[R-sig-ME] glmmADMB error

[Ben Bolker]

Nathan Lemoine <lemoine.nathan at ...> writes:

> I'm not sure if this belongs here or on the R-SIG-MAC listserve.

  I'm not sure either, but I think possibly on the R-SIG-MAC listserve.

>  I'm  
> trying to run glmmADMB to analyze some data on my Mac that have been  
> collected over the past few years. The actual data itself is, I think,  
> unimportant for the issue I'm having. I've successfully installed the  
> glmmADMB package, but the function glmmadmb() does not run on my  
> computer. Here's the output I get from attempting to run glmmadmb():
> 
> library(glmmADMB)
> model1.l<-glmmadmb(recruits~deer*weeds*use+(1| 
> age),family="nbinom",zeroInflation=TRUE,data=recruits)
> 
> dyld: unknown required load command 0x80000022
> Error in glmmadmb(recruits ~ deer, data = recruits) :
>    The function maximizer failed
> 
> I'm not really sure what this error means, I've not seen anything like  
> it before. It doesn't appear to be an issue with R, but rather  
> something deeper. I've tried running the examples in the vignette,  
> I've tried zeroInflation=FALSE, family="poisson", and a number of  
> other options. I get the same error every time. I've done a little  
> digging, and it seems this is a common error message when older Mac  
> operating systems attempt to run software developed for newer versions  
> of OS X.
> 
> I'm running Max OS X 10.5.8, R v 2.13.2, and glmmADMB v 0.6.7.1. I'm  
> not sure if this version of glmmADMB was written for OS X 10.6, but  
> has anyone encountered this issue and found a fix for it? I've tried  
> several fixes, including those I found on this thread:


> 
> http://groups.google.com/group/admb-users/browse_thread/thread/0df5779586e45b9b
> 
> but it appears that the files on that thread were compiled for OS X  
> 10.6 and may still not be compatible with my OS.

  Can you try the workarounds suggested at

http://asqueella.blogspot.com/2010/12/dyld-unknown-required-load-command.html

  ?

  If any of them work, that will be a big clue.

> I currently have a workaround for my data analysis that let's me  
> analyze the data, although it's less than satisfactory and I'd like to  
> give glmmADMB a try to see if it matches my workaround.


  This is actually very helpful, because someone else reported
a similar problem: I (mis)diagnosed it as a 32/64 bit problem,
but it could be a 10.6/10.5 problem.

  Actually, let's take this off-list -- please try running
with debug=TRUE, and try the workarounds above, and let me know
how you do.

  Ben Bolker