[QE-users] Different ELF values with QE 7.0

[Paolo Giannozzi]

The third and fourth commits of this list:
   https://gitlab.com/QEF/q-e/-/commits/develop/PP/src/elf.f90
extended ELF to spin-polarized cases. The factor "fac" was changed from
    fac = 5.d0 / (3.d0 * (3.d0 * pi**2) ** (2.d0 / 3.d0) )
to
    fac = 5.d0 / (3.d0 * (6.d0 * pi**2) ** (2.d0 / 3.d0) )
These commits however introduced a bug (an out-of-bound error) in the
spin-unpolarized case that I fixed in the latest commit (first of the
list). I did not change the factor "fac", though. If you revert to the
previous value of "fac", you get the same numbers as before for the
spin-unpolarized case, or at least, this is what I observed for a single
case.

Having already wasted too much time on something I have no interest in, I
am waiting for instructions from ELF-friendly people about the correct
value of such a factor.

Paolo

On Tue, Mar 15, 2022 at 1:53 PM Roman Lucrezi via users <
users at lists.quantum-espresso.org> wrote:

> Hey,
>
> has anybody else experienced this behaviour in the ELF calculation with QE
> v.7.0?
>
> Kind regards,
>
> Roman Lucrezi
> Institute of Theoretical and Computational Physics
> Graz University of Technology, Austria
>
> --------------------------------------------------------------------------
> Roman Lucrezi via users Tue, 08 Mar 2022:
>
> Hi everybody,
>
>
> recently I used QE v.7.0 to recalculate the 3D electron localization function
> (ELF) in a system I had already worked on with QE v.6.5, and obtained
> quantitatively quite different results: e.g. for specific 2D cuts I get
> differences in the range of 0 to 0.25 (for ELF(v.7.0)-ELF(v.6.5)), where
> the v.7.0 results are always higher. I have rerun everything with the
> exact same input files, trying v.6.5, v.6.7MaX and v.7.0: v.6.5 and
> v.6.7MaX give the same results, v.7.0. is different as described above. For
> each version I ran a non-polarized (nspin = 1) scf calculation with scalar-relativistic
> PBE ONCV pseudos, the system is metallic and I used a mp smearing. After
> that, I used pp.x for the ELF and an XSF file with this input (and
> varying prefixes):
>
> &inputPP
>    prefix   = 'v6_5',
>    outdir   = './tmp'
>    plot_num = 8
>    filplot  = 'v6_5.elf.dat'
> /
> &plot
>    iflag         = 3,
>    output_format = 5
>    fileout       = 'v6_5.elf.xsf'
> /
>
>
> Do you have any idea on what could cause this discrepancy? I saw in the release
> notes that there were some problems with ELFs in the spin-polarized case,
> and there is also a line for v.7.0, saying that the spin-UNpolarized ELFs
> were grossly wrong in v.6.8. Does this affect only v.6.8, or each version
> up to v.6.8, meaning that only v.7.0 gives correct results?
>
> Kind regards,
>
> Roman Lucrezi
> Institute of Theoretical and Computational Physics
> Graz University of Technology, Austria
>
> _______________________________________________
> 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
> _______________________________________________
> 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/20220315/f94d3c68/attachment.html>