[R-sig-ME] glmmADMB v 0.7.1

Ben Bolker bbolker at gmail.com
Fri Jan 13 18:32:17 CET 2012


dave fournier <davef at ...> writes:

> 
> When you run glmmadmb in R, R writes ADMB's files and runs the glmmadmb 
> program
> via system(). There is an option to keep these files and then to run 
> glmmadmb from outside
> R. You can then read the results back into R. Doing it this way (i.e. 
> exiting R before
> running glmmadmb) will give you a bit more memory. It may be enough.
> That is the simplest first thing to try I think.
> 
> 
  specifically, you should be able to specify 

..., save.dir="tmp", admb.opts=admbControl(run=FALSE), ...

which will create the temp directory with everything you need in it.
Then run ADMB from the terminal/command window.

  Here's an example:

data(bacteria,package="MASS")
bacteria$present <- as.numeric(bacteria$y)-1
## run to generate files
glmmadmb(present ~ trt + I(week > 2), random = ~ 1 | ID,
                     family = "binomial", data = bacteria,
         save.dir="tmp",admb.opts=admbControl(run=FALSE),
         debug=TRUE)
## now run glmmadmb outside of R ...
## ./glmmadmb -maxfn 500 -maxph 5 -noinit -shess
## run to read in data
result <- glmmadmb(present ~ trt + I(week > 2), random = ~ 1 | ID,
                     family = "binomial", data = bacteria,
         save.dir="tmp",admb.opts=admbControl(run=FALSE))




More information about the R-sig-mixed-models mailing list