[CP2K-user] [CP2K:19956] NON_LOCAL dispersion in ENERGY_CORRECTION ?

[Jürg Hutter]

Hi

Yes, the procedure you described is available in CP2K as the
ENERGY_FUNCTION DCDFT
option in the ENERGY_CORRECTION method. If the nl vdW is only
in the energy correction functional, only first derivatives are needed.
Second derivatives would be needed only for the option
ENERGY_FUNCTION HARRIS.

Unfortunately, both options will need the update of the XC functional
call in the energy correction routines. This is no big deal, but still needs a
couple of days work (including testing).

regards
JH

________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Ralf Frischmann <ralf.p.frischmann at gmail.com>
Sent: Thursday, February 22, 2024 3:44 PM
To: cp2k
Subject: Re: [CP2K:19956] NON_LOCAL dispersion in ENERGY_CORRECTION ?

Hi JH,

thanks for the quick reply and the clarification. Wasn't 100% sure whether my assumption above were actually correct.

Another possible option for getting NL dispersion in these cases could be to apply the dispersion NL functional in a post-SCF-type fashion (quantum chemistry codes ORCA and psi4 allow this as an option). Then it should be possible to take RHO from a converged HFX-SCF procedure and "stuff it" into the existing NL energy+gradient routines prior to the DFT response calculation, so with that no response calculation should be needed for the NL dispersion part.

Having the post-SCF NL option in general would also be a nice and "fairly easy" to do CP2K feature. Relative errors introduced by that are usually very small (~0.0x kcal/mol) in gas phase calculations, and substantial computational savings might result from it.

Regards,

Ralf






On Thursday, February 22, 2024 at 11:12:00 AM UTC+1 Jürg Hutter wrote:
Hi

thank you for raising this issue. I have opened a new feature request issue on
github (#3277) and have disabled nl-vdW functionals in EC calculation in the
current cp2k version.
The main problem is the required 2nd derivative of the functionals for
the gradient calculation.

regards
JH

________________________________________
From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of Ralf Frischmann <ralf.p.f... at gmail.com>
Sent: Thursday, February 22, 2024 5:57 AM
To: cp2k
Subject: [CP2K:19950] NON_LOCAL dispersion in ENERGY_CORRECTION ?

Dear all,

are there plans already to implement a working NON_LOCAL dispersion correction for the &ENERGY_CORRECTION part?

It appears to me that currently only the PAIR_POTENTIAL type delivers a finite Edisp value in the CP2K output file, although input files with a NON_LOCAL &VDW_POTENTIAL specifications get accepted with no further warning from the code, but do return Edisp = 0.d0 in the EC part of the output.

An inspections of the SUBROUTINE ec_disp(qs_env, ec_env, calculate_forces) in the file energy_corrections.F with a hard-wired call to the function calculate_dispersion_pairpot() further supports that assumption.

Thanks in advance for any clarification on that subject!

Regards,

Ralf





--
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... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/327a60fd-8cde-47cf-9144-be6a910ceb04n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/327a60fd-8cde-47cf-9144-be6a910ceb04n%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
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+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/9a7afc94-7e9b-4130-bb24-0a0c7aa62979n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/9a7afc94-7e9b-4130-bb24-0a0c7aa62979n%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
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+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB07593B63F069A93AFAD9676E9F562%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.