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

[dave fournier]

Assuming that you got the formula right (I don't have much interest in 
arcane R stuff),
the problem is that ADMB programs were designed to be run from the 
command line.
The program needs the command line option

                 -mno  N

where N >= 58531,  lets say N=60000 to make it simple.
The R part of glmmadmb should be able to figure this out from the number 
of random effects,
but that is not implemented to date.  Maybe someone wants to do that?

Failing that you can run the model by hand by saving the temporary files and
running the model from the command line such as

      xxxx -mno 60000

replace xxx with the current name of the exe.
Then you can find the usual output in  a file named xxxx.std