[QE-users] Phonon calculations with different parallelization

[Paolo Giannozzi]

On 3/21/24 11:03, Chirantan Pramanik wrote:

> The phonon frequencies are not the same

basically, they are. The differences are small (a few cm-1 at most) and 
mostly affect low-lying frequencies. Self-consistency is not very tight 
(1E-14) and may already explain the observed differences.

Paolo

> Regards,
> Chirantan
> 
> ------------------------------------------------------------------------
> *From:* Paolo Giannozzi <paolo.giannozzi at uniud.it>
> *Sent:* Thursday, March 21, 2024 11:34 AM
> *To:* Quantum ESPRESSO users Forum <users at lists.quantum-espresso.org>; 
> Chirantan Pramanik <chirantan.pramanik at weizmann.ac.il>
> *Subject:* Re: [QE-users] Phonon calculations with different 
> parallelization
> On 3/21/24 03:45, Chirantan Pramanik wrote:
>> 
>> My Phonon calculations provide different frequency values for different 
>> parallelization on the cluster especially for the total number of 
>> processors used. Is itpossible or I am doing something very wrong? As 
>> vibrational frequency is gauge invariat how it can be possible?
>> 
>> Thanks,
>> Chirantan
> 
> It is not possible, with one exception: zero-frequency acoustic modes
> that have non-zero frequencies; these can vary a lot, but they are
> anyway set to zero by imposing the acoustic sum rule.
> 
> It might be a bug, but it is hard to say more without knowing what you
> exactly computed on which code version
> 
> Paolo
> -- 
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 206, 33100 Udine Italy, +39-0432-558216

-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine Italy, +39-0432-558216