[QE-users] Negative dr2 error in DFT+U

[Timrov Iurii]

Dear Dominik,


This is the first time that I hear about negative Ueff. What I would do, I would just compute Ueff from first principles using the hp.x code of Quantum ESPRESSO:

Phys. Rev. B 98, 085127 (2018).

Best regards,
Iurii

P.S.: Please do not forget to add your affiliation when posting to the pw_forum.


--
Dr. Iurii Timrov
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
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 Dominik Gresch <greschd at gmx.ch>
Sent: Tuesday, October 8, 2019 2:43:58 PM
To: users at lists.quantum-espresso.org
Subject: [QE-users] Negative dr2 error in DFT+U

Dear QE community,

I am trying to run DFT+U calculations (lda_plus_u_kind=0) with a negative Ueff = U - J. However, this frequently results in the following error being raised:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine mix_rho (1):
     negative dr2
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

I have traced this to the ns_ddot function in PW/src/scf_mod.f90, which calculates a contribution to dr2 from the Hubbard term. Due to the negative U, this value also becomes negative. Consequently, the error can be avoided by changing Hubbard_U(nt) to ABS(Hubbard_U(nt)) in this function (in two places).

My question is this:

- Does this change make sense, or is the sign of ns_ddot relevant?

- Does it make sense physically to use negative U? These are empirical values, to match a known band structure.

Thank you and best regards,

Dominik Gresch
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20191008/cb8414da/attachment.html>