[QE-users] How to add Hubbard U manifold labels to a PP?

[Iurii TIMROV]

Dear Chris,


> <PP_CHI.7 type="real" size="  1110" l="3" occupation=" 7.0000" columns="3">
>  <PP_CHI.8 type="real" size="  1110" l="3" occupation=" 0.0000" columns="3">


This is the pseudopotential generated using the ATOMPAW code, right? I discussed with Natalie Holzwarth (the developer of the ATOMPAW code) and she explained to me that the second channel is actually an unbound empty state (its occupation is zero). So you should not add a "4F" label to it. You can set it e.g. to 0F (or anything else) so that the code does not apply the +U correction to it.


So you should have something like this:

<PP_CHI.7 type="real" size="  1110" l="3" occupation=" 7.0000" columns="3" label="4F">
<PP_CHI.8 type="real" size="  1110" l="3" occupation=" 0.0000" columns="3" label="0F">


This will be automatically handled by the new versions of the ATOMPAW code.


BTW, in my recent fix related to Hubbard_occ there was a bug. Please check this bug fix: https://gitlab.com/QEF/q-e/-/merge_requests/1976/diffs


HTH


Iurii


--
Dr. Iurii TIMROV
Senior Research Scientist
Theory and Simulation of Materials (THEOS)
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Christoph Wolf <wolf.christoph at qns.science>
Sent: Sunday, October 9, 2022 11:54:10 AM
To: Quantum Espresso users Forum
Subject: Re: [QE-users] How to add Hubbard U manifold labels to a PP?

Dear Iurii,

Thank you very much for your fix. I am still facing an issue here, probably specific to my PPs - I wonder if you can help me out:

These PPs have two "partial waves" per wave function, e.g. two 5d and two 4f waves. It seems that this somehow leads to an issue with the occupations as follows:

My calculations applies a U on 4f and 5d:
HUBBARD ortho-atomic
U Gd-4f 9.0
U Gd-5d 6.0

The neutral configuration of Gd is 4f7 5d1, and this is reflected in the scf output:

     STARTING HUBBARD OCCUPATIONS:

     =================== HUBBARD OCCUPATIONS ===================
     ------------------------ ATOM  144 ------------------------
     Tr[ns(144)] (up, down, total) =   7.00000  0.00100  7.00100
     Atomic magnetic moment for atom 144 =   6.99900
     SPIN  1
     eigenvalues:
       1.000  1.000  1.000  1.000  1.000  1.000  1.000
...

     Tr[ns(144)] (up, down, total) =   0.50050  0.50050  1.00100
     Atomic magnetic moment for atom 144 =   0.00000
     SPIN  1
     eigenvalues:
       0.100  0.100  0.100  0.100  0.100

However, after scf convergence:

End of self-consistent calculation

     =================== HUBBARD OCCUPATIONS ===================
     ------------------------ ATOM  144 ------------------------
     Tr[ns(144)] (up, down, total) =   0.17333  0.17427  0.34761
     Atomic magnetic moment for atom 144 =  -0.00094

 Background part
     Tr[ns(144)] (up, down, total) =   0.13969  0.14315  0.28284
     Atomic magnetic moment for atom 144 =  -0.00345

which looks like the 4f shell lost all its electrons.

If I check the lowdin charges from projwfc.x:

     Atom # 144: total charge =  17.7579, s =  2.1425, p =  6.0956, d =  1.9544, f =  7.5654

(now slight variations between the different schemes of calculating the charges always happen but this is clearly something else).

Now strangely enough I used the old version of the U+V code and recompiled with lanthanide background channels set to "5d" and occupation set to "1" and this problem does not happen - but recompiling it every time is a bit tedious and the new way in QE 7.1 is very convenient.

Just for completeness, the pseudo has wfcs like this:

 <PP_CHI.7 type="real" size="  1110" l="3" occupation=" 7.0000" columns="3">
 <PP_CHI.8 type="real" size="  1110" l="3" occupation=" 0.0000" columns="3">

(here for the 4f). I don't know exactly why this is built this way (the paper argued this generally makes them softer and better behaved I believe), but it seems to me that the new Hubbard code prints the occupations of CHI.8 and not CHI.7.

Thank you for your assistance with this matter!

Best regards,
Chris

--
Group Leader "Theory of Quantum Systems at Surfaces"
IBS Center for Quantum Nanoscience
Seoul, South Korea
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221009/25636b53/attachment.html>