[QE-users] Core wavefunction(Lorenzo Paulatto)

[emin klc]

Dear Lorenzo,

Thank you for your detailed response. I appreciate it. Everything is so
clear now.

Best regards,

Mehmet Emin Kilic, PhD

On Mon, Jun 3, 2019 at 4:41 AM Lorenzo Paulatto <paulatz at gmail.com> wrote:

> My workflow for Xpectra calculations:
>
> 1. first of all do everything with the standard pseudopotential with the
> full core, this is much cheaper (no need of supercell) and easier (no
> need to generate pseudopotentials), all you need is the 1s wavefunction
> in a text file, if the pseudopotential already contains GIPAW data, you
> can extract it, otherwise if the code was generated with ld1.x, you cna
> grab the input (it is at the beginning of the file) add
>   lsave_wfc=.true.
> to the input (check the INPUT_LD1 manual), and regenerate it as
>   ld1.x < input.txt
> with the all-electron data included.
>
> If the pseudo did not come from ld1, you have to speak with the person
> who generated it to see if gipaw data can be included, or use another
> pseudo.
>
> 2. The upf2plotcore script does not work with most UPF files, use the
> attached extract_core.f90, you can put it in upftools/ together with the
> Makefile, just run
>    extract_core.x file.UPF
> and it will produce one file for each core wavefunction (which you pick
> depending on the edge you want)
>
> 3. pw.x: if you have inequivalent atoms of the emitting type, take care
> to assign different types to each one, i.e. S1, S2 so you can get all
> the spectra,
> 3b. xspectra.x: no need to use a supercell, or do any nscf calculation.
> You may want to specify a finer grid of k-points directly in xspectra,
> you can set the right after the namelists:
> ...
>   &cut_occ
>      cut_desmooth=0.1,
>      cut_stepl=0.01,
>   /
> 4 4 4 1 1 1   <-- this is the grid used by xspectra
>
>
> 4. Now you want to refine the spectra, by including the core-hole, grab
> the ld1 input from above, and make a pseudo with only 1s1 electron, be
> careful to not forget the rest of the core/valence configuration, i.e.
> something like this
>   config='[Ne] 3s2 3p4 3d-1',
> may become
>   config='1s1 2s2 2p6 3s2 3p4 3d-2 4s-2 4p-2 4d-2',
> also, change the name of the output file. Normally you doo not need to
> change anything else, but sometimes reducing the pseudization radii is
> required. Check the output of ld1.x to see how much the core-hole
> contracts the orbitals, and reduce the radii accordingly.
>
> 5. You never need the core wavefunction from the pseudopotential with
> the core hole, because the XANES transition is from the unperturbed 1s
> to the perturbed valence, you always use the core wavefunctions from the
> unperturbed pseudopotential.
>
> 6. When doing the pw.x calculation you can set tot_charge=+1 or
> tot_charge=0 or anything in between, you may have a look at
> PhysRevB.98.214104 for a discussion on this. You will need a supercell
> to avoid interactions between core-holes (at least 5-7 Å), but you can
> start with a unit cell to see if it works, and do the expensive
> supercells when everything is sorted out.
>
> hope this helps
>
>
> On 6/1/19 4:49 PM, emin klc wrote:
> > Dear Lorenzo,
> >
> > Thank you for your email and suggestion.
> >
> > The calculation flow for XAS with Xspexctra as follows:
> > Firstly, prepare the GIPAW Pseudopotential.
> > Secondly, extract the core wavefunction
> > Thirdly, prepare input file and run SCF
> > Finally, prepare input file and run Xspectra.
> >
> > Considering the calculation flow, the problem (or unclear part) is to
> > get the core wavefunction.
> > As far as I understand, there are two ways to extract core wave
> functions:
> >
> > One is to use "upf2plotcore.sh" script. (upf2plotcore.sh < Ga.PBE.PP.UPF
> >  > Ga.wfc),  which is your suggestion. Am I correct?
> > If yes,
> > upf2plotcore.sh < Ga.PBE.PP.UPF(without core hole) > Ga.wfc
> > upf2plotcore.sh < Ga.PBE.PP.UPF(with core hole) > h_Ga.wfc.
> > which one is correct for spectroscopy.
> >
> > The other is to extract from all electron (AE) wavefunction which also
> > includes core wave functions. So, can we use the second way to extract
> > core wavefunction?
> > If yes, why the output file by ld1.x does not include 1S core wfc
> > information using my input.
> > Your thoughts are welcomed.
> >
> > Best regards,
> >
> > Mehmet Emin Kilic, PhD
> > -----------------------------
> >
> > Hello,
> > If you want to do spectroscopy with the xspectra code, it does not work
> > like that. You do not need an all-electron pseudopotential, but a pseudo
> > generated with a 1s1 core.
> >
> > --
> > Lorenzo Paulatto
> >
> > On Fri, 31 May 2019, 07:35 emin klc, <klcmemin at gmail.com  <
> https://lists.quantum-espresso.org/mailman/listinfo/users>> wrote:
> >
> >>/Dear QE experts, />//>/I have tried to generate a Ga Pseudopotentials
> for theoretical />/spectroscopy by ld1.x. />/Firstly, I need to have all
> electron (AE) wave functions including 1s 2s />/2p 3s 3p core states and
> valence states. />/I'm having some troubles since the output file (ld1.wfc)
> only gives 2s 2p />/3s 3p as a core states except for 1s, but which is
> important for K edge />/spectroscopy. />/I include below my input file for
> generating all electron wave functions. />//>/Could you say what is wrong
> with my input? />/----------------------- />/&input />/title = 'Ga',
> />/iswitch = 1, />/rel = 1, />/zed = 31.0, />/config = '[Ar] 4s2 4p1
> 3d10.0', />/dft = 'PBE' />/------------------------ />//>/Mehmet Emin
> Kilic, PhD />
> >
> >
> >
> > _______________________________________________
> > Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso
> )
> > users mailing list users at lists.quantum-espresso.org
> > https://lists.quantum-espresso.org/mailman/listinfo/users
> >
>
> --
> Lorenzo Paulatto - Paris
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190603/518cc5fd/attachment.html>