[QE-users] Problem in vc-relax calculation with Hubbard U in magnetic system

[Iurii TIMROV]

Dear Thomas,


> ...the difference might come from some details in the implementation of +U or from some things related to the pseudo


Indeed, the difference in pseudos might have an impact. But also there are important differences in the DFT+U implementations:

- in VASP the PAW Hubbard projectors are used

- in QE the atomic or ortho-atomic Hubbard projectors are typically used (but you can also try the PAW projectors but this requires pseudos with AE wavefunctions, see PWCOND/examples/example03)


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: Kazume NISHIDATE <nisidate at iwate-u.ac.jp>
Sent: Monday, November 28, 2022 12:21:47 AM
To: Quantum ESPRESSO users Forum
Cc: Vivekanand Shukla; Iurii TIMROV
Subject: Re: [QE-users] Problem in vc-relax calculation with Hubbard U in magnetic system

Dear Thomas Brumme,

I'm not familiar with the +U method and this is a general view.

- One possibility is the difference in the relaxation route achieving
  the most stable state. In some cases, the system may be trapped in
  quasi stable state.

  Suppose that you get structure A in QE and structure B in VASP, both
  starting from the original structure G, you can compare the relative
  energies by performing the static scf calculations for A, B, and G.

  If both of the codes return the same order of energies, such as E(A)
  < E(B) < E(G), you can safely identify the stable structure (in this
  case the structure A). Be careful about the k-points and cutoff
  energies. The number of PWs of A must be different from that
  of B and you must maintain the accuracy by increasing the k-points
  and the cutoff.

- Are you using dense k-points mesh and high cutoff energies in the
  relaxation process too?

- Of course the difference of the pseudo/PAW potentials may influence
  the results. This is the reason why the VASP has variety of
  pseudo/PAW potentials for some elements. Try the same calculations
  using different pseudo/PAW sets.

- Try the VASP relaxations without +U option. Will it return the same
  structure with +U?



> 2022/11/28 0:43、Thomas Brumme <thomas.brumme at tu-dresden.de>のメール:

> In VASP the relaxation is not restricted and since both codes use more or less the same theory, the difference might come from some
> details in the implementation of +U or from some things related to the pseudo. Do you have any feeling/idea if changing to a different
> pseudo or something different concerning +U might influence the relaxation?
>
> Kind regards
>
> Thomas


西館数芽
Kazume NISHIDATE Ph.D

Department of Systems Innovation Engineering,
Graduate School of Science and Engineering, Iwate University
4-3-5 Ueda, Morioka, Iwate, 020-8551 JAPAN
Phone:+81-19-621-6391
kazume.nishidate at gmail.com, nisidate at iwate-u.ac.jp
https://sites.google.com/site/nisidatelab/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221128/54b9a7ab/attachment.html>