Aki Vehtari

@davharris: That's a good question. Unfortunately, I don't know the answer and I have to think about this. The problem is that when you predict in blocks, you are making...

1) Can you report `devtools::session_info("loo")` If you are using Windows, check if this known but obscure problem in Windows multi core is affecting you https://discourse.mc-stan.org/t/problem-parallelizing-rstanarm-loo/6433/10 . 2) We could give...

Seems easy to test with pbapply. Should define min_time.

@jgabry I would like to change also loo object part `diagnostics$n_eff` to `diagnostics$ess`, but can we do that?

Just report 0, or also store in loo object `$diagnostics$ess` value 0?

Why NaN instead of NA? Currently if any k>0.7, we report MCSE as NA I don't have a strong preference for 0, but I would assume 0 would produce fewer...

> but I think you said that MCSE returned NaN? Where? In the above comment, I write "Currently if any k>0.7, we report MCSE as NA"

> I just think NA/NaN is better than 0 since it will cast a more explicit error and hence be more defensive. People would not accidentally use the value. NA...

> Sorry. I see ” We are already reporting for MCSE NaN if any khat>0.7, so I propose we report…” in the comment above? Oh, in the first post! I...

I'm fine with these. We discussed also rbind and cbind type functions, but I guess it's not indexing. Pinging @dpsimpson, too.