Molcas Forum

Anugraha

Member

Registered: 2025-12-17

Posts: 22

Active orbitals rotated out of active space in CASPT2 workflow

Dear Molcas Developers,
I am running a CASPT2 calculation on the ground state of my molecule.

 & SEWARD 
 & RASSCF 
  Fileorb = project.ScfOrb
  symmetry =1
  Iterations = 200 100
  Spin = 1
  nActEl = 8 0 0
  Inactive = 25 1
  Ras2 = 0 7
  ciroot = 1 1 1 
 & CASPT2
  Multistate = 1 1 
 &RASSI
  Nr of Job=1 1 ; 1
  EJob
 & Grid_it 
 All

Initially, I performed an SCF calculation to inspect the molecular orbitals and determine a suitable active space. I modified the SCF orbital file and selected the following orbitals for my active space: orbitals 2, 3, 4, 5, 7, 8, and 9 in symmetry 2.

However, after the CASPT2 calculation, I found that one of the active orbitals(number2) no longer has the character I originally selected from the SCF. In addition, the original active orbital now appears in the inactive space as number 1 in symmetry 2.

Could you please advise how I can ensure that the active orbitals I selected before the calculation remain in the active space throughout the CASSCF/CASPT2 procedure?

Thank You
Anugraha

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,204

Re: Active orbitals rotated out of active space in CASPT2 workflow

If the orbital has occupation very close to 2, it's probably just lower in energy with the other orbital. You could include more roots in the state average, if you reach the state where electrons are excited from your wanted orbital, its average occupation will decrease and it will "want" to stay in the active space.

Anugraha

Member

Registered: 2025-12-17

Posts: 22

Re: Active orbitals rotated out of active space in CASPT2 workflow

If the orbital has occupation very close to 2, it's probably just lower in energy with the other orbital. You could include more roots in the state average, if you reach the state where electrons are excited from your wanted orbital, its average occupation will decrease and it will "want" to stay in the active space.

Sir, I included orbitals numbered 1 and 2 in symmetry 2 in the active space in addition to the other selected active orbitals because in a previous CASPT2 calculation without orbital 1, the orbital that I originally selected as number 2 lost its character and was replaced by orbital 1 in the active space.

& SEWARD
& RASSCF
Fileorb = project.ScfOrb
symmetry =1
Iterations = 200 100
Spin = 1
nActEl = 10 0 0
Inactive = 25 0
Ras2 = 0 8
ciroot = 1 1 1
& CASPT2
Multistate
1 1
& Grid_it
All

Here is the natural occupation number after caspt2

Natural orbitals and occupation numbers for root  1
      sym 2:   1.999245   2.000000   1.961959   1.904359   1.906207   0.035683   0.098592   0.093955

As you can see, two of the active orbitals have occupation numbers very close to 2. Does this indicate that my active space may be inappropriate? Should I include additional roots in the state-averaged CASSCF instead of adding orbital 1 to the active space?