CAMd Summer school 2008

Databar

Setting up your UNIX environment

The first time you use the databar computers, you must do this:

$ ~jjmo/summerschool/setup.sh
$ source ~/.bashrc

That will set up the environment for you so that you can use ASE, GPAW, VMD and matlpotlib.

Warning runnig ~jjmo/summerschool/setup.sh owervrites users ~/.bashrc ~/.emacs ~/.pythonrc ~/.vmdrc and ~/.matplotlib directory.

Notes

  • Useful links: Userguides FAQ Unix USB-sticks

  • Editors: emacs, vim, nedit (MS Windows/Macintosh-like environment). Python syntax

  • Printer: gps1-308. Terminal: lp -d gps1-308 filename

  • E-mail client: Thunderbird is the default mail client in the databar and configured with your summer school e-mail (camd0??@student.dtu.dk).

  • To open a pdf-file: acroread filename

Niflheim

Frontend nodes

Log in to niflheim:

ssh school1.fysik.dtu.dk

or:

ssh school2.fysik.dtu.dk.

Same password as handed out for the databar. Please use school1 if the number in your userid is odd and school2 if it is even.

Copying files from gbar to niflheim

You can copy files from the Gbar to niflheim with scp. If you are on niflheim:

scp hald.gbar.dtu.dk:path/filename .

will copy filename to your present location. Likewise:

scp school1.fysik.dtu.dk:path/filename .

will copy filename from Niflheim to your present location at the Gbar.

GPAW

Use the gpaw-qsub.py command to submit GPAW jobs to the queue.

SIESTA

Siesta is installed on Niflheim, so you need to log in to the Niflheim front-end nodes as described above in the Niflheim section. Furthermore you have to set two environment variables by adding the following two lines to your ~/.bashrc file:

export SIESTA_PP_PATH=~mvanin/asesiesta
export SIESTA_SCRIPT=~mvanin/asesiesta/run_siesta.py

and source it by typing:

$ source ~/.bashrc

To submit a job to Niflheim, use the qsub command:

$ qsub -l nodes=1:ppn=1:switch5 filename.py

Octopus

Octopus is installed on the ‘q’ opteron nodes on Niflheim. The way to run jobs is the following: Create inp file in the working directory as described in the tutorial, and then run octopus_run.py. To use various octopus utilities such as oct-cross-section and oct-broad you need to do:

source /usr/local/openmpi-1.2.5-pathf90/bin/mpivars-1.2.5.sh

first. Submitting jobs to the queue is done by:

qsub -l nodes=2:ppn=4:switch5 octopus_run.py