Author Topic: mpshift fails for medium sized problems  (Read 8614 times)

resofidentity

  • Full Member
  • ***
  • Posts: 63
  • Karma: +0/-0
mpshift fails for medium sized problems
« on: January 18, 2007, 09:25:34 am »
Hello,

I have encountered frequently problems with mpshift for calculations using more than 500 basis functions
and b3lyp or mp2 methods. I suppose some files are getting too large or any memory problems, but I never was able
to localize the problem exactly. Here is what I got with an B3LYP/TZVPP/M4 with 728 cartesian bfs:

  1. iteration
==============
  csxx   -44.70155023  csyy   -53.54877794  cszz   -53.47960021
 errxx    -5.08463341 erryy    -8.19357229 errzz    -8.19080721
  converged: none
  from getini : eof encountered at record  1

 MODTRACE: no modules on stack

 fatal error in getini.
 mpshift ended abnormally


Has anyone experienced similar problems?

Thanks
             R

uwe

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 405
  • Karma: +0/-0
Re: mpshift fails for medium sized problems
« Reply #1 on: January 22, 2007, 11:36:10 am »
Hi,

this is written in the documentation, first page in the chapter about NMR chemical shifts:

3. it is not possible to run the program in the fully direct mode when doing
an SCF, MP2 or a DFT (using hybrid functionals) run, so you will have to
perform a statistics run of Dscf before calling Mpshift, or just set the size
of the $twoint file to a non-zero value

Regards,

Uwe

resofidentity

  • Full Member
  • ***
  • Posts: 63
  • Karma: +0/-0
Re: mpshift fails for medium sized problems
« Reply #2 on: January 29, 2007, 09:32:17 am »
Hello Uwe,

thank You! I should have been reading the manual more carefully! Now its working.

R