[CP2K-user] [CP2K:22220] Re: Wavelet Poisson solver - Huge change in total energy when increasing the plane-wave cut-off energy from 1200 to 1260 Ry

[Mattia Perrone]

Hi all 

I can confirm that also in my case (different system and different  setup) 
I observe the same unexpected behavior, when the FFT grid size becomes 
larger than 640.

I wanted to ask if a solution to the problem had been found in the meantime

Thanks!

mattia


On Monday, December 29, 2025 at 10:50:33 AM UTC+1 Jürg Hutter wrote:

> Hi
>
> yes, I can reproduce your problem using a small molecule in a 30x30x30 box 
> with
> cutoffs 1200 and 1260.
>
> These settings result in internal FFT grids of size 625 (works) and 648 
> (fails).
> Some tests show that up to 640 (cutoff 1230) it works, everything larger 
> fails.
>
> This needs a deeper look into the wavelet poisson solver. Maybe we should 
> open
> a bug report at the CP2K page on Github.
>
> regards
> JH
>
> ________________________________________
> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of 
> Torstein Fjermestad <tfjer... at gmail.com>
> Sent: Monday, December 29, 2025 10:18 AM
> To: cp2k
> Subject: [CP2K:22012] Re: Wavelet Poisson solver - Huge change in total 
> energy when increasing the plane-wave cut-off energy from 1200 to 1260 Ry
>
> Dear Frederick,
>
> Thanks for the piece of advice.
> I generated the input file by setting the name of the variable and its 
> value like this:
> @SET <name> <value>
>
> In the QS section, I explicitly set the following variables:
> @SET eps_def 1.0E-10
> @SET extrapol PS
> @SET precOM 1.0E-5
> &QS
> EPS_DEFAULT ${eps_def}
> EXTRAPOLATION ${extrapol}
> EPS_PGF_ORB ${precOM}
> EXTRAPOLATION_ORDER 3
> &END QS
>
> I followed your advice, and I decreased the value of EPS_DEFAULT to 
> 1.0E-12 and the value of EPS_PGF_ORB to 1.0E-6.
> This did not have any notable effect on the results.
> (For the calculation with a cut-off of 1260 Ry, the energy after the first 
> SCF iteration was -4394.5092346069 <(509)%20234-6069> Hartree with the 
> new settings and -4394.5090808424 Hartree with the previous settings. 
> However, these values are already a few thousand Hartree below the 
> converged value of the calculation with a cut-off of 1200 Ry 
> (-1389.249071606565622 Hartree))
>
> Should I decrease EPS_DEFAULT and EPS_PGF_ORB further, or are other 
> variables I should change?
>
> Thanks.
>
> Best regards,
> Torstein
>
>
>
> tirsdag 23. desember 2025 kl. 11:54:48 UTC+1 skrev Frederick Stein:
> Dear Torstein,
> You not only need to adjust the CUTOFF but also EPS_DEFAULT (and if set 
> explicitly the other thresholds) in the QS section. How did you obtain the 
> input file? I would never set all parameters. Especially EPS_DEFAULT sets a 
> variety of other parameters if these are not set explicitly.
> Best,
> Frederick
>
> Torstein Fjermestad schrieb am Montag, 22. Dezember 2025 um 20:09:02 UTC+1:
> Dear all,
>
> As part of a basis set convergence test, I gradually increase the plane 
> wave basis set cut-off energy. The atomic system is a cluster model of 
> methanol adsorbed in a zeolite. I isolate the system from its neighboring 
> unit cells by using the wavelet Poisson solver.
> I use a cubic box of 30 Å. The size of the system is about 21 Å in the 
> longest direction. The atom-centered basis set is TZV2P.
> When I increase the cut-off from 900 Ry to 1200 Ry, the total energy 
> varies in a range less than 5 kJ/mol. However, when I increase the cut-off 
> from 1200 Ry to 1260 Ry, the total energy is after 6 SCF iterations 
> 1.38E+07 kJ/mol lower!
> Because of the massively different total energy, I stopped the calculation.
>
> I have attached the output files of the two calculations. The complete 
> input can be found at the start of the output files.
>
> What could be the cause of this behavior?
>
> Thank you very much for your help.
> Please tell me if you need more information to understand the problem 
> better.
>
> Best regards,
> Torstein Fjermestad
>
>
> --
> 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 visit 
> https://groups.google.com/d/msgid/cp2k/9cbaa759-1d2e-41c9-a4ad-5e376f6398f3n%40googlegroups.com
> <
> https://groups.google.com/d/msgid/cp2k/9cbaa759-1d2e-41c9-a4ad-5e376f6398f3n%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 visit https://groups.google.com/d/msgid/cp2k/7f19f214-6499-4ad7-9fc3-42456834dd58n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260505/ec1943f5/attachment.htm>