Gaussian

Warning

Gaussian is proprietary software. Make sure you meet the requirements for it's usage.

Available Modules¶

Mahuika

module load Gaussian/09-D.01

Description¶

The Gaussian series of programs provides state-of-the-art capabilities for electronic structure modelling. It is used by chemists, chemical engineers, biochemists, physicists and other scientists worldwide. 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's models can be applied to both stable species and compounds which are difficult or impossible to observe experimentally (e.g., short-lived intermediates and transition structures).

Licensing requirements¶

Gaussian is made available to researchers under closed-source, commercial licence agreements with individuals, research groups or institutions. Whether you have access to Gaussian, which versions you have access to, and under what conditions, will vary depending on where you work or study.

For the sake of compliance with Gaussian licence agreements, we maintain a special Gaussian UNIX group. Only members of this group may access and use Gaussian. Contact our Support Team to join the Gaussian group.

All University of Auckland staff and students are in the Gaussian group automatically. If you are not a staff member or student at the University of Auckland, we will add you to the Gaussian group if we are satisfied that you require access to Gaussian to carry out your research and that your institution's Gaussian licence agreement permits you to use Gaussian on a computer that is not owned by or housed at your institution. We may at any time remove you from the Gaussian group if we believe these conditions are no longer met.

If you have any questions regarding your eligibility to access Gaussian or any particular version or installation of it, please Contact our Support Team.

Example jobs¶

The following job submission script is intended for use on Mahuika. Please note that it has a memory requirement built in: at least 2 GB for Gaussian itself, plus a further 2 GB as a buffer zone, for a minimum request of 4 GB (4,096 MB).

Any Gaussian input file must end with a blank line. We also recommend specifying a checkpoint file using the %Chk directive, as a saved checkpoint file facilitates recovery and restart if your Gaussian job fails or is killed by the scheduler. In this case, the value of the checkpoint file is a placeholder (as are the number of cores and the memory) and is replaced with a real value when the Slurm job starts.

Shared Memory¶

#!/bin/bash -e

#SBATCH --job-name=H2O_shared_memory
#SBATCH --account=nesi99999
#SBATCH --time=00:15:00
#SBATCH --cpus-per-task=8       # Note, Gaussian will use twice the number of CPUs specified by --cpus-per-task.
#SBATCH --hint=nomultithread
#SBATCH --mem=8G

module load Gaussian/09-D.01

INPUT_FILE="H2O.gjf"

GAUSSIAN_MEM="$((${SLURM_MEM_PER_NODE} - 2048))"

# It is reconmended to prepare a job-specific scratch directory
export GAUSS_SCRDIR="/nesi/nobackup/${SLURM_JOB_ACCOUNT}/gaussian_job_${SLURM_JOB_ID}"
mkdir -p "${GAUSS_SCRDIR}"

cat << EOF > $INPUT_FILE

%CPU=$(taskset -cp $$ | awk -F':' '{print $2}')
%Mem=${GAUSSIAN_MEM}MB
%Chk=${INPUT_FILE}.chk

# HF/6-31G(d) Opt=ModRedun Test

water geo optimisation HF/6-31G(d)

0 1
H
O 1 0.95
H 2 0.95 1 109.0


EOF

srun g09 < "${INPUT_FILE}"

Distributed Memory¶

#!/bin/bash -e

#SBATCH --job-name=H2O_distributed_memory
#SBATCH --account=nesi99999
#SBATCH --time=00:15:00
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=4
#SBATCH --hint=nomultithread
#SBATCH --mem=4G

module load Gaussian/09-D.01

INPUT_FILE="H2O.gjf"

GAUSSIAN_MEM="$((${SLURM_MEM_PER_NODE} - 2048))"

# It is reconmended to prepare a job-specific scratch directory
export GAUSS_SCRDIR="/nesi/nobackup/${SLURM_JOB_ACCOUNT}/gaussian_job_${SLURM_JOB_ID}"
mkdir -p "${GAUSS_SCRDIR}"

cat << EOF > $INPUT_FILE

%LindaWorkers=$(for n in $(srun hostname | sort -u);do printf "${n}:${SLURM_NPROCS},"; done)
%Mem=${GAUSSIAN_MEM}MB
%Chk=${INPUT_FILE}.chk

# HF/6-31G(d) Opt=ModRedun Test

water geo optimisation HF/6-31G(d)

0 1
H
O 1 0.95
H 2 0.95 1 109.0


EOF

srun g09 < "${INPUT_FILE}"

Further notes¶

Setting the memory and number of cores¶

It is important to ensure that the memory and number of cores in the Gaussian input file itself are consistent with what you set in your job submission script.

The key properties are %CPU and %Mem:

%CPU should be set to the number of CPU cores you intend to use, matching the value of the -c or --cpus-per-task directive in the Slurm job file. Because Slurm assigns which CPUs of the compute node your jobs lands on, the 'taskset' command is needed to identify what CPUs are available to your job.
%Mem should be set to the amount of memory you intend to use. It should be about 2 GB (2,048 MB) less than the value of --mem in the Slurm job submission script. Note that --mem is interpreted as being in MB rather than GB unless otherwise specified (i.e., with a "G" on the end).

If you use the example Slurm script and template .gjf file provided above (with appropriate modifications for your chemical system and desired calculation), this should happen automatically.

Saving temporary working files (for advanced users)¶

If you want Gaussian's temporary files (*.inp, *.d2e, *.int, *.rwf and *.scr) to be written to a particular directory, you can achieve this by setting the GAUSS_SCRDIR environment variable in your job submission script, for instance:

export GAUSS_SCRDIR=/nesi/nobackup/nesi99999/mahuika_job_123456

This should happen automatically if you use an appropriately modified script based on the example job submission script given above.