Molcas Forum

jxzou

Member

Registered: 2020-09-11

Posts: 14

Question and trick about calculating molecular polarizability

Dear (Open)Molcas developers,

I'm trying to calculate the CASSCF molecular polarizability using OpenMolcas. I have a few questions and would appreciate your advice. As far as I know, currently there are three approaches to calculate the CASSCF polarizability:

(1) Use `&MCKINLEY`, perform the harmonic frequency calculation.
The molecular polarizability will be printed as byproduct. It seems that some density-matrix perturbation theory trick is applied when calculating the frequency, because I see

     Linear response function is computed for root no. =   1

 Process perturbation number                      1
      Perturbation no:    1 converged in   36 steps.

 Process perturbation number                      2
      Perturbation no:    2 converged in   36 steps.
...

in OpenMolcas output. I do not know whether it can be called "calculate the molecular polarizability in an analytic way" in such case.

(2) Use `&LOPROP`.
The molecular polarizability will be calculated by automatically calling FFPT. That is to say, the molecular polarizability is calculated using numerical differentiation.

(3) Use `&FFPT`, which is to manually perform six CASSCF single-point calculations using different (and small) electric fields, and perform numerical differentiation to the obtained dipole moments at different field strengths.

My question is:
(i) For three approaches above, is the polarizability all calculated numerically, or is it calculated analytically in approach (1)?
(ii) Is there any more efficient way or cost-effective way to calculate the polarizability? For example, we obtain many quantities from a frequency calculation, but sometimes I only want the polarizability.

Thanks a lot.

Best wishes,
Jingxiang

Last edited by jxzou (2026-06-11 17:33:26)