[ADMB Users] Installed glmm, admb 0.6 beta can't find function "glmm.admb"

[Ben Bolker]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

  [forwarding back to users group, for posterity ...]

On 11-07-29 05:31 PM, Shawn McCracken wrote:
> Hello Ben,
> 
> Thanks for your response.
> 
> So I just tried it again and here is what I get:
> 
>> fm <- glmmadmb(y ~ Base * trt + Age + Visit + (Visit | subject), data
> = epil2, family = "nbinom")
> Error in process_randformula(formula, data = data) : 
>   all grouping variables must be factors

  glmmadmb is now fussier than before about insisting that grouping
variables be factors. I could have made the software smarter, but it's
easier to make users convert their grouping variables into factors ...
(so just use epil2$subject <- factor(epil2$subject))

>> fm <- glmmadmb(y ~ Base * trt + Age + Visit + (Visit | subject),
> family = "nbinom", data = epil2)
> Error in process_randformula(formula, data = data) : 
>   all grouping variables must be factors

   same again (this looks identical, except for switching the order of
the last two arguments, which should never matter as long as they're
both named)

>> fm <- glmmadmb(y ~ Base * trt + Age + Visit + (Visit | subject),
> family = nbinom, data = epil2)
> Error in glmmadmb(y ~ Base * trt + Age + Visit + (Visit | subject),
> family = nbinom,  : 
>   object 'nbinom' not found
> 

  "nbinom" needs to be in quotation marks.

> I tried changing the syntax a bit and no luck. I would offer to compile
> a TPL file but wouldn't have a clue where to start. Also, in your
> document dated April 26,2011 under 'Overview of user guides and package
> vignettes it still shows glmm.admb in the example model. 

  Oops, will fix.

> Alright, I just
> realized the 32-bit version of R is also installed on my computer and so
> I tried that. Same errors as above. I know you like to keep these posts
> on the mailing lists, should I resubmit with our email test copied in?

  Can you try

epil2$subject <- factor(epil2$subject)
fm <- glmmadmb(y ~ Base * trt + Age + Visit + (Visit | subject),
    family = "nbinom", data = epil2)

 with the 32-bit version?

> 
> I am so very eager to get to try the new version out with multiple
> random effects! Will it also handle nested effects?

  I think so (but again, not thoroughly tested, so let me know how it goes).

   One warning is that the new version will be a bit slower than the old
for single-random-effect cases -- we can probably fix that, but I will
have an easier time twisting Hans and Dave's arms to work on it if we
can get the package out and get people using it (and complaining).

  cheers
    Ben


> 
> Thank you,
> 
> Shawn
> 
> 
> On Fri, Jul 29, 2011 at 4:10 PM, Ben Bolker <bbolker at gmail.com
> <mailto:bbolker at gmail.com>> wrote:
> 
> 
>  I can't seem to find the original message now, so the thread may get
> broken, but ...
> 
>  The name of the primary function has changed from "glmm.admb" to
> "glmmadmb" in the new release.
> 
>  However, I'm sorry to say that I think the development version of the
> package won't work for you on a MacOS 64-bit machine because we don't
> have the binary built.  (If someone wants to compile a TPL file from the
> current SVN version of ADMB (10.1 won't do) on a 64-bit MacOS platform
> and send me the results, I can send the TPL file).
> 
>  We are on the verge (I hope, with breath held and fingers crossed) of
> having the Buildbot automatically make these so I can incorporate them
> in the package ... Johnoel has been working hard ...
> 
>  Ben
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk4zJqEACgkQc5UpGjwzenMi8gCgiqtdfzRDkOnuhvuppm1yHmSB
vMwAn0Rl5uOdHDoKntmQoQyp5yG59Krw
=cJhA
-----END PGP SIGNATURE-----