Skip to content

error in glmmTMB when adding glmmTMBControl: unused arguments

4 messages · Ben Bolker, Mollie Brooks, Diego Pavon

#
Dear list

I am trying to run this model

M1.zi <- glmmTMB(Abundance ~ Temp_std + NEness_std + WithinNatura +
Winter_std
                 + WithinNatura * Winter_std + NEness_std * Winter_std +
Temp_std * WithinNatura
                 + WithinNatura * Winter_std  * NEness_std
                 + (1|fWinter/site) + (1|species),
                 family = poisson,
                 ziformula = ~ 1,
                 data = AllSpecies)

but after a couple of days (I have 3.5 million data points), I finally got
an error:

Error in optimHess(par.fixed, obj$fn, obj$gr) :
  gradient in optim evaluated to length 1 not 14
In addition: There were 11 warnings (use warnings() to see them)


Then I tried to add the CONTROL argument following
https://cran.r-project.org/web/packages/glmmTMB/glmmTMB.pdf:

M1.zi <- glmmTMB(Abundance ~ Temp_std + NEness_std + WithinNatura +
Winter_std
                 + WithinNatura * Winter_std + NEness_std * Winter_std +
Temp_std * WithinNatura
                 + WithinNatura * Winter_std  * NEness_std
                 + (1|fWinter/site) + (1|species),
                 family = poisson,
                 control = glmmTMBControl(optCtrl = list(iter.max = 1000,
eval.max = 1000), profile = TRUE, collect = FALSE),
                 ziformula = ~ 1,
                 data = AllSpecies)


For some reason, I get this error message:

Error in glmmTMB(Abundance ~ Temp_std + NEness_std + WithinNatura +
Winter_std +  :
  unused argument (control = glmmTMBControl(optCtrl = list(iter.max = 1000,
eval.max = 1000), profile = TRUE, collect = FALSE))


Does anyone have an idea how to specify the control arguments? I suspect
this is some small thing I am missing but I can't see it anymore...

Thank you very much for your time and help.

Best

Diego
#
A reproducible example would be useful.  This works (with the
development version
of glmmTMB, but I'd be surprised if it broke since the last CRAN release ...)

library(glmmTMB)

m1 <- glmmTMB(count~ mined + (1|site),
              zi=~mined,
              family=poisson, data=Salamanders,
  control=glmmTMBControl(optCtrl=list(iter.max=1e3,
                                      eval.max=1e3),
                         profile=TRUE))

what's your sessionInfo()?

Maybe obvious, but it would probably be efficient to do some
troubleshooting/model-checking on a
smaller subset of your data, if you haven't already ...

On Mon, Feb 12, 2018 at 3:04 AM, Diego Pavon
<diego.pavonjordan at gmail.com> wrote:
#
Yes, I agree that the control argument (as you specified it) appears to be working on the development version of glmmTMB.
For example
(m3 <- glmmTMB(count~spp + mined + (1|site), 
  zi=~spp + mined, 
  family=nbinom2, Salamanders, control = glmmTMBControl(optCtrl = list(iter.max = 1000,
eval.max = 1000), profile = TRUE, collect = FALSE)))

The error you get
should have produced a message including "See vignette('troubleshooting')". If it didn?t, then something is broken, so please report it in the issue tracker on GitHub or let me know.

thanks,
Mollie

???????????
Mollie E. Brooks, Ph.D.
Research Scientist
National Institute of Aquatic Resources
Technical University of Denmark

  
  
#
Dear Mollie and Ben

After getting admin rights to my laptop, I managed to update glmmTMB from
0.1.0 to the latest 0.2.0 and the error message disappeared.

@Mollie, I will report to you if after running the model, the error message
I showed above still shows.

Cheers

Diego
On 12 February 2018 at 15:07, Mollie Brooks <mollieebrooks at gmail.com> wrote: