Fixed band occupations during SCF

18 views
Skip to first unread message

Nicholas Winner

unread,
Nov 23, 2021, 4:43:07 PM11/23/21
to cp2k
I am trying to mimic a technique shown in the literature with other codes.

Given bands in VB and CB, and an in-gap state with N electrons, an ionized state can be simulated without the need for a charged supercell by taking some n < N electrons from the in gap state and moving them to LUMO. A similar procedure can be performed for VB and holes.

I'd like to be able to do this in CP2K so I can try this using OT+ADMM of hybrid calculations. This is quite straight forward in many codes, but CP2K may have some difficulties with it according to what I've seen on here.

There was another conversation a few years ago regarding this: https://groups.google.com/g/cp2k/c/WV2VqCd1z9Q/m/JX7OgUmQCgAJ

But I am unable to follow some of the discussion, and the last message refers to some FORTRAN code that I assume is supposed to be inserted into CP2K source somewhere.

Could anyone shed some light on this?

Thanks
Nick

Krack Matthias (PSI)

unread,
Nov 24, 2021, 11:33:54 AM11/24/21
to cp...@googlegroups.com

Hi Nick

 

The code looks like a stand-alone program reading a CP2K WFN restart file and writing the information in the same format to a new file. In between the code manipulates the orbital occupations of HOMO and LUMO, i.e. these values are set to 0 and 1, respectively. Check in read/write subroutines in https://github.com/cp2k/cp2k/blob/master/src/qs_mo_io.F for details. Most of the code might have been copy-pasted from there. I guess that the re-written file is then used to restart a CP2K run with a modified orbital occupation.

 

HTH

 

Matthias

__________________________________________
Paul Scherrer Institut
Dr. Matthias Krack
OHSA/D07
Forschungsstrasse 111
5232 Villigen PSI
Switzerland

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/63fb747b-1bee-4ea0-9a56-5186d8e98a70n%40googlegroups.com.

Reply all
Reply to author
Forward
0 new messages