R session aborted very often

[Vítor Margato]

I've been using RStan on a daily basis and I get the "bomb warning" (R session aborted) very often, almost daily.

(For reference, my computer is a MacBook Pro, 2.9 GHz Intel Core i5, 8 GB 1867 MHz DDR3, and I think it's unlikely that this is caused by something specific to it.)

Is this normal? Does it happen more often when one uses relatively complex models? Is there any simple way to make it happen less often?

Thanks very much!

Vitor

[Ben Goodrich]

On Friday, January 27, 2017 at 12:21:57 PM UTC-5, Vítor Margato wrote:

I've been using RStan on a daily basis and I get the "bomb warning" (R session aborted) very often, almost daily.

(For reference, my computer is a MacBook Pro, 2.9 GHz Intel Core i5, 8 GB 1867 MHz DDR3, and I think it's unlikely that this is causes by something specific to it.)

Is this normal? Does it happen more often when one uses relatively complex models? Is there any simple way to make it happen less often?

Can you say more about (or post) the code that causes RStudio to bomb?

Ben

[Vítor Margato]

Hi again, Ben!

Sadly, I think I can't. That's not to say it's an incredibly innovative model that I wanna keep to myself. It's just that I think it technically belongs to my employer, so I'm not sure I'm allowed to share it publicly.

I'm aware it's not very helpful (to say the least) to ask for tips without posting the model, so all I wanted was to know whether this sort of problem happened often to others, and particularly whether this was a typical situation that could be overcome by some minor, standard change.

Thanks again.

Vitor

[Bob Carpenter]

To answer your question, RStan should never cause R to
abort. If it does, that's a bug. It it happened to one
of the developers using RStan, we'd file an issue with
a bug report.

So given that this never happens to us, it's hard to say
more without a reproducible example.

- Bob

> On Jan 27, 2017, at 12:21 PM, Vítor Margato <vitorma...@gmail.com> wrote:
>
> I've been using RStan on a daily basis and I get the "bomb warning" (R session aborted) very often, almost daily.

> (For reference, my computer is a MacBook Pro, 2.9 GHz Intel Core i5, 8 GB 1867 MHz DDR3, and I think it's unlikely that this is causes by something specific to it.)

> Is this normal? Does it happen more often when one uses relatively complex models? Is there any simple way to make it happen less often?
> Thanks very much!
> Vitor
>

> --
> You received this message because you are subscribed to the Google Groups "Stan users mailing list" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to stan-users+...@googlegroups.com.
> To post to this group, send email to stan-...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Ben Goodrich]

On Friday, January 27, 2017 at 1:03:08 PM UTC-5, Vítor Margato wrote:

Sadly, I think I can't. That's not to say it's an incredibly innovative model that I wanna keep to myself. It's just that I think it technically belongs to my employer, so I'm not sure I'm allowed to share it publicly.

I'm aware it's not very helpful (to say the least) to ask for tips without posting the model, so all I wanted was to know whether this sort of problem happened often to others, and particularly whether this was a typical situation that could be overcome by some minor, standard change.

It has happened on Windows, in which case you can install rstan from GitHub. It probably has little to do with the Stan program but rather the way in which the Stan program is being called.

Ben

[Aki Vehtari]

Since the last release I've had lot of crashes with a simple linear model, but I haven't figured out reproducible scenario (and thus no issue), although it seems to happen mostly after modifying a script and then rerunning the script and when it runs rstan it crashes. I mention it here, just to let you know there's at least one other user with similar problem.

Running R on terminal, 

rstan 2.14.1

R version 3.3.2 (2016-10-31)

Linux 4.4.0-59-generic #80-Ubuntu SMP Fri Jan 6 17:47:47 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux

Aki

[Bob Carpenter]

Thanks for letting us know. These kinds of issues are really
frustrating if they get reported and we can't reproduce them.
I haven't had any trouble with 2.14 (other than Anaconda messing
with my path when I installed Jupyter, but that's not Stan's fault).

- Bob

[Stephen Martin]

I've had this issue in the past. I recompiled rstudio and rstan and it seems to happen less (at least, it hasn't happened recently).

It would typically happen when running summary() on the stanfit. Rstudio would bomb out and I'd have to run the model again. I can't reliably reproduce it though. I think it happened at least once a day when testing a larger model of mine (but it was nowhere near running out of memory, so I don't think that's it).

[andre.p...@googlemail.com]

To answer your question, RStan should never cause R to
abort.  If it does, that's a bug.  It it happened to one
of the developers using RStan, we'd file an issue with
a bug report.  

So given that this never happens to us, it's hard to say
more without a reproducible example.

I faced also to some crashes in RStudio using RStan in Ubuntu 16.04.

I discovered, it usually happens when I not start from a "fresh" environment,

that means I run already some models in that environment before, and

when I move windows, especially plots. I cannot reproduce the crashes,

but they happen at then end of the simulation processes.

Now I "clean workspace" and "restart R", reload my necessary data start the

RStan run and it seem to be fine.

Andre

[Vítor Margato]

Yes that happens to me too!
Well, anyway, it's nice to know it happens to others.

Thanks for your feedback.