[CP2K-user] [CP2K:21813] Phonon Calculations in CP2K

[Ashley Dickson]

Hi Matthias,

thanks for the response. For your information I am using the finite 
difference method here. Supposing this was the issue, how would you 
recommend solving it? Presumably I can't explicitly freeze the orbital 
occupations for the displaced configurations? Perhaps including the 
wavefunction of the initial geometry as an initial guess might keep the 
correct spin state.

Also, I wonder if the anisotropic nature of YBCO is the issue here. It is 
not strictly magnetic, although the copper chains/planes may act as such 
and mess up the forces. Would it be beneficial to include spin in my 
calculations maybe? 

Many thanks 
Ashley 
On Tuesday, 9 September 2025 at 14:44:39 UTC+1 Krack, Matthias wrote:

> Hi Ashley
>
>  
>
> Regarding the CP2K side, you should check if all the finite difference 
> force calculations converged properly to equivalent (magnetic?) states 
> during an atomic displacement. You do not report the applied computational 
> method and I can only guess, but I would have a look at the Cu 3d 
> occupations of the YBCO before and after an atomic displacement.
>
>  
>
> Best
>
>  
>
> Matthias
>
>  
>
> *From: *cp... at googlegroups.com <cp... at googlegroups.com> on behalf of 
> Ashley Dickson <ashleydi... at gmail.com>
> *Date: *Tuesday, 9 September 2025 at 12:39
> *To: *cp2k <cp... at googlegroups.com>
> *Subject: *[CP2K:21810] Phonon Calculations in CP2K
>
> Hello everyone, 
>
>  
>
> I'm unsure if this is the right place to post this issue as I am using 
> phonopy to generate phonons with CP2K, however I thought I might have a bit 
> more luck finding someone with experience using phonopy specifically with 
> CP2K. The issue I'm having is large negative phonon modes, and large drift 
> force reported by phonopy in the y direction (~0.25 eV/A). I know that I 
> shouldn't be getting negative modes as my structure is certainly 
> dynamically stable. my workflow is as follows:
>
>  
>
> phonopy --cp2k -d --dim="6 6 2" -c energy.inp 
>
> for i in energy-supercell-*; do
>     bn=${i%.*}  # Remove file extension
>     num=${bn#*supercell-}  # Extract the number after 'energy-supercell-'
>
>     if [[ -z ${num} ]]; then
>         continue
>     fi
>     dir="$num"
>
>     if [[ -d $dir ]]; then
>         echo "Skipping: Directory $dir already exists."
>         continue  # Skip to the next iteration instead of exiting
>     fi
>
>     mkdir "$dir"
>
>     mv "$i" "$dir/energy.inp"
>   #  if [[ "$dir" != "001" ]]; then # Keep Wavefunction file from first 
> simulation to make others quicker
>  #       cp 001/MIN-supercell-001-RESTART.wfn 
> "$dir/MIN-supercell-$num-RESTART.wfn"
> #    fi
>
>     cd "$dir"
>     srun cp2k.psmp -i energy.inp -o result.out
>     cd ../
>
> done
>
> phonopy --cp2k -f $(for i in {1..13}; do printf "%03d/MIN-supercell-%
> 03d-forces-1_0.xyz " $i $i; done)
>
> phonopy --cp2k -c energy.inp -p -s --dim="6 6 2" --band " 0.0 0.0 0.0  0.0 
> 0.0 0.5  0.0 0.5 0.5  0.0 0.5 0.0  0.5 0.5 0.0  0.5 0.0 0.0  0.5 0.0 0.5 
>  0.5 0.5 0.5" --mesh 41 41 41
>
>  
>
>  
>
> I have tried various supercell sizes, and have found that 4x4x2 should be 
> more than converged enough. My energy.inp file has the coordinates of the 
> relaxed unitcell printed as follows:
>     &CELL
>       A 3.870961 0 0
>       B 0 3.9540005 0
>       C 0 0 11.8602555
>                 PERIODIC XYZ
>     &END CELL
>    &COORD
> Ba 1.9354787033 1.9770037827 9.6771343800
> Ba 1.9354787999 1.9770022809 2.1442401159
> Y 1.9354802394 1.9770060545 5.8953066752
> Cu -0.0000006631 0.0000091908 7.6100822750
> Cu -0.0000006426 0.0000055768 4.1724491018
> Cu -0.0000078256 0.0000008848 -0.0048129326
> O -0.0000004062 1.9770007276 -0.0014139445
> O 1.9354801904 0.0000011797 7.3249015553
> O 1.9354801707 0.0000009008 4.4580654293
> O -0.0000001259 1.9770034877 7.3279693035 <(327)%20969-3035>
> O -0.0000001304 1.9770024288 4.4553471390
> O -0.0000006180 0.0000004257 9.9643054004
> O -0.0000006737 0.0000003017 1.8887411088 <(888)%20741-1088>
>       &END COORD
>
>  
>
> My geometry has been relaxed in a 6x6x2 supercell, with an RMS_FORCE 
> setting of 1e-8 with the CELL_OPT method. The sum of atomic forces at the 
> end lines up with this. I have converged my plane wave cutoff and relative 
> plane wave cutoff (how converged does this need to be?). I worry it may be 
> an issue with negative coordinates in the unit cell, however the periodic 
> XYZ command should wrap these as I understand? 
>
>  
>
> If anyone has any suggestions as to why my phonons look quite bad I'd be 
> very thankful.
>
>  
>
> Thanks in advance 
>
> Ashley 
>
>  
>
> -- 
> 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.
> To view this discussion visit 
> https://groups.google.com/d/msgid/cp2k/d5c45f96-776c-4803-90b0-2584e1cd8f4an%40googlegroups.com 
> <https://groups.google.com/d/msgid/cp2k/d5c45f96-776c-4803-90b0-2584e1cd8f4an%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/d8d75fed-c55e-4288-adff-cdb9e9e31e16n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250909/84292870/attachment-0001.htm>