Skip to main content
Home > Documentation > Software >gaussian

gaussian

Versions and Availability

▶ Display Module Names for gaussian on all clusters.

▶ Module FAQ?

▶ Did not find the version you want to use??

About the Software


Usage

Gaussian is run from the command line, and does not provide a graphical interface. Thus interactive and batch job usage is the same. The TCP Linda extension is required run Gaussian in parallel using more than 1 node per job. Currently only LSU has such a license.

Please refer to the FAQ on Common Problems below or the Gaussian User Manual for Memory Requirements for the your gaussian job.

An input file is used to specify the desired calculations. It may be as simple as:

%chk=water.chk

# HF/6-31G(d)

water energy              Title section

0   1
O  -0.464   0.177   0.0
H  -0.464   1.137   0.0
H   0.441  -0.143   0.0

Please refer to the program documentation for details.

One an input file has been created, the next step is creating a PBS or SLURM job file.

▶ QSub FAQ?

An example PBS batch job file follows:

 #!/bin/tcsh
 #PBS -A your_allocation
 # specify the allocation. Change it to your allocation
 #PBS -q checkpt
 # the queue to be used.
 #PBS -l nodes=1:ppn=4
 # Number of nodes and processors
 #PBS -l walltime=1:00:00
 # requested Wall-clock time.
 #PBS -o g09_output
 # name of the standard out file to be "g09_output".
 #PBS -j oe
 # standard error output merge to the standard output file.
 #PBS -N g09test
 # name of the job (that will appear on executing the qstat command).
 #
 # cd to the directory with Your input file
 cd ~USER/g09test
 #
 # Change this line to reflect your input file and output file
 g09 water.inp

An example SLURM batch job file follows:

 #!/bin/bash
 #SBATCH -A loni_loniadmin1
 # specify the allocation. Change it to your allocation
 #SBATCH -p checkpt
 # the queue to be used.
 #SBATCH -N 1
 #SBATCH -n 48
 # Number of nodes and processors
 #SBATCH -t 2:00:00
 # requested Wall-clock time.
 #SBATCH -o slurm-%x-%j.out-%N
 # name of the standard out file to be "slurm-g16-100460.out-qbc185".
 #SBATCH -J g16_job
 # name of the job (that will appear on executing the squeue command).
 #
 # cd to the directory with Your input file
 cd $SLURM_SUBMIT_DIR
 #
 # Change this line to reflect your input file and output file
 g16 myinput.com

Multi-node Job Submission

Below is an example PBS job script for running gaussian/g16-b01 on QB2

(Ref: https://github.com/ResearchComputing/Documentation/wiki/Gaussian#parallel-jobs) 

#!/bin/bash
#PBS -N g16job
#PBS -l nodes=2:ppn=20
#PBS -l walltime=01:00:00
#PBS -A your_allocation

module load gaussian/g16-b01

for n in $(cat $PBS_NODEFILE | uniq);
do
    echo ${n}
done | paste -s -d, > nodes.$PBS_JOBID

# the next line prevents OpenMP parallelism from conflicting with Gaussian's internal parallelization
export OMP_NUM_THREADS=1

# increases the verbosity of Linda output messages
export GAUSS_LFLAGS="-v"

cd $PBS_O_WORKDIR
date
g16 -p=20 -w=$(cat nodes.$PBS_JOBID) myinput
date

Below is an example SLURM job script for running gaussian/g16-b01 on QB3

(Ref: https://curc.readthedocs.io/en/latest/software/gaussian.html) 

#!/bin/bash
#SBATCH -p checkpt
#SBATCH -N 2
#SBATCH -n 48
#SBATCH -c 1
#SBATCH -t 2:00:00
#SBATCH -A loni_loniadmin1
#SBATCH -J g16
#SBATCH -o slurm-%x-%j.out-%N

module load gaussian/g16-b01

cd $SLURM_SUBMIT_DIR

for n in `scontrol show hostname | sort -u`; do
 echo ${n}
done | paste -s -d, > nodes.$SLURM_JOBID

# the next line prevents OpenMP parallelism from conflicting with Gaussian's internal parallelization
export OMP_NUM_THREADS=1

# increases the verbosity of Linda output messages
export GAUSS_LFLAGS="-v"

g16 -p=48 -w=$(cat nodes.$SLURM_JOBID)  myinput.com

#End-of-file (EOF)

Contents of myinput(or myinput.com):

#P b3lyp/6-31g* test stable=(opt,qconly)

Gaussian Test Job 135:
Fe=O perpendicular to ethene, in triplet state.

0 3
X
Fe X  RXFe
C1 X  RXC  Fe  90.
C2 X  RXC  Fe  90.  C1  180.
O  X  RXO  C1  90.  Fe 0.
H1 C1 RCH  C2 CCH   Fe  Angle1
H2 C1 RCH  C2 CCH   Fe -Angle1
H3 C2 RCH  C1 CCH   Fe  Angle2
H4 C2 RCH  C1 CCH   Fe -Angle2

RXFe  1.7118
RXC   0.7560
RXO   3.1306
RCH   1.1000
Angle1 110.54
Angle2 110.53
CCH   117.81

Below are examples for using g09

TCP Linda is required to run gaussian jobs on more than one node

#!/bin/bash
#PBS -q checkpt
#PBS -l nodes=2:ppn=16
#PBS -l cput=00:20:00
#PBS -l walltime=00:20:00
#PBS -o output-file
#PBS -j oe
#PBS -V
#PBS -N jobtest

export WORK_DIR=$PBS_O_WORKDIR
cd $WORK_DIR

cat $PBS_NODEFILE | sort | uniq > /tmp/.nodes.$PBS_JOBID
export GAUSS_LFLAGS="-nodefile /tmp/.nodes.$PBS_JOBID"

g09 < input.inp > output.log

In the sample input above, INPUT.inp, the Link 0 directives should be

%UseSSH
%chk=/work/mmcken6/g09.chk
%mem=16mw
%nprocshared=16  Note: must match ppn in the job submission script
%NprocLinda=2    Note: must match nodes in the job submission script
(rest of input)
...

Note that the "%UseSSH" directive is necessary for Linda jobs, which may fail otherwise. Alternatively, you can add the "-opt Tsnet.Node.lindarsharg: ssh" flag to the g09 command, which has the same effect for Linda jobs.

Resources

▶ Common Problems FAQ?

Last modified: September 17 2021 12:09:19.