[CP2K-user] [CP2K:22326] Troubleshooting: changing periodicity and index to radix array not found

[Frederick Stein]

No, your output file states that the given Poisson solver doesn't implement 
stress tensors for 2D-periodic systems. I am pretty sure that all Poisson 
solvers implement stress tensors for 3D-periodic systems but not 
necessarily for 1D or 2D systems. In a few cases, this has changed over the 
years.

An alternative would be to describe your system as a 3D-periodic system if 
the boy in the non-periodic direction carries enough vacuum.
Best

Michela Cavalieri schrieb am Mittwoch, 15. Juli 2026 um 18:01:49 UTC+2:

> Please scrap the first part of my question. I am seeing now that they 
> should match. I will leave it at this: trying NUMERICAL stress tensor. Will 
> update here if it works or if I get an updated error message. Thanks :)
>
> On Wednesday, July 15, 2026 at 12:00:17 PM UTC-4 Michela Cavalieri wrote:
>
>> Hello Dr. Stein, thank you for your reply! I am actually a little 
>> confused about this comment you made: " I am not sure which Poisson solver 
>> implements stress tensors with PERIODIC XY. ". I thought MT, IMPLICIT and 
>> WAVELET solvers were available for 2D systems. Does this only refer to the 
>> electrostatics? In other words, could I use PERIODIC XY unit cell (CELL) 
>> but PERIODIC XYZ electrostatics in POISSON_SOLVER?
>>
>> Funnily enough, while I was trying to understand this, I found this thread 
>> from last year where the two of spoke about a similar issue 
>> <https://groups.google.com/g/cp2k/c/-pJubMFm70Y>! I am now trying 
>> NUMERICAL stress tensor following your advice (left the system in 2D for 
>> now).
>>
>> Thank you,
>> On Wednesday, July 15, 2026 at 9:44:24 AM UTC-4 Frederick Stein wrote:
>>
>>> Dear Michela,
>>> These are great news! You may try to reduce the print level again to 
>>> have a less verbose CP2K. I guess that the missing restart file is related 
>>> to CP2K aborting before finishing the first optimization step. You will 
>>> need to employ PERIODIC XYZ instead of PERIODIC XY. I am not sure which 
>>> Poisson solver implements stress tensors with PERIODIC XY.
>>> Best,
>>> Frederick
>>>
>>> Michela Cavalieri schrieb am Mittwoch, 15. Juli 2026 um 15:35:46 UTC+2:
>>>
>>>> Good morning Dr. Stein and CP2K community!
>>>>
>>>> I have an update. With the library you suggested, the job ran (thank 
>>>> you!!). However, no .restart file was generated, so I am a little confused.
>>>>
>>>> The changes I made, as mentioned earlier, include the new library you 
>>>> suggested + PRINT_LEVEL HIGH.
>>>>
>>>> The files that were generated include:
>>>> - .out and .err (attached)
>>>> - a number of localLog.out files
>>>> - a number of nonbonded_nl.out files
>>>> - ELF, efield, hartree, ELECTRON_DENSITY, TOTAL_DENSITY, WFN cubes
>>>> - a restart.wfn file
>>>>
>>>> but no restart!!! Now, my geo_opt is not done, so I was planning on 
>>>> restarting and I would normally write the file myself, but coordinates, 
>>>> cell dimensions, velocities etc were not printed as individual files or as 
>>>> part of the output. Is there anything I should fix in my original script to 
>>>> ensure a restart gets printed? I could maybe just switch back to 
>>>> PRINT_LEVEL LOW like I had before.
>>>>
>>>> Notably for the output and error files,
>>>> 1. The .err reports an OpenFabrics error.
>>>> 2. The .out reports some problems, such as: "Stress Tensor not tested 
>>>> for 2D Systems" (aborted after); "Berry phase moments for non uniform MOs' 
>>>> occupation numbers not implemented".
>>>>
>>>> Kindly let me know if I should upload anything else!
>>>>
>>>> Thank you, 
>>>>
>>>> Michela
>>>> On Tuesday, July 14, 2026 at 9:40:29 AM UTC-4 Michela Cavalieri wrote:
>>>>
>>>>> Dr. Stein, thank you so much for your help! Sorry for the long time to 
>>>>> respond on my end as well. For some reason, I did not receive an email 
>>>>> notification! I am trying FFTW3 now and increased the print level to HIGH. 
>>>>> I will update you as soon as possible.
>>>>>
>>>>> Thank you very much!
>>>>> Michela
>>>>>
>>>>> On Thursday, July 9, 2026 at 11:25:07 AM UTC-4 Stein, Dr. Frederick 
>>>>> wrote:
>>>>>
>>>>>> Dear Michela, 
>>>>>> Sorry for the long time to respond. I guess it is just a very special 
>>>>>> question. 
>>>>>> Your system may be too large for the internal FFT library. Please 
>>>>>> consider 
>>>>>> PREFERRED_FFT_LIBRARY FFTW3. I have just run the calculation with 
>>>>>> FFTW3 but 
>>>>>> replaced your system with hydrogen such that the underlying FFT-grids 
>>>>>> should 
>>>>>> be the same. If you still observe an error, could you increase the 
>>>>>> print 
>>>>>> level and send a suitable output file. 
>>>>>> Best, 
>>>>>> Frederick 
>>>>>>
>>>>>

-- 
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/4ae1cd77-0937-4f79-9692-14aba1bac6f7n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260715/54384691/attachment.htm>