[CP2K-user] [CP2K:17984] Vibrational analysis best practices

Xavier Bidault jazzquark at gmail.com
Fri Nov 4 18:05:11 UTC 2022


Hi Eric,

I would recommend this paper: https://doi.org/10.1063/5.0041511
where we used CP2K/PHONOPY for phonon calculations of molecular crystals.
We used a large cutoff of 1500 Ry for CP2K, but you can use 600 Ry, which
is already tightly converged. But DO NOT use lower than 600 Ry for CHON
systems. You can do your own convergence tests, of course.
REL_CUTOFF is not that important (60 Ry or default is good).
But NGRIDS 5 is important! Faster calculations.
The driving convergence parameter is MAX_FORCE 1E-5. You can keep the
others at 1E-3.
Yes, EPS_SCF 1E-8.
Yes, NLCC.
I think SR is not bad, but do some comparison (optimization and phonon
calculations) with non-SR to see if SR is accurate enough.
PBE-D2 or PBE-D3(BJ)? Look at my concerns in the paper. We ended up with D2
with the pseudo potentials, but D3(BJ) with all-electron calculations.
GAPW is for all-electron calculations.

Best,
Xavier

On Fri, Nov 4, 2022 at 7:29 AM Eric Patterson <eric.v.patterson at gmail.com>
wrote:

> Dear all,
>
> I am soon to embark on a series of optimizations and harmonic frequency
> calculations for a set of molecular solids (mostly C, H, N, O organics).  I
> have been combing this forum for advice, and would like to check that the
> following are general best practices for these calculations.  Any advice
> and corrections greatly appreciated!
>
> I plan to use PBE-D3(BJ)/DZVP-MOLOPT-SR-GTH.
>
> I believe I understand how to select CUTOFF and REL_CUTOFF, and plan to
> use EPS_SCF 1E-8.
>
> It seems best to use NLCC pseudopotentials with GPW, correct?
> Also use smoothing?  If so, which is recommended?
> Use USE_FINER_GRID, or not?
>
> If using GAPW, still use NLCC?
> Same questions regarding smoothing and USE_FINER_GRID.
>
> I have read that extremely tight convergence on the geometry is
> recommended.  Do the following seem appropriate?
>
>     MAX_FORCE 4.5E-6
>     RMS_FORCE 3.0E-6
>     MAX_DR 3.0E-5
>     RMS_DR 1.5E-5
>
> Anything else experienced users might wish to share?  I appreciate any
> advice as it could save me a ton of time benchmarking various options.
>
> Thank you,
> Eric
>
> --
> 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/d527e09a-28eb-41d3-b751-2e3ae555b0e3n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/d527e09a-28eb-41d3-b751-2e3ae555b0e3n%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/CAD0N%2BNVDhq8WCzP9qg418ZSGRFiJwiNsrHrRC8_6gr%3DW67Cz2Q%40mail.gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221104/dab6f229/attachment.htm>


More information about the CP2K-user mailing list