Martin Modrák

Hi, `SBC` uses `future_lapply` from the `future.apply` package to run the fits. It appears (I didn't test it) that `future.apply` integrates well with `progressr` to show progress - see https://github.com/HenrikBengtsson/future.apply/issues/34#issuecomment-549011124...

Thanks for reporting. I just pushed a workaround to make this code path not give an error (hopefully - please test). The code will now report which underlying cmdstanr fits...

@maxbiostat No need for a full MWE. What would however help is when the package gives you a message like: ``` Fits for datasets 1,5,113 had NA for n_failed_chains. ```...

I just tried it myself and both versions finished in less than 20 seconds... Are you using multiprocessing? I've used the default ```r library(future) plan(multisession) ``` Maybe some of the...

> how can this be finished in twenty seconds..? I have a quite powerful 20 core PC at work 😁 > I am using SBC.min_chunk_size = 5; can increasing this...

> FYI, from here it was mentioned that future_apply will be deprecated. This concerns the `future_lapply` function in the `future` package. We are using `future.apply::future_lapply()` as recommended on that page.

> updated version doesn't support it What do you mean by "doesn't support" - do you get an error? Or does it take too long?

Partially addressed in acfd50fab2193ff96463d5af6bcae7c20d2951c1

> How big is stantarget package? No idea :-) Just know that it exists and has some users (judging from forum discussions) > I see great potential in using SBC...

I think the source of the confusion is that posterior ranks start with 0, because "rank is defined as the number of draws < simulated value" so the lowest rank...