Error for grouped correlated random effects in multivariate Poisson model

Interfaces brms
1

Grouping correlated random effects using (1|cor|gr(random effect, by = group)) is currently throwing the following error.

Chain 1: Rejecting initial value:
Chain 1: Error evaluating the log probability at the initial value.
Chain 1: Exception: poisson_log_lpmf: Log rate parameter[1] is nan, but must not be nan! (in
‘modelf3c2f4b3b47_c5d728f3d7cb11eab4bdaf4b6d4561d8’ at line 81)

I have attempted to specify various starting values and priors to no avail. Here is a minimally reproducible example on my machine. I uninstalled and reinstalled both rstan and brms beforehand.

df ← data.frame(y1 = sample(seq(1,100)),
y2 = sample(seq(1,100)),
re = rep(seq(1:20), each=5),
group = rep(c(1,2), each=50))
f1 ← bf(y1 ~ 1 + (1|cor|gr(re, by = group)) ) + poisson()
f2 ← bf(y2 ~ 1 + (1|cor|gr(re, by = group)) ) + poisson()
m ← brm( f1 + f2 + set_rescor(FALSE), data=df, chains=1)

Please also provide the following information in addition to your question:

  • Operating System: Windows 10
  • brms Version: Current development version 2.12.3
2

Thanks. This was caused by a typo which should now be fixed in the brms github version.

1 Like
3

Awesome, thanks so much Paul.