<html>
<body>
<font size=3>All sorry,<br><br>
Just looked at the help file for glmm.admb. Should have done this
before I posted. I see that the ability to pass args is
there. But even more interesting to me is the "save.dir"
argument that Ben mentions below. That is really nice and seems
designed just for this. I am very curious what that is actually
doing beneath the surface since my interest is for other admb
applications not just the glmmADMB package in R. When I look
through available admb argsI don't see one that does this. Is this
changing env variables or...?<br><br>
Jim<br><br>
<br><br>
At 04:55 PM 7/29/2011, Ben Bolker wrote:<br>
<blockquote type=cite class=cite cite="">-----BEGIN PGP SIGNED
MESSAGE-----<br>
Hash: SHA1<br><br>
Yes, separating the examples into separate subdirectories should
do<br>
the trick. I *think* you should also be able to specify the
"save.dir"<br>
argument to take care of this, but I admit I haven't tested it
...<br><br>
Ben Bolker<br><br>
<br>
On 11-07-29 04:03 PM, Weihai Liu wrote:<br>
> We can run the same admb program for different dataset by using the
job<br>
> array, also called pleasantly parallel on our hpcc, it works fine if
you<br>
> don't need each job within the array to communicate with other
job<br>
> during the run. More details see<br>
> link
<a href="https://wiki.hpcc.msu.edu/display/~colbrydi@msu.edu/2011/07/08/Pleasantly+Parallel+Presentation" eudora="autourl">https://wiki.hpcc.msu.edu/display/~colbrydi@msu.edu/2011/07/08/Pleasantly+Parallel+Presentation</a>,<br>
> click the pdf for more slides.<br>
> <br>
> I am not using glmmadmb, but you can try to separate each dataset in
its<br>
> own directory, then call the program or submit the job for
each<br>
> directory, since admb has not implement parallel yet and create
the<br>
> temporary files during its run. <br>
> <br>
> <br>
> weihai<br>
> <br>
> On Fri, Jul 29, 2011 at 2:38 PM, Tara Crewe <tcrewe2@uwo.ca<br>
>
<<a href="mailto:tcrewe2@uwo.ca" eudora="autourl">mailto:tcrewe2@uwo.ca</a>>>
wrote:<br>
> <br>
> Good afternoon,<br>
> <br>
> I'm running glmmADMB (0.5-2) with negative
binomial and zinegbin<br>
> distribution on count data in R (version
2.12.2). Currently I'm<br>
> using a cluster
(<a href="http://www.sharcnet.ca/" eudora="autourl">www.sharcnet.ca</a>
<<a href="http://www.sharcnet.ca/" eudora="autourl">http://www.sharcnet.ca</a>>)
to submit<br>
> multiple datasets to be analyzed by the same
R program in parallel<br>
> (i.e., serial farming).<br>
> <br>
> Other R packages work fine in parallel this
way (e.g., glmer), and<br>
> the code for ADMB works fine on a single
dataset. However, as soon<br>
> as I submit a second dataset to be analyzed
in parallel, ADMB fails<br>
> with the following error messages:<br>
> <br>
> Error in dvar_matrix write<br>
> Error in glmm.admb(ObservationCount ~ c.year
+ poly(c.day, 2) +<br>
> poly(c.hour, :<br>
> The function maximizer
failed<br>
> Calls: glmm.zinb.trnd.admb.yr ->
glmm.admb<br>
> In addition: Warning messages:<br>
> 1: In system(cmd2, intern = !verbose)
:<br>
> line 1 may be truncated in call
to system(, intern = TRUE)<br>
> 2: In system(cmd2, intern = !verbose)
:<br>
> line 1 may be truncated in call
to system(, intern = TRUE)<br>
> 3: In system(cmd2, intern = !verbose)
:<br>
> line 1 may be truncated in call
to system(, intern = TRUE)<br>
> 4: In system(cmd2, intern = !verbose)
:<br>
> line 1 may be truncated in call
to system(, intern = TRUE)<br>
> 5: running command './nbmm -maxfn 500 ' had
status 1<br>
> Execution halted<br>
> <br>
> I searched these error messages online, and
haven't found any<br>
> documentation. Question is, is it
possible to run ADMB in<br>
> parallel? I suspect (with my very
limited knowledge) that the error<br>
> may have to do with how temporary files are
written/stored by ADMB?? <br>
> <br>
> I haven't attached sample data, because for
now I'd simply like to<br>
> know if parallel processing is possible with
ADMB? If so, is there<br>
> documentation that could help me out?
I am also happy to supply<br>
> data and R code if it will help.<br>
> <br>
> Thanks so much,<br>
> <br>
> Tara<br>
> <br>
> --<br>
> Tara L. Crewe, PhD Candidate<br>
> <br>
> Department of Biology<br>
> University of Western Ontario<br>
> 1151 Richmond Street<br>
> London, Ontario, Canada,<br>
> N6A 3K7<br>
> <br>
> Tel: 519-661-2111 x 84646
<tel:519-661-2111%20x%2084646><br>
> email: tcrewe2@uwo.ca
<<a href="mailto:tcrewe2@uwo.ca" eudora="autourl">mailto:tcrewe2@uwo.ca</a>><br>
>
_______________________________________________<br>
> Users mailing list<br>
> Users@admb-project.org
<<a href="mailto:Users@admb-project.org" eudora="autourl">mailto:Users@admb-project.org</a>><br>
>
<a href="http://lists.admb-project.org/mailman/listinfo/users" eudora="autourl">http://lists.admb-project.org/mailman/listinfo/users</a><br>
> <br>
> <br>
> <br>
> <br>
> _______________________________________________<br>
> Users mailing list<br>
> Users@admb-project.org<br>
>
<a href="http://lists.admb-project.org/mailman/listinfo/users" eudora="autourl">http://lists.admb-project.org/mailman/listinfo/users</a><br><br>
-----BEGIN PGP SIGNATURE-----<br>
Version: GnuPG v1.4.10 (GNU/Linux)<br>
Comment: Using GnuPG with Mozilla -
<a href="http://enigmail.mozdev.org/" eudora="autourl">http://enigmail.mozdev.org/</a><br><br>
iEYEARECAAYFAk4zHkIACgkQc5UpGjwzenMx9QCggIQZ5tlzSe2yZ/5P7Yw95TUA<br>
Ec8AoIWiNx66p34/72m01oBstbg6JQzL<br>
=eDyF<br>
-----END PGP SIGNATURE-----<br>
_______________________________________________<br>
Users mailing list<br>
Users@admb-project.org<br>
<a href="http://lists.admb-project.org/mailman/listinfo/users" eudora="autourl">http://lists.admb-project.org/mailman/listinfo/users</a>
</blockquote>
<x-sigsep><p></x-sigsep>
Jim Bence<br>
Dept. of Fisheries and Wildlife <br>
Michigan State University<br>
<a href="http://www.msu.edu/user/bence/" eudora="autourl">http://www.msu.edu/user/bence/</a></font></body>
</html>