Gaussian is an electronic structure program, starting from the fundamental laws of quantum mechanics, Gaussian predicts the energies, molecular structures, vibrational frequencies and molecular properties of molecules and reactions in a wide variety of chemical environments.
Gaussian is available as a module on Apocrita.
To run the default installed version of Gaussian, simply load the
module load gaussian
then run Gaussian, depending on the version the binary will be named differently:
# Gaussian g09_D01 g09 < input-file > output_file # Gaussian g16 g16 < input-file > output_file
QMUL have a cluster licence for Gaussian installed on Apocrita, this licence only allows use by QMUL employees.
QMUL have a licence for GaussView, requests for this must go via the IT Services Helpdesk.
When the Gaussian module is loaded in a job, it will automatically set
GAUSS_PDEF (for Gaussian g09_D01) or
GAUSS_CDEF (for Gaussian g16) to
the value of
SGE_BINDING, which contains a list of the CPU cores bound
to the job. These variables ensure that Gaussian will only use the cores
requested by the job script.
Note that any core settings on the command line or in the input file will override this.
Please be aware that if you use the
nproc option, this must use the
NSLOTS (number of cores requested by the job script), to avoid
overloading the compute nodes.
If Gaussian returns a Segmentation Violation then you may need to give it more memory. Memory for Gaussian is specified in two places, the job script and the .com file:
#$ -pe smp 4 #$ -l h_vmem=2G
The %mem request must be 1-4G less than the h_vmem request
Gaussian uses about 1-4GB already so the memory stated in the
must be less than the SGE request.
If your jobs run out of memory try setting
%mem to a lower value.
It's important to note that whilst the job script
h_vmem is multiplied by
number of cores, the
.com file is not and the full value must be specified.
Here is an example job running on 4 cores.
#!/bin/bash #$ -cwd #$ -j y #$ -pe smp 4 #$ -l h_rt=1:0:0 #$ -l h_vmem=2G module load gaussian g16 < input-file > output_file