Confusion about INLA on HPC cluster

[Xicheng Xie]

Hello everyone,

I was running inla on the HPC cluster, but it kept giving me the same error over and over again. I ran the simplest model, which is inla(formula, family = "gaussian", data = df_res[[1]]), but it returns

Error in if (...length() && any(...names() == "Dimnames")) .Object@Dimnames <- fixupDN(.Object@Dimnames) : missing value where TRUE/FALSE needed

 *** inla.core.safe:  inla.program has crashed: rerun to get better initial values. try=1/2 

Error in if (...length() && any(...names() == "Dimnames")) .Object@Dimnames <- fixupDN(.Object@Dimnames) : 

  missing value where TRUE/FALSE needed

Does anyone encounter the same problem as me when running inla on the HPC cluster? Please let me know how to fix it!!!

Thanks so much,

Xicheng

[Steffi]

Strangely, I just discovered the same problem. Even for the simplest model in INLA I get the following error messages. No way to run whatsoever.

I re-installed r-base-core and INLA, but nothing changed. Any hints are welcome!

Steffi

Error in if (...length() && any(...names() == "Dimnames")) .Object@Dimnames <- fixupDN(.Object@Dimnames) : missing value where TRUE/FALSE needed *** inla.core.safe: inla.program has crashed: rerun to get better initial values. try=1/2 Error in if (...length() && any(...names() == "Dimnames")) .Object@Dimnames <- fixupDN(.Object@Dimnames) : missing value where TRUE/FALSE needed *** inla.core.safe: inla.program has crashed: rerun to get better initial values. try=2/2 Error in if (...length() && any(...names() == "Dimnames")) .Object@Dimnames <- fixupDN(.Object@Dimnames) : missing value where TRUE/FALSE needed Error in inla.core.safe(formula = formula, family = family, contrasts = contrasts, : *** Fail to get good enough initial values. Maybe it is due to something else.

[Finn Lindgren]

The first report about this error message on the list I could find was
https://groups.google.com/g/r-inla-discussion-group/c/O-Vn_LFJkqw/m/nTJ86HErBAAJ
and wasn't traced to a specific function or package, but the root
cause was clear; a version of some package, likely Matrix, since
fixupDN() is an unexported function in that package, was installed on
a pre-4.2 version of R, but the ...length() and ...names() features
weren't introduced to R in previous versions, such as 4.1, and the
previous reports of this issue were resolved by upgrading R to a 4.2.x
version.

If you rerun with inla(..., safe = FALSE) you should be able to
produce traceback() output. From that, we may be able to pinpoint the
source of the problem.

Finn

> --
> 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/a59a1d2a-54ab-4cf7-ac35-635b938f7f44n%40googlegroups.com.



--
Finn Lindgren
email: finn.l...@gmail.com

[Finn Lindgren]

Also, please provide your complete sessionInfo() so we can see what
package versions are used.
Finn

[Havard Rue]

Normally on HPC you also need another than the default, like the centos7 one which you can install with

inla.binary.install()


Also using R 4.3 would help as we do not build with 4.2 anymore 

-- 

Håvard Rue 

hr...@r-inla.org

To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/CAHC4gzZujyqMtJFjEG9i1FNTCx3KLJ5x3LFKM2Xc1t0hER-jcQ%40mail.gmail.com.

[Steffi]

Thank you all for the extremely fast and useful help. Upgrading to R 4.3 and re-installing everything has resolved the issue.

All the best and thank for your fantastic work as always,

Steffi