I've been trying to sample from multiple Stan models in parallel, and chains terminate unexpectedly every single time.

I tried all kinds of workarounds without any success, and eventually found the following bug report:

https://github.com/stan-dev/cmdstan/issues/1169

Containers seemed like overkill; then I discovered `unname` and found that it worked! What a relief. I still have no idea *why* this is necessary ...

#StanMCMC #cmdstan #cmdstanr

Jaw. Dropped.

@cjvanlissa has a tutorial in #ESMARConf2023 blowing my mind in terms of production level. There are good slides, *two* camera views of the speaker, and even attention-preserving cuts.

The subject was the #Rstats package {pema}, for penalized meta-regression in #MetaAnalysis with #StanMCMC

https://youtu.be/Bd_hCE0yRSQ