PAW on the Cobalt-Cluster

The general runscript paw_run

    ~program/paw/scripts/paw_run

If you have the path to the scripts directory in your standardpath you just type

    paw_run test

for test.inp being your input file.  This will NOT submit the job to a queue but just run it interactively on the
machine you are currently logged in. The script produces some output about the things it does, which might be
redirected in another file (as <>.echo). It also gives in the <>.host file on which machine it is running, which is
usefull when paw_run is used in the queueing-environement (see "other utilities" for more info about that)

The script performs roughly the following actions (when started with "paw_run test"):
 

• paw.atomparams masterfile is copied into the current dir. if necessary (user edited file is NOT overwritten)
• test.cntl and test.strc are generated
• a working directory  is generated called /local/scratch/<username>/test
• if it is NOT a conv_psi run a test.rstrt file is expected and copied to the working directory
• The job is run with the standard executable ~program/paw/bin/paw
• The test.prot is copied back (the working directory is NOT removed!!!)
• Depending on some ekywords other files are automatically copied back to the homedirectory