It happened to me that some chains do nothing (dont consume CPUs) before a quick sampling.
I have read this topic: Windows only -- stan() is taking a really long time to return · Issue #121 · stan-dev/rstan · GitHub
and thinking that maybe it is a problem of rstan.
Any suggestion about trying CmdStan? thank you.
Worth a shot. If you’re locked to Windows, you might also try installing Ubuntu & Rstudio via WSL2 and running things that way.
you could try either CmdStanR or CmdStanPy if you’re having trouble with rstan - R Interface to CmdStan • cmdstanr and cmdstanpy – Python interface to CmdStan — CmdStanPy 0.9.76 documentation
these are (usually) easy to install and provide access to the latest version of Stan.
the only thing they can’t do that rstan does is provide access to the log gradients (see https://mc-stan.org/rstan/reference/stanfit-method-logprob.html)