Author Topic: BSSE - One atom calculations in parallel  (Read 4233 times)

Ian K

  • Jr. Member
  • **
  • Posts: 16
  • Karma: +0/-0
BSSE - One atom calculations in parallel
« on: August 19, 2010, 02:25:44 pm »
I'm trying to run BSSE calculations, and one of my fragments is one plutonium atom. So, I divided it all up, redefined the plutonium atom calculation to have the right multiplicity, and then tried running it (in parallel, using MPI), and got the following message:

it makes no sense to run one atom  calculations
 parallel, routine ATGRD


Now, running the same calculation in serial works, but my other fragment is appreciably larger, and I'd like to just be able to run the whole thing off jobbsse, in parallel. While it may be a little silly running a one atom calculation in parallel, it's not exactly senseless, especially in this context, so I was wondering whether there's a way to get around this short of defining another atom with no basis functions and no charge or something similarly hacky. (Also, why does this check seemingly arbitrarily refuse calculations in this way? Does it present problems in the decomposition algorithm or something similar?)

Arnim

  • Developers
  • Sr. Member
  • *
  • Posts: 224
  • Karma: +0/-0
Re: BSSE - One atom calculations in parallel
« Reply #1 on: October 11, 2010, 05:32:54 pm »
Hi Ian,

you are right! Such a case is not really considered.

If you add the following lines right under the #!/bin/sh in $TURBODIR/mpirun_scripts/ridft,
the calculations with the single atom will run sequentially:

 natoms=`sdg -b coord | wc -l`
 if [ "$natoms" = "1" ] ; then
    "$TURBODIR/bin/`sysname -s`/ridft"
    exit 0
 fi

Hope this helps,
Arnim

Ian K

  • Jr. Member
  • **
  • Posts: 16
  • Karma: +0/-0
Re: BSSE - One atom calculations in parallel
« Reply #2 on: October 13, 2010, 05:45:57 pm »
That appears to work nicely. Thanks, Arnim.