RE: https://hachyderm.io/@Mara/115544451317527305

I love doing this kind of work for some reason, it's like nesting but for a software project... for #MCMCStan I reimplemented a bunch of arcane gradient functions ages ago and I'm still tickled when people migrate some stale implementation to use those... my other pet project was attempting to make it easier to modify the output streaming from Stan and that one sadly never worked out 😢 It's hopeful work b/c it's not worth doing unless you think the project will stick around...

I recently helped out on an analysis by reworking a #MCMCStan model for a small analysis (longitudinal data on just like thousands of individuals) and it highlighted that it's a fun space to develop models in.

You have enough to make it interesting but not so much that you're playing data engineer half the time.

I am really disappointed that this is happening in the service of #AI when it really should be done to feed the world's insatiable need for Bayesian inference. #mcmcstan #stanlang #bayesian

https://wapo.st/4gIutOw

Sampling from a t distribution is harder than I thought. I wrote some code that implements a relatively new method by Shaw, Luu, and Brickman:
https://github.com/szego/truncated-student-t-rng

#stats #mcmcstan #mcmc

Anyone have a good way of taking cmdstanr optimization results, reshaping them into the list form required by `init`, and then starting sampling/optimization from that point?

#mcmcstan #bayes @mcmc_stan

🚨 New #JuliaLang package! StanLogDensityProblems.jl is a really basic package that implements the LogDensityProblems.jl interface for @mcmc_stan models, built on BridgeStan.jl. It also integrates with PosteriorDB.jl, which makes it really easy to benchmark a new inference method against a large number of models. #ProbProg #MCMCStan

https://github.com/sethaxen/StanLogDensityProblems.jl

There's a podcast interviewing the O.G. Bob Carpenter, give it a listen if you're a Bayes type. Real excited to listen when I get a moment!

https://learnbayesstats.com/episode/76-past-present-future-of-stan-bob-carpenter/

#mcmcstan

What's the best way to get the Hessian from #Stan models after optimizing?

#mcmcstan