UK Logo High Performance Computing November 23, 2009
Home News System Status User Policies Account Info Hardware Software Documentation FAQs Search
UKy HPC Frequently-Asked Questions

Frequently-Asked Questions related to:

For questions about this faq, please contact: help-hpc@uky.edu


FAQ Revised: Tuesday 26 April 2005 16:11:35

Table of Contents

1. Gaussian
2. Gaussian 2003

1. Gaussian

1.1. How do I submit a Gaussian job on the HP Superdome cluster ?

All Gaussian jobs must be submitted as batch jobs running under LSF. There is a C-shell script called batchg03 that automatically creates and submits a batch job to run Gaussian under LSF. To submit a Gaussian job using the batchg03 script, type:

    batchg03 filename [queue] [bsub options]

where

filename is the name of your Gaussian input file with or without the .com extension.

queue is the LSF queue to which you want to send your Gaussian job. Specifying a queue is optional; if no queue is specified, the job will be sent to gauss8l. The list of queues that are available for Gaussian runs can be found in the LSF Queue Table.

bsub options are optional. Any options from the bsub command (other than the -q option) may be used.

1.2. How do I combine multiple Gaussian jobs within a single input file ?

Multiple Gaussian calculations can be combined using the Link 1 command. To combine multiple-step jobs, separate the input for each successive job from that of the preceding job by the separator line:

    %Chk = frequency
    #B3LYP/6-31G(d,p)  Freq

    Frequency Calculation on XYZ at STP

    0 1
    . . .

    --Link1--
    %Chk = frequency
    #B3LYP/6-31-G(d,p)  Geom=Checkpoint  Guess=Read  Freq=(ReadFC,ReadIsotopes)

    Frequency Calculation on XYZ at Elevated Temperature

    0 1

    450.0 1.0

Note that a blank line must be included between the last line of the first job and the --Link 1-- separator line. Failure to include a blank line will result in a segmentation violation error when the Gaussian job is run.

Note also that if the checkpoint file from the first job is to be used in a subsequent job, the names specified with the %Chk command in the Link 0 section of each job must be identical. Otherwise, the second job will not be able to find the checkpoint file from the first job, which will cause the second job to fail.

1.3. I am running a Gaussian job on the HP Superdome cluster and I need to retrieve information from a checkpoint file that was created on a different computer operating system. Can I use this checkpoint file in my Gaussian job on the HP Superdome cluster ?

There is a Gaussian utility called chkmove that will convert a binary checkpoint file into and out of a (non-readable) text format and allow you to transfer your checkpoint file between different computer operating systems. To transfer a checkpoint file named my_checkfile.chk between machines ABC and the HP Superdome cluster:

  • Convert my_checkfile.chk from a binary to a text format on machine ABC by issuing the command:

    chkmove f my_checkfile.chk my_checkfile.xfr

    where f indicates that the direction of conversion is from a binary to a text format and the extension .xfr is the conventional extension for text formatted checkpoint files.
  • Transfer my_checkfile.xfr from machine ABC to NCX.
  • Convert my_checkfile.xfr from a text format back to a binary format on ABC by issuing the command:

    chkmove u my_checkfile.xfr my_checkfile.chk

    where u indicates that the direction of conversion is from a text to a binary format.

The chkmove utility is located in a release-dependent directory the HP Superdome cluster. Before running it, run "source /usr/local/env/gaussian.csh" to make sure the appropriate paths and environment variables are set - this is not done automatically, in case someone needs to use an earlier release of the Gaussian software for which the setup is somewhat different.

Note that chkmove will only work on checkpoint files from completed calculations. If your checkpoint file is from a failed job or from some type of intermediate calculation (e.g., unconverged optimizations or numerical frequencies), this utility will not convert your checkpoint file for transfer between different machines.

1.4. How do I convert my binary checkpoint file into a format that I can actually read ?

There is a Gaussian utility called formchk that will convert the data in a Gaussian 03 binary checkpoint file into a formatted form that is readable. To reformat the binary checkpoint file my_checkfile.chk into a format that is readable, issue the command:

formchk [-c] my_checkfile.chk my_checkfile.fchk

where the extension .fchk is the conventional extension for formatted checkpoint files on UNIX systems.

[-c] is an option used to cause the molecular mechanics atom types to appear in the formatted checkpoint file as strings rather than integers.

The formchk utility is located in a release-dependent directory on the HP SuperDome cluster.Before running it, run "source /usr/local/env/gaussian. csh" to make sure the appropriate paths and environment variables are set - this is not done automatically, in case someone needs to use an earlier release of the Gaussian software for which the setup is somewhat different.

1.5. I wrote a checkpoint file to my scratch directory; it wasn't there when the job finished. What happenned?

If it was written to your main scratch directory ( /scratch/user, where user is your userid, look in the chk_archive directory. Checkpoint files can be large, and people were not cleaning them up; much time was spent copying them around. Now, part of what LSF does when it copies files back up to the login host is to put any *.chk files it finds in a user's main scratch directory into the chk_archive directory, which isn't copied down to the execution host. If you need to use a *.chk file with a job, just move it out of the chk_archive directory before you submit the job.

1.6. How may I access the Gaussian utilities, such as formchk?

The Gaussian utilities are located in a release-dependent on the HP SuperDome cluster. Before running a utility, run "source /usr/local/env/gaussian.csh" to make sure the appropriate paths and environment variables are set - this is not done automatically, in case someone needs to use an earlier release of the Gaussian software for which the setup is somewhat different. You do not need to do this each time a utility is run - just once per login session. If you do not need to use an earlier release of the software, you can add it to one of your startup files (such as .cshrc) to have it done automatically when you log in.

1.7. How do I change the number of processors used from the default (8)?

There are two steps. First, use %nproc=X (where X is the desired number of processors) in your .com file specify the number of processors to Gaussian. Then use the -n X (where X is the number of processors) option when submitting the job with batchg03. The numbers should match. If the -n option is omitted, batchg03 will read your command file to get the number of processors to use. If you specify both %nproc and -n, but the numbers of processors requested do not match, the job will not be submitted.


2. Gaussian 2003

2.1. When will the UK HP system offer Gaussian 2003?

8/09/04

After considerable delay, we have received Gaussian 2003 from the vendor. It has been compiled from source code and is currently undergoing testing and validation. Michael Sheetz is the primary consultant for testing the computational results.

8/25/04 Gaussian 2003 has been released for general use.
There are some known issues with the Periodic Boundary Conditions (PBC) that have been reproduced on other platforms.


FAQ generated by: makefaq