diagnostic_posterior bug

[DominiqueMakowski]

I've encountered a few times a bug (that led to upstream bugs in parameters())

But I can't reproduce it... I don't know if it's a matter of like model convergence of what, but opening that issue in case someone confirms the bug or manages to reproduce it.

data <- lme4::sleepstudy
model <- brms::brm(Reaction ~ Days + (1 | Subject), data = data, refresh = 0, algorithm = "meanfield", seed = 333)

bayestestR::diagnostic_posterior(model)
#>     Parameter     Khat
#> 1      b_Days 4.609439
#> 2 b_Intercept 5.122897

^{Created on 2022-03-26 by the reprex package (v2.0.1)}

[strengejacke]

What's the issue here?

[DominiqueMakowski]

I got an error while running the example above, but I can't reproduce it 😕

[strengejacke]

Seems to be related to the sampling algorithm? If you remove "meanfield", more diagnostics are done.

[DominiqueMakowski]

no but the error was a complete error, it failed entirely... I'll post a reprex if I manage to reproduce it someday

[strengejacke]

closing for now, we can open a new issue if we encounter this problem again.