INLA gives error

[Amin Shn]

Hi, 

I tried running a Poisson model with 140 000 rows and three random effects, on of which is 'Besag' model on 36000 counties (so a gigantic adjacency matrix) and INLA gave me this error: 

taucs_malloc_stub fail to malloc 9085712 bytes 

Aborted (core dumped) 

Error in inla.inlaprogram.has.crashed() : The inla-program exited with an error. Unless you interupted it yourself, please rerun with verbose=TRUE and check the output carefully. If this does not help, please contact the developers at <he...@r-inla.org>.

Do I have a memory problem? Is the problem related to the size of my data? Is there any solution to it? 

Best,

Amin

[Elias T. Krainski]

Yes, it is a lack of memory. For this model size you might need a few dozens of Gb.

From the error message I can see that you are not using the (new) PARDISO library. You can try it following instructions from

  inla.pardiso()

Elias

--
You received this message because you are subscribed to the Google Groups "R-inla discussion group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to r-inla-discussion...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/d66a447f-7983-4d26-9f31-d3b1ea928d9dn%40googlegroups.com.

[Helpdesk]

...and use the newest testing version with R-4.1, adding

inla(..., inla.mode="experimental")

On Thu, 2021-11-04 at 18:52 +0300, Elias T. Krainski wrote:
> Yes, it is a lack of memory. For this model size you might need a few
> dozens of Gb.
> From the error message I can see that you are not using the (new)
> PARDISO library. You can try it following instructions from
>   inla.pardiso()
>
> Elias
>
>
> On Thu, 4 Nov 2021 at 18:44, Amin Shn <amins...@gmail.com> wrote:
> > Hi, 
> >
> > I tried running a Poisson model with 140 000 rows and three random
> > effects, on of which is 'Besag' model on 36000 counties (so a
> > gigantic adjacency matrix) and INLA gave me this error: 
> >
> > taucs_malloc_stub fail to malloc 9085712 bytes 
> > Aborted (core dumped) 
> > Error in inla.inlaprogram.has.crashed() : The inla-program exited
> > with an error. Unless you interupted it yourself, please rerun with
> > verbose=TRUE and check the output carefully. If this does not help,
> > please contact the developers at <he...@r-inla.org>.
> >
> > Do I have a memory problem? Is the problem related to the size of my
> > data? Is there any solution to it? 
> >
> > Best,
> > Amin

--

Håvard Rue
he...@r-inla.org

[Amin Shn]

Thank you, I tried the inla(..., inla.mode="experimental") but still have the same problem. I couldn't get a license to use PARDISO library. When I enter my information including my university in Ireland it says  “Your corporate or commercial entity is not yet included in the current list of corporate users” so I don't know how I can use the PARDISO to see if that could fix the problem. At the moment the only solution I can think of is to use a lower resolution adjacency matrix and fit the model on a smaller scale. 

Best,

Amin

--
You received this message because you are subscribed to a topic in the Google Groups "R-inla discussion group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/r-inla-discussion-group/Z9cQhNsBssc/unsubscribe.
To unsubscribe from this group and all its topics, send an email to r-inla-discussion...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/ad3db83c2853ac04eb18c0429cc42d4a73349893.camel%40r-inla.org.