[QE-developers] Bad scalability of supercells energy with Hybrid functional.

[Paolo Giannozzi]

Does this happen for hybrid functionals with USPP only? If so, use 
norm-conserving PPs. The hybrid+USPP/PAW combo isn't any faster and has 
some known problems

Paolo

On 11/14/24 16:25, jmasnaghetti wrote:
> Dear QE developer,
> 
> I'm a master student from the University of Milan and I've been working
> on simulating Silicon Germanium alloys with hybrid functional for my
> master thesis. I've already posted a similar question to the users
> forum, but I've been advised by my supervisors to post it to the
> developer as it may be related to something in the actual code of QE.
> Basically, my problem seems to be the fact that, running some tests, the
> energies per ion are not consistent between the classic 2 ions cells and
> bigger 8 or 16 atoms supercells if i use ultra-soft pseudos. The
> differences are of the order of tens of meV. For comparison, I've always
> set my convergence parameters so that the energies are correct within 1
> meV, and that seems to be the case for norm conserving pseudos, form
> whom the per ion energy differences between the conventional cell and
> supercell are of the order of 0.1 meV. I did not change any of the
> energy cutoffs between the conventional cell and the supercells and i
> tried to keep the products between the number of k points and atoms
> consistent across the different cells to prevent errors given by
> inconsistent Brillouin zone samplings. Because these discrepancies arise
> only for US pseudos, and they seem to get bigger as the 2 cutoff radii
> in the pseudos grow apart, we have thought that it might be some
> something related to the calculation of the augmentation charges.
> I've attached 2 input files as reference for the type of simulation I'm
> running.
> 
> Thank you in advance,
> Jacopo Masnaghetti.
> 
> 
> ________________________________________________
> The Quantum ESPRESSO community stands by the Ukrainian people
>   and expresses its concerns about the devastating effects that
> the Russian military offensive has on their country and on the
> free and peaceful scientific, cultural, and economic cooperation
> amongst peoples.
> _______________________________________________
> developers mailing list
> developers at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/developers

-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine Italy, +39-0432-558216