Author Topic: RI-SCS-MP2 optimization  (Read 13003 times)

kabel999

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
RI-SCS-MP2 optimization
« on: October 14, 2008, 02:45:33 PM »
Hello ,
I am newbie in Turbomole in 5.10 and I tried to optimize the molecule using RI-SCS-MP2 method. I prepared the input in old way using rimp2prep, and I added following lines into the control file
$ricc2
 scs
 geoopt=mp2

 However, the results seems to be identical as RI-MP2 ones (where I used mp2 instead scs in control file). The jobex script were running with -level ricc2  option.
 What I am doing wrong?
Thanks for your help!
Martin


Arnim

  • Developers
  • Sr. Member
  • *
  • Posts: 253
  • Karma: +0/-0
Re: RI-SCS-MP2 optimization
« Reply #1 on: October 15, 2008, 09:43:25 AM »
Hi,

if SCS is switch on, you should find the lines:
     Spin-Component Scaling will be applied...
          scaling factors:  C_os = 1.2  C_ss = 0.3
in the output.

Cheers,
Arnim

Stefan

  • Jr. Member
  • **
  • Posts: 15
  • Karma: +0/-0
Re: RI-SCS-MP2 optimization
« Reply #2 on: November 07, 2008, 01:37:43 PM »
Hello,

Your input must look like:
$ricc2
 scs
 geoopt model=mp2

And you must start your calculation with:
jobex -level cc2 -ri ...

Best regards, Stefan

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: RI-SCS-MP2 optimization
« Reply #3 on: January 11, 2009, 12:35:24 PM »
Dear Stefan,

jobex -level cc2 is fine. The "-ri" flag is ignored by all TM scripts if used with ricc2 since there is no CC2 code in TM which
could do the calculations with RI.
Actually, also "geoopt=mp2" should work (but you're right: the recommended input is "geoopt model=cc2").

kabel999: check in the output, whether the SCS scaling was applied and gradients were calculated. If yes
maybe the SCS-MP2 structure is so close to the MP2 structure  that they you don't see a significant difference with
the default convergence thresholds?

Christof