Molcas Forum

daishudaishu

Member

Registered: 2019-02-24

Posts: 22

What Natural orbital is used in CASPT2 output?

Dear all,
I am calculating SA-CASSCF+MS-CASPT2. I would like to know the character of the MS-CASPT2 roots. I find the following output. Do the u,d and 2 represent the occupation of SA-CASSCF natural orbitals, which could be visualized with GRID_IT and Luscus?
Thanks a lot!

  THE ORIGINAL CI ARRAYS ARE NOW MIXED AS LINEAR
  COMBINATIONS, GIVEN BY THE EIGENVECTORS.
  A NEW JOBIPH FILE NAMED 'JOBMIX' IS PREPARED.
********************************************************************************

++ Mixed CI coefficients:
  The CI coefficients for the MIXED state nr.   1
--------------------------------------------------------------------------------
 CI COEFFICIENTS LARGER THAN  0.50D-01
  Occupation of active orbitals, and spin coupling
  of open shells. (u,d: Spin up or down).
  SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
        Conf  SGUGA info        Occupation         Coefficient         Weight
          87  ( 2:1:   1/  11)   u22222222222d0       0.804559       0.647316
          88  ( 2:1:   2/  11)   u2222222222d20       0.108204       0.011708
          90  ( 2:1:   4/  11)   u2222222222dud       0.060648       0.003678
          91  ( 2:1:   5/  11)   u2222222222udd      -0.151888       0.023070
          92  ( 2:1:   6/  11)   u222222222202d      -0.070944       0.005033
          94  ( 2:1:   8/  11)   u22222222220d2       0.105945       0.011224
         155  ( 3:1:   1/  11)   u222222222d220       0.504303       0.254322
         156  ( 3:1:   2/  11)   u222222222d2ud      -0.073604       0.005418
         188  ( 4:1:   1/  10)   u222222222u2dd      -0.073096       0.005343
         190  ( 4:1:   3/  10)   u222222222udd2       0.058539       0.003427
         218  ( 5:1:   1/  10)   u222222222022d       0.090245       0.008144
         219  ( 5:1:   2/  10)   u22222222202d2       0.109157       0.011915

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: What Natural orbital is used in CASPT2 output?

If I'm not mistaken, those should be the same orbitals as the RASSCF: average pseudo-natural orbitals. The mixed CI vectors should be just a linear combination (according to the MS eigenvectors) of the input CI vectors.

daishudaishu

Member

Registered: 2019-02-24

Posts: 22

Re: What Natural orbital is used in CASPT2 output?

Dear Ignacio,
Thanks so much for your reply!

I have to identify the orbitals that contribute to certain MS-CASPT2 roots. Now I am trying to visualize the CASPT2 orbitals with the following script.

&GATEWAY
  coord  =  geom.xyz
  basis  = 6-31g*;  group=  E
RICD
&GRID_IT
FILEORB=input.Pt2Orb.1
all

I used ALTER and SUPSYM to calculate XAS with the core-excited state. But the output *.lus file doesn't show me any core orbitals. All of them look like valence orbitals.

However, the *lus file generated from the RASSCF orbital (*.RasOrb) does show core orbitals.

Do you mean that I could use the RASSCF natural orbitals to analyze the MS-CASPT2 roots?

Thanks a lot!

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: What Natural orbital is used in CASPT2 output?

The Pt2Orb files should contain the PT2 natural orbitals, i.e., they contain the perturbation correction. I'm not sure they contain the MS mixing. Are they different when you include the NOMULT keyword? The "Mixed CI coefficients" table, on the other hand, contains the mixing, but not the PT2 correction.

The Pt2Orb files should contain the core orbitals, even if they are fully occupied... If they don't, it might be a bug in GRID_IT or Luscus. Note that if you have core orbitals in the active space, you may want to change the number of frozen orbitals in CASPT2.