Bayz Manual

Using bayz in the AU-Foulum cluster

The AU-Foulum computing cluster runs the PBS system, google something like "linux PBS queue" or ask experts in the group with questions about the computing cluster.

When everything goes smooth ...

Larger computing task should not be run on the 'interactive' prompt, but should be submitted in the cluster queue. To do so:

Example bash script to run bayz

A minimum example script that runs bayz in your own home directory follows below. As this queue script runs bayz in your home directory, log and output file will accumulate their while bayz is running, but note that this can create a large output file in your home directory! This example is running a bayz script called 'analysis1.bayz'.

#!/bin/bash
#Script to run bayz in home directory (PBS_O_WORKDIR):
cd $PBS_O_WORKDIR
/usr/home/qgg/lucjanss/bayz/bayz analysis1

Call this bash script, for instance, run_bayz.sh, and in the directory where the bayz script is located, submit this bash script to the queue with 'qsub run_bayz.sh'. To run in different queues consult the experts that know about the queuing system.

A bash script where the output is not (initially) created in your home directory, but in a cluster temporary directory can look as follows:

#!/bin/bash
#Script to run bayz in the cluster scratch (temporary) directory
# The following is the location of a temporary directory for this queue job:
TMPDIR=/scratch/$USER/$PBS_JOBID
# Go to the temporary directory and copy the bayz file to run:
cd $TMPDIR
cp $PBS_O_WORKDIR/analysis1.bayz .
# run bayz:
/usr/home/qgg/lucjanss/bayz/bayz analysis1
# Now all output is in the temporary directory, copy it back to your home.
# Zip the output file first to reduce disk usage in your home dir.
gzip analysis1.out
cp analysis1.out.gz analysis1.lab analysis1.log analysis1.mod $PBS_O_WORKDIR