[QE-users] Hubbard-U QE vs VASP

Mon Oct 11 14:30:15 CEST 2021

I would like to add a few words on the comments by Iurii:

Using the switch "U_projection='ortho_atomic' " had already been tested in 
my study of meta-stable states.

It does not alter the main arguments in my case; but what I had found was 
that using the normalized localized atomic orbitals slightly changes the 
electron densities and thereby slightly shifts the total energies of the GS 
and meta-stable states; but does not alter the hierarchy of the states.

I would like to thank Iurii for the reference provided. 

Mahmoud Payami

NSTRI, AEOI, Tehran, Iran

Email: mpayami at aeoi.org.ir
Phone: +98 (0)21 82066504
--------------------------------------------------------

From: Iurii TIMROV via users <users at lists.quantum-espresso.org>
To: Quantum ESPRESSO users Forum <users at lists.quantum-espresso.org>
Date: Mon, 11 Oct 2021 08:41:59 +0000
Subject: Re: [QE-users] Hubbard-U QE vs VASP

Dear All,

In addition to what Matteo explained, I would like to point your attention 
to the following article which is relevant (in my opinion) for this 
discussion (QE vs VASP): https://aip.scitation.org/doi/10.1063/1.4945608

Greetings,

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 Matteo 
Cococcioni <matteo.cococcioni at unipv.it>
Sent: Monday, October 11, 2021 10:22:08 AM
To: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Hubbard-U QE vs VASP

Dear Jhon,

I would be quite surprised if there is a sign-convention mismatch between 
the two implementations.
Please keep in mind that, even if you use the same type of pseudootentials, 
there is no guarantee that the same U should lead to the
same result (actually this is not the case most often). The value of U 
depends on all the details of the calculations and of the pseudopotnetial.
Of course, It would be interesting to compare the two codes on exactly the 
same pseudopotential but I am afraid this is not (yet) possible.
If you want to compute it (which I recommend) please consider using the hp.x 
code (now part of the QE distribution) that allows you to determine its 
value
using a dfpt implementation of linear response theory which is very 
efficient and user-friendly (PRB 103, 045141 (2021); PRB98, 085127 (2018)). 
Another thing I recommend to use with QE is orthogonalized atomic orbitals 
(U_projection_type = 'ortho-atomic') for which the code now has forces and 
stresses
(PRB 102, 235159 (2020)).
Please keep in mind also (and pehaps this makes up for most of the 
differences you observe) that VASP uses the the projectors of the 
seudopotentials  (of PAW type
mostly) to define the atomic occupations then used in the Hubbard 
correction. In QE atomic wavefunctions are instead mostly used to project 
Kohn-Sham states on.
If you want a stricter comparison between the two perhaps you should use 
U_projection_type = 'pseudo' in QE (use PPs containing wavefunctions - see 
INPUT_PW.txt
for more details).

Best regards,

Matteo

Il giorno lun 11 ott 2021 alle ore 01:37 Jhon Gonzalez 
<jhon.gonzalez at usm.cl> ha scritto:
Hi,

Perhaps this is a bit off-topic, there is an issue when comparing LDA+U 
results in 2D-materials between Quantum-ESPRESSO and VASP.
For V2C monolayers, CrI3 mono- and bi-layers, the magnetic stability changes 
and it is impossible to reproduce the results.

For the V2C monolayer with U = 4 eV, while with quantum-ESPRESSO (US and 
PBE), I find that the ground-state 
(GS) solution is FM and the AF solution is about 1 eV above; with VASP U=4 
eV (LDATYPE=2), I find that the GS solution 
is AF and the FM state is 1 eV above (in agreement with 
dx.doi.org/10.1021/jp507336x). 

Following the discussion on the implementation of Hubbard's interaction: 
https://lists.quantum-espresso.org/pipermail/users/2020-May/044521.html
There, it is suggested that in QE the Hubbard-U tends to decrease the gap, 
while in VASP it tends to open it.
And inspired by the discussion 
http://grandcentral.apam.columbia.edu:5555/tutorials/dft_procedures/linear_response_u/index.html 

there they mention an "empirically is a difference in sign convention". I 
tried a VASP calculation with U = -4 eV, and I found an 
FM GS-solution followed by an AF solution 0.3 eV above, following the QE 
stability order.

>From my experience comparing QE results with CrI3 experiments, it seems that 
QE uses the proper sign convention for the U term. 
However, I do not have any evidence for this.

I am currently exploring the use of hybrid functionals to address this 
issue, is there an elegant way to settle this dispute?

Best,

Jhon W. González

Departamento de Física
Universidad Técnica Federico Santa María
VALPARAISO-CHILE
_______________________________________________
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

--
Matteo Cococcioni
Department of Physics
University of Pavia
Via Bassi 6, I-27100 Pavia, Italy
tel +39-0382-987485
e-mail matteo.cococcioni at unipv.it
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20211011/5fbcac1b/attachment.html>