[ADMB Users] gradient_structure problem with nbinom glmm.admb in R

[Ben Bolker]

On 06/15/2011 09:42 AM, Rafael Mares wrote:
> Thank you Mollie, Dave and Hans for your replies and help.
> 
> As Mollie pointed out, the problem was simpler than I imagined. Just
> in case this is helpful for anyone else:
> 
> I was using the latest stable release of glmmADMB (0.5-2), which I
> mistakenly thought could handle 2 random effects (maybe it can, but
> not in the way I was specifying the model). I've now tried the alpha
> release (0.6) in linux (I think it doesn't work on windows and didn't
> work for me on a mac), and it seems to be doing what I want it to do:
> incorporating 2 random effects to my glmm, like so.
> 
> glmmadmb(dur ~ 1 + (1 | group) + (1 | id), data = dat, family =
> "nbinom", easyFlag = TRUE, verbose = TRUE)
> 
> 
> No more MAX_NVAR_OFFSET problems.
> 
> Thanks again and sorry for the confusion.
> All the best,
> 

  I've added an 'extra.args' argument to the both the stable and the
alpha versions of glmmADMB (it may take ~ 24 hours for the change to
propagate to the versions of the packages that are available from R-forge).

  It's great that you got the alpha release to work.  I'm still a little
bit hung up on the issue of getting all the appropriate versions of the
binaries compiled & included ...

  Ben