Cmdstan 2.24 release candidate now available

Yes, that needs to be highlighted.

CmdStan release notes so far:

CmdStan 2.24.0


  • CmdStan User’s Guide now online - updated and revised
  • Utility command stansummary allows user-specified quantile reporting,
    improved output format, better command-line options handling
  • Many makefile improvements:
    • precompiled headers
    • detect compiler option dependencies
    • improved messaging


  • Generated quantities for models with non-scalar parameters

comments and edits here:

My bad. I’ll add something to the Math notes.

@rok_cesnovar announcement text can be something like “The new ODE interface is intended to make it easier to specify the ODE RHS by avoiding packing and unpacking schemes required with the old interface. It is important to switch to the new interfaces because the old interfaces are significantly slower than they were previously (we have seen up to 30%)”

Updated top-level comment and the draft I have for the actual release. Thanks.

So much good stuff! Thanks to everyone who contributed! :)


Thanks everyone! For anyone who wants to try out the release candidate with CmdStanR you can install it easily using

rc_2_24 <- ""
cmdstanr::install_cmdstan(release_url = rc_2_24, cores = 4)

You may need to manually set the path the next time you use CmdStanR using

# replace with your path

because I think we have it default to the latest installed release (not release candidate).


I’m playing around with this in cmdstanr and I needed to add the -Wno-ignored-attributes flag or my Rstudio gets crazy slow and bogged down.

Can we make this flag default until that warning goes away? It’s an Eigen/compiler interaction right?

Something like:

model <- cmdstan_model("test.stan",
                       cpp_options = list(CPPFLAGS = "-Wno-ignored-attributes"))

Without this even something simple like:

parameters {
  real a;

model {
  a ~ normal(0, 1);

eats up my Rstudio session.


@bbbales2 have you tried this from the command line just using the CmdStan makefile?
I think the makefile is sometimes recompiling more things than it should, but I haven’t yet tried to systematically make this happen -

I ran across this just compiling a very few different models to get output to test stansummary with.

Very awesome! Should we also tweet this out via the @mcmc_stan account for more visiblity?

1 Like

Yes, we should tweet this… we need testing NOW.

Refreshed 200+ core HTCondor pool in need of some jobs for stress testing. Would happily contribute some resources to testing out compile and run times.

is your friend. Maybe just use the stat_comp_benchmark stuff.

Though… these models are really tested well already. The example model git would probably be better to go through for larger test coverage.

I see the 2.24-rc tag in the Cmdstan repository but not in stan-dev/stan.

Does this mean there’s no rc for stan or math, just Cmdstan?

Indeed, we saw no use-case of making a Stan RC. Same for Math. If you have a need for a Stan RC we could build it next time.

the stan rc is essentially defined by the stan git hash tag which got tagged by cmdstan. So it’s implicitly there.

Yes, pystan 3/httpstan would use a tagged release. In our build process
we download a release (just like one downloads a stanc binary).

To give you a sense of how we use the tagged releases, here are two
lines from our Makefile:

curl --silent --location$(STAN_VERSION).tar.gz -o $@

curl --silent --location$(MATH_VERSION).tar.gz -o $@

Ok, let me know if you would already benefit from having a Stan & Math RC now and would use it this week. If so, I can ask @serban-nicusor to build when he has the time. Otherwise I just made a note to build a RC for Stan & Math for the next release cycle.

1 Like

@rok_cesnovar @jonah how do you feel about -Wno-ignored-attributes in cmdstanr until the Eigen issue goes away?

I guess so far it’s only me suffering from that issue, but man, there is soooo much output that prints to the point that it bogs down my computer.


You are far from the only one suffering form that output. Its annoying to me as well.

I am going to open a PR in math to make that flag be the default flag. Its a problem for anyone with g++ 8 or recent clang.

Just need to double check that it doesnt cause a problem with the old compilers that might not have that flag (they probably do have it, just to double check). Will spin up an AWS instance to install the old compilers.


Thanks! Will the docs also update the current/old HMM model example to show how to use the new functions?