[QE-users] Stress in hybrid functional calculations

[Paolo Giannozzi]

The stress converges _very_ slowly with cutoff. The ultimate test for
stress is to compare it with the numerical derivatives of the energy,
computed at a _fixed_ number of plane waves. This has been done more than
once and I am quite sure that the stress is correctly computed.  There are
still some problems in variable-cell calculations, though, of unclear
origin: https://gitlab.com/QEF/q-e/-/issues/204.

Paolo

On Thu, Oct 14, 2021 at 1:17 PM Audrius Alkauskas <audrius.alkauskas at ftmc.lt>
wrote:

> Dear all,
>
> We have started testing the Quantum Espresso implementation of screened
> Fock exchange (HSE functional), benchmarking against VASP results (the
> input file below).
>
> As a test case, we took simple bulk diamond. If we calculate the energy
> as a function of the lattice constant, the resulting equilibrium lattice
> parameter (3.555 A) and bulk modulus are very similar to the results
> from VASP, which is nice. However, the calculated stress for the
> equilibrium lattice parameter is still large (~30 kBar).  The stress is
> linear as a function of the lattice constant and is ~0 for the lattice
> constant 3.585 A, yielding an offset of about 0.03 A with respect to
> what I assume is the "correct" HSE value. The value of 3.585 A is much
> closer to the PBE result.  Are there still some issues when calculating
> stresses with hybrids or we do something wrong? Energies seem OK,
> stresses not (the error is not huge, but still...)
>
> If anybody has any insights, I would be happy to get receive them.
>
> Best,
> Audrius
>
> ***
> Input
> ***
>
> &CONTROL
>     calculation      = 'scf'
>     title            = 'carbon-cubic-hse-3.556'
>     verbosity        = 'high'
>     nstep            = 100
>     tstress          = .true.
>     tprnfor          = .true.
>     outdir = '/home/vytautas/HSE/C/relax/3.556'
>     etot_conv_thr    = 1e-06
>     forc_conv_thr    = 1e-05
>     pseudo_dir       = '.'
> /
> &SYSTEM
>     ibrav            = 1
>     celldm           = 6.719866103476117
>     nat              = 8
>     ntyp             = 1
>     nbnd             = 24
>     ecutwfc          = 100
>     ecutrho          = 400
>     ecutfock         = 400
>     input_dft        = 'hse'
>     exxdiv_treatment = 'gygi-baldereschi'
>     x_gamma_extrapolation = .true.
>     nqx1             = 4
>     nqx2             = 4
>     nqx3             = 4
>     celldm(1)        = 6.719866131544066
> /
> &ELECTRONS
>     electron_maxstep = 300
>     conv_thr         = 1e-08
> /
> &IONS
> /
> &CELL
> /
>
> ATOMIC_SPECIES
> C 12.011 C.upf
>
> K_POINTS automatic
> 4 4 4  0 0 0
>
> ATOMIC_POSITIONS angstrom
> C 0.0000000000 0.0000000000 0.0000000000
> C 1.7780000000 1.7780000000 0.0000000000
> C 0.8890000000 0.8890000000 0.8890000000
> C 2.6670000000 2.6670000000 0.8890000000
> C 0.0000000000 1.7780000000 1.7780000000
> C 1.7780000000 0.0000000000 1.7780000000
> C 0.8890000000 2.6670000000 2.6670000000
> C 2.6670000000 0.8890000000 2.6670000000
>
>
>
> --
> Audrius Alkauskas
> Center for Physical Sciences and Technology (FTMC)
> Vilnius, Lithuania
>
> email: audrius.alkauskas at ftmc.lt
> phone: +370 612 22281
> www.puntukas.com / www.ftmc.lt
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 206, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20211014/d6378f6e/attachment.html>