[molpro-user] Reference weight value for CASPT2 calculations.

[Hans-Joachim Werner]

Dear Dawid,

RS3 computes the third-order energy using the first-order (CASPT2) wave function. The CASPT2 calculations are done in intermediate normalization, and the squared norm of the total wave function is seen under  SQ.NORM (first column) in the print of the iterations. Thus, < Ref|Ref>=1/sqnorm, and you can take sqrt(1/sqnorm) as
the coefficient of the reference function |Ref> in the first-order wave function. At the end of a multi-state calculation, the program solves the SG equation HC=SCE in the basis of the first-order wave functions (so for 3 states it's a 3x3 eigenvalue problem). This yields a transformation between the states, similar to MS-CASPT2 (but at a higher level). The transformed states are orthogonal. The caspt2 and caspt3 energies for the transformed wave functions are the „ROTATED“ values. The dipole and transition moments are computed as well and printed for the original as well as for the rotated wave functions. Note that the printed dipole moment expectation values and transition moments are always computed using normalized wave functions.

The weights of the references <Ref(ROT)|RefF(ROT)> in the rotated wavefunctions are printed after "Norm of relaxed reference (rot)“. 

I had tried the calculation you had sent earlier, but the third state did not converge.

Best regards
Joachim

---
Prof. Dr. Hans-Joachim Werner
Institute for Theoretical Chemistry
University of Stuttgart
Pfaffenwaldring 55
D-70569 Stuttgart
Tel: +49 711 / 685 64400
Fax: +49 711 / 685 64442
email: werner at theochem.uni-stuttgart.de


> Am 11.12.2015 um 16:21 schrieb Dawid das <addiw7 at googlemail.com>:
> 
> Dear Molpro Experts,
> 
> I have performed CASPT3 calculations with CASSCF wavefunction as a reference.
> I used following input:
> 
> ***,PSB3
> memory, 450, M
> SET, CHARGE=1
> gexpec, dm
> gdirect
> geomtyp=xyz
> symmetry,nosym
> geometry={
> 14
> PSB3
> C         -0.0002566337        0.0012806335       -0.0045289738
> C         -0.0030717618       -0.0004852300        1.3489960614
> C          1.2442813204       -0.0012856474        2.0756133424
> C          1.4475005421       -0.0039130467        3.4356099462
> C          0.3842489326       -0.0058656017        4.3728298332
> N          0.5682361422       -0.0095509528        5.6709166926
> H          2.4654050705       -0.0048405031        3.8177998498
> H          2.1426826325       -0.0002301120        1.4591434928
> H         -0.6530291517       -0.0042819065        4.0474832878
> H         -0.9514175607       -0.0016008345        1.8796761931
> H         -0.2127208483       -0.0110877687        6.3195543342
> H          1.5004228539       -0.0110165037        6.0776308342
> H         -0.9240107645        0.0015068704       -0.5729784019
> H          0.9255382266        0.0022226034       -0.5733504920
> }
> basis=roos(3s2p1d/2s1p)
> rhf
> put, molden, PSB3.scf.molden;
> {multi
>   rotate, 19.1, 20.1, 0
>   rotate, 24.1, 30.1, 0
>   occ, 25, 0
>   closed, 19, 0
>   wf, 44, 1, 0
>   state, 3
> }
> put, molden, PSB3.rasscf.molden;
> {rs3, shift=0.1, ipea=0.25
>   state, 3
> }
> 
> Now I would like to know what was the coefficient of the reference wavefunction (CASSCF)
> when I calculated states 1, 2 and 3 with CASPT3 method. I cannot find it in my 
> *log file.
> In other words, how much of the reference has contrivution to the CASPT3 energy.
> 
> Best wishes,
> Dawid Grabarek
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user