HPC Gaussian README
Supercomputer documentation is always a work in progress! Please email questions, corrections, or suggestions to the HPC support team at firstname.lastname@example.org as usual. Thanks!
See also the Gaussian Frequently Asked Question list.
Getting Started with Gaussian
You should be familiar with the DLX cluster at the University of Kentucky, Red Hat LINUX, bash, and the vi editor.
Gaussian 09, distributed by Gaussian, Inc., is one of the most diverse quantum chemistry codes available, because of the wide range of properties it can compute. The code was originally developed by John Pople at Carnegie-Mellon University in 1970. Slater-type orbitals can be used in G98, G03, and G09 by selecting a STO-nG basis set, which is the default basis set in Gaussian. Some of the properties Gaussian 09 can compute include molecular structure, vibrational frequencies for systems in gas phase and solution, ground and excited states, reaction mechanisms, and electronic transitions. It can also perform semi-empirical geometry optimizations, making it possible to work with larger molecules.
On the DLX cluster Gaussian 09 is normally run on the Hi-Mem nodes, using 8 cores by default. Each Hi-Mem node has four sockets with eight cores each for a total of 32 cores per node. Each node has 512 GB of memory. A different numbers of cores can be used by specification in the input file. Gaussian is designed to run on a single node. Gaussian 09 has a companion software package, GaussView, which can be used to generate input files and display structures, though this is not required. The input file can be constructed with any Unix text editor, along with a list of atomic coordinates.
For more detailed Gaussian 09 information see the vendor web page at http://www.gaussian.com/g_prod/g09.htm.
For a general introduction to Gaussian, beginning users are advised to read the first two chapters of the book Exploring Chemistry with Electronic Structure Methods (Foresman and Frisch, Gaussian, Inc. 1998), and any section related to their work. Though the book is outdated, the basics are still the basics and it is still available on Amazon.
There is also an online help feature, but Gaussian must be loaded for this to work. Entering the ghelp command will display general information and a list of topics available. Entering ghelp topics will display the list of topics.
Setting Up Gaussian 09
The Gaussian module must be loaded with the module load gaussian command before anything associated with Gaussian can be run. This can be done by manually, once per logon, or you may have it loaded automatically each time you logon by editing the .bashrc file in your home directory. Add the following code:
module load gaussian
When you submit a job with the batchg09 it will load the Gaussian module automatically.
Gaussian 09 input Files
Each Gaussian job requires an input file, known as an input deck, specifying the type of calculation, basis set, and molecular specification of the input data. Input files contain ASCII text only and are prepared using a text editor. Gaussian input files require a filename with a .gjf or .com extension, for example, ethane.gjf and water.com.
An example input file for a Restricted Hartree-Fock single point energy calculation outputting orbital coefficients on a ZnO cluster using the 6-31G Gaussian basis set is listed below.
# RHF/6-31G* Pop=Full Test
All orbital coefficients for 44 atom ZnO cluster
O 0.000000 9.670617 9.787500
O 3.350000 9.670617 9.787500
O 6.700000 9.670617 9.787500
O -1.675000 12.571802 9.787500
O 1.675000 12.571802 9.787500
O 5.025000 12.571802 9.787500
O 0.000000 7.736494 12.397500
Zn 0.000000 7.736494 10.440000
O 1.675000 6.769432 15.007500
Zn 1.675000 6.769432 13.050000
O 3.350000 7.736494 12.397500
Zn 3.350000 7.736494 10.440000
O 5.025000 6.769432 15.007500
Zn 5.025000 6.769432 13.050000
O 6.700000 7.736494 12.397500
Zn 6.700000 7.736494 10.440000
O -1.675000 10.637679 12.397500
Zn -1.675000 10.637679 10.440000
O 0.000000 9.670617 15.007500
Zn 0.000000 9.670617 13.050000
O 1.675000 10.637679 12.397500
Zn 1.675000 10.637679 10.440000
O 3.350000 9.670617 15.007500
Zn 3.350000 9.670617 13.050000
O 5.025000 10.637679 12.397500
Zn 5.025000 10.637679 10.440000
O 6.700000 9.670617 15.007500
Zn 6.700000 9.670617 13.050000
O -1.675000 12.571802 15.007500
Zn -1.675000 12.571802 13.050000
O 0.000000 13.538864 12.397500
Zn 0.000000 13.538864 10.440000
O 1.675000 12.571802 15.007500
Zn 1.675000 12.571802 13.050000
O 3.350000 13.538864 12.397500
Zn 3.350000 13.538864 10.440000
O 5.025000 12.571802 15.007500
Zn 5.025000 12.571802 13.050000
The first line indicates the number of processors. The second, the amount of memory. The memory specification can be in KB, MB, or GB.The third line renames the .chk file so it will not be automatically deleted at the completion of the run, the fourth indicates the type of calculation and basis set. Line five must be left blank. The sixth line is a title. Line seven is also blank. The remaining lines contain the molecular charge, spin multiplicity, atom types, and Cartesian coordinates. Input text is not case sensitive and extra whitespace, such as spaces or tabs, is allowed within a line. Sections of the input file are separated by blank lines. Coordinates can be specified in a Cartesian, a Z-matrix format, or a mixed format. A blank line must be included at the end of the input file or else Gaussian 09 will not run.
Consult the Gaussian User's Reference for proper keyword usage and input file format.
By default Gaussian jobs go to the gauss queue, but if you need memory between 8 GB to 16 GB, use the gauss_bigmem queue when submitting the job as described below. If you need memory larger than 16 GB consult with HPC help first.
Running a Gaussian job
All jobs must be run in batch mode which means you will have to run a script to start the job. Use the batchg09 seedname script to submit a Gaussian job.
The memory requirements can be specified in the Gaussian input file along with the number of processors. The default values are 2.5 GB. and 8 cores. Running a job on more than 8 processors actually reduces efficiency in some cases.
You can monitor the job's progress in the queue using the squeue command.
Gaussian jobs must be run from scratch disk. To do this copy all of the input files to your scratch disk (/home/userid/scratch), then change directory to the scratch disk (cd /scratch). Run the batchg09 seedname script to submit the job. After the job finishes you must copy all of the files back to your home disk and clean up your scratch disk. Then you can process the output stored in the seedname.log file.