Molcas Forum

ChiTienHsieh

Member

Registered: 2022-03-29

Posts: 2

Localisation problem (Partition in fragments)

Dear Molcas community,

I encountered a problem when trying to localize the active space orbitals, my example is from this paper, J. Phys. Chem. B 2008, 112, 45, 14099–14102. In Figure 2., the author localized the orbitals into the metal center and ligand using MOLCAS-7. I somehow cannot achieve this kind of localization with OpenMolcas’ &LOCALISATION program. Instead of partitioning into Fe/ligand, I would have Fe, pi-bonding, and sigma-bonding quite localized to atom/bond. I have tried Pipek–Mezey, Boys, Cholesky and PAO methods, but none of them generates the orbitals I want.

In a nutshell, I would like to know how to localize orbitals into 2 fragments, instead of localizing orbitals into atoms/bonds.

" />

The image link is appended, in case the link is not working.
https://imgur.com/PIZcUNh

My &LOCALISATION input is also appended.

&LOCALISATION
  NORBITALS
  12
  NFROZEN
  91

Last edited by ChiTienHsieh (2022-03-30 06:36:20)

ChiTienHsieh

Member

Registered: 2022-03-29

Posts: 2

Re: Localisation problem (Partition in fragments)

This problems comes into 2 parts.
First, the symmetry (Cs) needed to be applied, this way the orbitals all look good and symmetrized.
Second, I used to localize all 12 orbitals (orbitals in the Active Space), but it's not necessary. If I only localize four orbitals like the picture (and they are partially occupied, that's why they are chosen), I can get the orbitals I want. Since the final CAS energy remains the same, it indicates no matter how I choose the orbitals in active space to localize, the over all wavefunction is invariant.