Molcas Forum

jxzou

Member

Registered: 2020-09-11

Posts: 12

Zero occupation numbers found in (X)MS-CASPT2 SO states

Dear (Open)Molcas developers,
I've performed a XMS-CASPT2 calculation based on a set of SA-CASSCF(6e,4o)/ANO-RCC-VTZP converged MOs. State interaction is also considered using the &RASSI module. But I find that the natural orbital occupation numbers of an SO (spin-orbit) state becomes all zero for state 8~10. If this is reasonable, would you please briefly explain the reason or theories involved? My input file is

&GATEWAY
Basis set
...
noCD
AMFI

&SEWARD
Relativistic= R02O

&RASSCF
CIMX= 200
Tight= 5d-8 5d-6
Charge= 0
Spin= 1
nActEl= 6
RAS2= 4
CIroot= 4 6 1
CIonly
FILEORB= xxx_ANO-RCC-VTZP_XMS.INPORB

&CASPT2
MAXITER= 200
XMultiState
4 1 2 3 4

>> COPY $Project.JobMix $Project.S0.JobMix

&RASSCF
CIMX= 200
Tight= 5d-8 5d-6
Charge= 0
Spin= 3
nActEl= 6
RAS2= 4
CIroot= 3 5 1
CIonly
FILEORB= xxx_ANO-RCC-VTZP_XMS.INPORB

&CASPT2
MAXITER= 200
XMultiState
3 1 2 3

>> COPY $Project.JobMix $Project.T0.JobMix
>> COPY $Project.S0.JobMix JOB001
>> COPY $Project.T0.JobMix JOB002

&RASSI
TRDI
EJOB
SpinOrbit
NatOrb
10

I try to calculate the lowest 4 singlet states as well as 3 triplet states, and generate SO states using these pure states.

The natural orbital occupation numbers of state 7 (in xxx_ANO-RCC-VTZP_XMS.SiOrb.7) are

#OCHR
* OCCUPATION NUMBERS (HUMAN-READABLE)
  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000
  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000
  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000
  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  2.0000  1.0000  1.0000
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
...

which I think is normal.

But the natural orbital occupation numbers of state 8 (in xxx_ANO-RCC-VTZP_XMS.SiOrb.8) are

#OCHR
* OCCUPATION NUMBERS (HUMAN-READABLE)
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000  0.0000
...

This problem can also be found when using MS-CASPT2. Also, if you find anything in my input file which can be improved, please do not hesitate to tell me.

Thanks!
Best wishes,
Jingxiang

Last edited by jxzou (2025-09-12 03:31:10)

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,165

Re: Zero occupation numbers found in (X)MS-CASPT2 SO states

There are only 7 spin-free states. I don't know why it's creating more files than states, or why it's not crashing...

If you want the orbitals for SO states, there's the (undocumented) SONOrb keyword:

SONOrb
10
1
2
3
4
5
6
7
8
9
10

jxzou

Member

Registered: 2020-09-11

Posts: 12

Re: Zero occupation numbers found in (X)MS-CASPT2 SO states

Thank you, Prof. Ignaco. I'll try `SONOrb`. As for the number of orbital files, I have

_ANO-RCC-VTZP_XMS.RasOrb
_ANO-RCC-VTZP_XMS.RasOrb.1
_ANO-RCC-VTZP_XMS.RasOrb.2
...
_ANO-RCC-VTZP_XMS.RasOrb.6
_ANO-RCC-VTZP_XMS.SiOrb.1
_ANO-RCC-VTZP_XMS.SiOrb.2
...
_ANO-RCC-VTZP_XMS.SiOrb.10

If my understanding is correct, this is reasonable because I calculated 6 singlet roots and 5 triplet roots for CASSCF. The final spin-pure state orbital file should be .RasOrb.6. And I calculated 4 singlet roots and 3 triplet roots for XMS-CASPT2 and subsequent state interaction, which leads to 4+3*3=13 SO states. So the final orbital file for SO states should be .SiOrb.13. Since I specify only 10 SO states, the final orbital file turns out to be .SiOrb.10. Am I right?

Oh, maybe you mean that `NatOrb` is used to calculated for spin-free states, not for SO states, so the number of `.SiOrb.n` files is supposed to be 7. Once I use `SONOrb`, the number of `.SONOrb.n` files is supposed to be 10 or 13.

Last edited by jxzou (2025-09-12 08:36:12)

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,165

Re: Zero occupation numbers found in (X)MS-CASPT2 SO states

You computed only 4 singlet and 3 triplet states with CASPT2, which is what you are using for RASSI. NatOrb (*SiOrb.*, *SsOrb.*) is only for spin-free states, so it will be limited to 7 states. SONOrb  (*SOOrb.*, *SOSOrb.*) will give you 13 SO-coupled states.

Even if you had used the CASSCF wave functions, I think the RASSI will be limited to the roots used in the state average.

jxzou

Member

Registered: 2020-09-11

Posts: 12

Re: Zero occupation numbers found in (X)MS-CASPT2 SO states

Your answer does a lot of help. I misunderstood `NatOrb`. Just to make sure, if I modify the final few lines into

&RASSI
TRDI
SONT
9
1 2
1 3
1 4
1 5
1 6
1 7
1 8
1 9
1 10
ARGU
EJOB
SpinOrbit

These keywords tell OpenMolcas to calculate SO-NTOs between the SO-coupled state 1 and SO-coupled states 2~10. Spin-free states are not involved here. Am I right?

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,165

Re: Zero occupation numbers found in (X)MS-CASPT2 SO states

Yes, that should do it.