Author Topic: Unexpected RI-MP2 single-point energy after optimization  (Read 11960 times)

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Unexpected RI-MP2 single-point energy after optimization
« on: December 03, 2007, 05:34:55 PM »
Hi all,

In the course of some benchmarking, I have faced with an unexpected and non-trivial issue when re-calculating the RI-MP2 energy of an optimized (at the same RI-MP2 level !) structure. After carring out a typical optimization with jobex -statpt -level mp2 -ri , I took the last (and optimized) structure with t2x -c and prepared a single-point calculation input at (exactly) the same level using define (for the dscf) and rimp2prep (for the rimp2). Surprisingly, the dscf energy of this single-point calculation was quite different from that obtained in the last step of the optimization (-2439.3820282307 versus -2439.546401018 hartrees). I guess that I must be doing something wrong, but I really do not know what it could be...Any solution, please?

Thank you very much in advance

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Re: Unexpected RI-MP2 single-point energy after optimization
« Reply #1 on: December 04, 2007, 10:10:32 AM »
Hi,

I can give some more info about the aforementioned inconsistency. After some probes, I decided to substitute the mos file (the molecular orbitals) freshly generated by eht in define by the one obtained at the end of the optimization. This trick allowed to calculate the same single-point energies (both SCF and MP2) previously obtained in the optimization. Well, I realize that this is really tricky because one usually wants to calculate single-point energies a a higher level but without optimizing the structure (at that given level) first. Anyway, I will try to go further...

Bye,

Gonzalo

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: Unexpected RI-MP2 single-point energy after optimization
« Reply #2 on: December 04, 2007, 11:47:08 AM »
Hi,

this sound as if you have for this molecule two (close lying) solutions for the SCF equations... In UHF calculations this might be due to different numbers of half-occupied orbitals. In a closed-shell calculation this usually only happens if you spatial symmetrie and the orbital occupation obtained from the EHT guess doesn't lead to the correct SCF ground state, or the order the orbitals at the HOMO/LUMO gap changed between the start and end points of you geometry optimization.

Regards,
               Christof

PS: you don't need to use rimp2prep to set up the control file for rimp2. you can do everyting within the mp2 menu of define (which is the recommended way to do it).

gojimeo

  • Newbie
  • *
  • Posts: 9
  • Karma: +0/-0
Re: Unexpected RI-MP2 single-point energy after optimization
« Reply #3 on: December 05, 2007, 01:36:33 PM »
Hi Christof

OK, thank you very much for your valuable comments. I will try to generate a better initial guess for the mos.

Regards,

Gonzalo