Molcas Forum

Yafu Guan

Member

Registered: 2020-11-26

Posts: 2

antisymmetric derivative overlap

Dear Molcas users and developers,

Molcas or OpenMolcas seem to be able to provide the antisymmetric derivative overlaps or half-derivative AO overlap integrals
(see Eq. 22 of https://doi.org/10.1021/acs.jctc.6b00384), which are necessary to obtain the CSF part of derivative couping.
I'm wondering how to invoke the calculations of them in Molcas and how to access them through Molcas interface.

Thank you.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,011

Re: antisymmetric derivative overlap

The overlap derivatives are not stored or available for printout. They are computed and contracted on the fly with the density matrices to compute the gradients. If you compute the non-adiabatic coupling with the NAC keyword, you'll get f(CSF), but not S.

Yafu Guan

Member

Registered: 2020-11-26

Posts: 2

Re: antisymmetric derivative overlap

Thank you so much, Ignacio.

felixp

Member

Registered: 2015-11-06

Posts: 27

Website

Re: antisymmetric derivative overlap

Hi Ignacio, thanks for getting back!

The idea is that Yafu wanted to use the interface to Columbus to compute the f(CSF) from Columbus. So, if we feed the density matrix from Columbus into alaska, then it should be possible to get f(CSF), I guess. Similarly to how we get f(CI). Right?

Cheers
Felix

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,011

Re: antisymmetric derivative overlap

Yes, it should be possible, but as a programmer/hacker, not at a user (at least for the moment). If you store the right matrices and flags in the RunFile, ALASKA doesn't need to know where the density matrices come from.