Please try to keep responses replied to the Amber list, and not individual
responders.
On Wed, Feb 27, 2013 at 9:29 AM, Rasha Alqus
<rasha.alqus.manchester.ac.uk>wrote:
> Dear Amber users,
>
> i would be thankfull if you could explain to me please how i do this
> (Also, the command-line flags that nmode takes will not be recognized
> byyour nab program -- you already hard-coded the file names into your nab
> program, so you can eliminate all of the command-line arguments)
>
One thing I didn't catch in your last email: what you were trying to use as
a nmode input file was in fact a nab program.
The instructions were present in my last email, but I will summarize here:
1) You must compile your NAB program using "nab". You called this nab
program G17-vac-nmde.in. I would recommend changing that file name
to G17-vac-nmde.nab -- to demonstrate that it is a nab program.
mpinab G17-vac-nmde.nab
2) Use this new "a.out" file in your job script -instead- of nmode. So
your job file looks like this:
### SGE Job Stuff
#$ -cwd
#$ -V
### The next line submits to the amber11 pe and tells SGE how many
processors are required
#$ -pe amber11 4
cd $SGE_O_WORKDIR
mpiexec -n $NSLOTS ./a.out
Notice the SGE_O_WORKDIR command -- this will automatically change into the
directory that you submitted the job from (make sure you compile your nab
program inside this directory and that the files you provided in your nab
program are present in that directory).
One note, your input file imposes very strict convergence criteria on the
Newton-Raphson minimization. For big systems, this is likely too strict,
and your system will never finish minimizing (but maybe not). This is
something to keep in mind while debugging.
Good luck,
Jason
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Feb 27 2013 - 07:30:04 PST
