How to best handle a large forcing vector for an ODE system

Here’s the thread @wds15 is talking about. It has code and examples from the two of us:

I wrote a poster on the subject and an implementation for Torsten (a extension of Stan for pharmacometrics), which I have here. I also wrote a manuscript, but the journal I submitted it to was not interested.

@andrewgelman spoke about writing a case study paper or extending documentation on ODE-based models, following this episode. It could be an extensive update on the (relatively simple) StanCon article I wrote on ODE-based models in 2017, and an opportunity to pool everyone’s expertise.