work continues, standalone documentation. Met with Pyladies. Planning on some meetups.

Leah C: Just graduated with PhD. Linking Bayesian causal inference.

Sean: Compiler work, benchmarked tensor flow prob with Stan. Same speed on LR models. Measuring gradients/sec. Planning offsite developers meeting.

Sean reporting for dev ops person, Nick: Working on benchmark platform.

Sean reporting for consultant Andrew A: Working cmake and mono-repo.

Matthijs: Autodiff, semantics paper, bug fixes for new compiler.

Lauren: Finishing papers. Working out talks.

Jonah: Taught 3 day in Michigan, very good workshop. Editing papers. Rpackage work.

Breck: StanCon continues, $100k contribution from Facebook to NumFOCUS matching Schmidt funds. 20% to matching goal. Lots of SGB activity.

Bob: Worked with Shira on modeling. Finished autodiff tests refactor. Waiting on PR review. Working on speeding up compilation.

Report on Math meeting: went through PRs that are still open.

general discussion

Bob: Do we allow Stan derived packages in Case Studies?
Breck, Jonah and Sean will look into a general solution to allowing this.

Andrew: What do with a 0 successes on 50 trials?