Having Issues With Example3_Bays2009 brms Script

[mdrcboston]

Hello I'm having trouble with the brms scripts. I'm trying to run the brms script "Example3_Bays2009.R" to replicate the results, however, I get this error message for each of the 6 chains:

"SAMPLING FOR MODEL 'anon_model' NOW (CHAIN 1).
Chain 1:
Chain 1: Gradient evaluation took 0.080665 seconds
Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 806.65 seconds.
Chain 1: Adjust your expectations accordingly!
Chain 1:
Chain 1:
[1] "Error in sampler$call_sampler(args_list[[i]]) : "
[2] " c++ exception (unknown reason)"
error occurred during calling the sampler; sampling not done
[[1]]
Stan model 'anon_model' does not contain samples."

Do you know why this may be happening? I can successfully upload the Bays data file into R and everything works up until I try to fit the mixture model using brms. As a result, there are plot outputs and the summary does not print.

I'm having no trouble with the bmm version of Example3, just the brms script.

Hope you can help, thanks!

[GidonFrischkorn]

Hi @mdrcboston,
thanks for creating this issue, and I am sorry that you ran into issues with running the examples. Unfortunately, I could not reproduce your problem, I have tried fitting the model specified in "Example3_Bays2009.R" for brms both with using cmdstanr and rstan as backend. For me sampling started without any issues.

Can you provide some further details regarding your R and STAN setup? More specifically, which interface from R to STAN are you using (rstan or cmdstanr and which versions). As you have no issues with using with the bmm example it could be that when you are fitting the brms example rstan is selected as default backend, whereas bmm uses cmdstanr as backend as default if possible.

@venpopov do you have other ideas what might cause this issue?

[mdrcboston]

Hi @GidonFrischkorn,

Thank you for your reply! I was able to successfully run the brms model!

When I specified that brms use cmdstanr as the backend, the model finally worked and I was able to replicate the results.