gdistsamp() makes R exit

[Marc Kery]

Dear all,

 

we’re using gdistsamp() to fit HDS models with imperfect availability to some bird count data from Bialoviecza forest in Poland. Admittedly, our current model is mis-specified: we have 5 primary occasion separated by 2 weeks and within each, two secondary occasions of 5 mins back to back. Thus, the proper model for availability would respect the two temporal scales of the sampling, and specify a different availability parameter among primary than among secondary occasions. We’re working on this in JAGS, but for now wanted to explore the data in unmarked with the mis-specified with 5 * 2 = 10 primary occasions. Now, on 3 different machines, and with Rstudio as well as with proper R, we have observed very frequently that R just exits during the call to gdistsamp(). This happens quite unpredictably, i.e., not always with the same model, sometimes after the first model we fit after launching R and sometimes only after the fourth model … but for about 50% or more of our attempts to fit a model. Has anybody else observed this ? Any ideas what we could improve (other than the model) ?

 

Thanks and best regards -- Marc

 

 

[Ken Kellner]

Hi Marc,

There was a similar bug report a couple years ago. Because the crashes happened randomly and didn't give any errors (meaning it must be in the C++ code) I struggled to identify the bug and never tracked it down. I think missing values played some role here.

Can you try setting engine = "R"? My guess is that should run consistently, though slower.

If you can package up a minimal example that you know crashes sometimes from your data and send it to me that would be helpful, but only if you have time.

Ken

On Wed, Oct 18, 2023 at 09:48:21AM +0000, Marc Kery wrote:
> Dear all,
>

> we're using gdistsamp() to fit HDS models with imperfect availability to some bird count data from Bialoviecza forest in Poland. Admittedly, our current model is mis-specified: we have 5 primary occasion separated by 2 weeks and within each, two secondary occasions of 5 mins back to back. Thus, the proper model for availability would respect the two temporal scales of the sampling, and specify a different availability parameter among primary than among secondary occasions. We're working on this in JAGS, but for now wanted to explore the data in unmarked with the mis-specified with 5 * 2 = 10 primary occasions. Now, on 3 different machines, and with Rstudio as well as with proper R, we have observed very frequently that R just exits during the call to gdistsamp(). This happens quite unpredictably, i.e., not always with the same model, sometimes after the first model we fit after launching R and sometimes only after the fourth model ... but for about 50% or more of our attempts to fit a model. Has anybody else observed this ? Any ideas what we could improve (other than the model) ?

>
> Thanks and best regards -- Marc
>
>

> --
> *** Three hierarchical modeling email lists ***
> (1) unmarked (this list): for questions specific to the R package unmarked
> (2) SCR: for design and Bayesian or non-bayesian analysis of spatial capture-recapture
> (3) HMecology: for everything else, especially material covered in the books by Royle & Dorazio (2008), Kéry & Schaub (2012), Kéry & Royle (2016, 2021) and Schaub & Kéry (2022)
> ---
> You received this message because you are subscribed to the Google Groups "unmarked" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+u...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/unmarked/ZR0P278MB08698FC636BA67E8106690E4EBD5A%40ZR0P278MB0869.CHEP278.PROD.OUTLOOK.COM.

[Marc Kery]

Dear Ken,

Apologies for not doing our homework before posting... and thank you very much for the reminder. My colleague has now set engine = "R" as you suggest and indeed, that solves the problem. We will send you an example with data and code offline.

Best regards -- Marc


-----Ursprüngliche Nachricht-----
Von: unma...@googlegroups.com <unma...@googlegroups.com> Im Auftrag von Ken Kellner
Gesendet: Mittwoch, 18. Oktober 2023 14:05
An: unma...@googlegroups.com
Betreff: Re: [unmarked] gdistsamp() makes R exit

To view this discussion on the web visit https://groups.google.com/d/msgid/unmarked/ZS/KA8tqwUBWFA4Y%40COYOTE.