[QE-developers] Status of realspace USPP treatment

[Tobias Klöffel]

Hi Paolo,

thank you, so we have to check ourself.

For the g-space HFX newd/usdens routines we have some changes ready that 
enhances performance quite a bit (down from >2days to ~8 hours). I am 
currently thinking about ionic forces with USPP and HFX but that might 
still need a few more thoughts on how to do it efficiently.

As soon as it is ready, I will create a PR for you to review.

Best
Tobias


On 11/9/25 5:47 PM, Paolo Giannozzi wrote:
> On 11/9/2025 4:50 PM, Tobias Klöffel wrote:
>>
>> Hi all,
>>
>> we are currently looking into HFX with USPP and wonder what's the status
>> of the USPP real space implementation.
>
> It works, but it is not very efficient in parallel execution (MPI) and 
> no systematic testing has been performed
>
>> Is there a paper out that compares the real space and G space
>> implementations in QE? 
>
> Not that I know
>
>> What is the current best practice to use the real space treatment?
>> I noticed three input variables:
>> tbeta_smoothing
>> tq_smoothing
>> real_space
>> tqr
>
> tqr = use real-space implementation for augmentation charges Q_lm(r)
> real_space = as above, for projectors \beta_l(r)
> tq_smoothing = use Fourier smoothing to improve the precision of the 
> real-space algorithm for augmentation charges Q_lm(r)
> tbeta_smoothing = as above, for projectors \beta_l(r)
>
>> Should all of them be enabled?
>
> In principle, they should
>
>> Also there are some notes that ecutfock should work with USPP in the
>> code but the code ignores ecutfock for USPP. What's the status of using
>> ecutfock with USPP?
>
> I don't think it was ever implemented
>
> Paolo
>
>>
>> Best
>> Tobias
>>
>> -- 
>> Dr. Tobias Klöffel
>> =======================================================
>> Erlangen National High Performance Computing Center (NHR at FAU)
>> Erlangen Regional Computing Center(RRZE)
>> Friedrich-Alexander-Universität Erlangen-Nürnberg
>> Martensstr. 1
>> 91058 Erlangen
>>
>> Room: 1.133
>> Phone: +49 (0) 9131 / 85 - 20101
>>
>> =======================================================
>>
>> E-mail: tobias.kloeffel at fau.de
>>
>> ________________________________________________
>> The Quantum ESPRESSO Foundation stands in solidarity with all 
>> civilians worldwide who are victims of terrorism, military 
>> aggression, and indiscriminate warfare.
>> _______________________________________________
>> developers mailing list
>> developers at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/developers
>

-- 
Dr. Tobias Klöffel
=======================================================
Erlangen National High Performance Computing Center (NHR at FAU)
Erlangen Regional Computing Center(RRZE)
Friedrich-Alexander-Universität Erlangen-Nürnberg
Martensstr. 1
91058 Erlangen

Room: 1.133
Phone: +49 (0) 9131 / 85 - 20101

=======================================================

E-mail: tobias.kloeffel at fau.de