Author Topic: mpshift crashes with segfault in huge S9 symm calc.  (Read 3415 times)

resofidentity

  • Full Member
  • ***
  • Posts: 63
  • Karma: +0/-0
mpshift crashes with segfault in huge S9 symm calc.
« on: October 24, 2011, 12:37:37 pm »
Dear TM users,
I try to to run a GGA-DFT mpshift -job for a molecule with 1008 electrons. The molecule has S9-symmtry. When i understand the manual correctly this symmetry should be supported by mpshift. However, during or after the initial CAO calculation step the mpshift crashes with a short segfault on std-err, no further messages. I use 6.2/Linux/AMD-Opteron. Has anyone seen something like that before?
Thanx for reading
  r

ps. i also set $gimic

resofidentity

  • Full Member
  • ***
  • Posts: 63
  • Karma: +0/-0
Re: mpshift crashes with segfault in huge S9 symm calc.
« Reply #1 on: October 24, 2011, 04:37:01 pm »
pps.: It also crashes without $gimic and in c1. :-(

Now I tried it on top of a SCF calculation instead of DFT, this time I got an error message:

 Inversion problem in dvdson(osinv)
 
 MODTRACE: no modules on stack

 fatal error in dvdson.
 mpshift ended abnormally
 mpshift ended abnormally
program stopped.
« Last Edit: October 24, 2011, 06:38:24 pm by resofidentity »

uwe

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 405
  • Karma: +0/-0
Re: mpshift crashes with segfault in huge S9 symm calc.
« Reply #2 on: October 24, 2011, 09:01:03 pm »
Hi,

S9 would not work anyway, since it contains reducible e representations which are not supported by mpshift (energy, gradient, etc. do all work though). But it usually prints out an error message and recommends to retry the calculation with lower symmetry.

If it also fails in C1, there must be a different problem. Please send the input to the Turbomole Support. If the error is still present in the newest Turbomole release, the support can hopefully debug and fix it.

Regards,

Uwe