[QE-users] Self-consistent calculation of Hubbard U parameter diverges

[Dorde DANGIC]

Dear Iurii,

Thanks for clarification.

Kind regards,

Dorde

On Oct 11, 2023 10:14 AM, Iurii Timrov via users <users at lists.quantum-espresso.org> wrote:

Dear Dorde,


The electronic configuration of Lu is [Xe<https://en.wikipedia.org/wiki/Xenon>] 4f14 5d1 6s2


You are using QE v6.8 with the old Hubbard input syntax. The Hubbard manifold is hard-coded in this version and it is 4f for Lu (hubbard_l = 3). So you are trying to compute U for the 4f shell which is fully occupied.


The HP code is based on linear-response theory (DFPT to 1st order) which can be used for open shells only. If you try to use it for closed shells (like you do), the U parameter will diverge in the self-consistent cycle. This is indeed what you observe. Please have a look at this paper: K. Yu and E.A. Carter, J. Chem. Phys. 140, 121105 (2014).


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 Dorde DANGIC <dorde.dangic at ehu.eus>
Sent: Wednesday, October 11, 2023 9:57:58 AM
To: users at lists.quantum-espresso.org
Subject: [QE-users] Self-consistent calculation of Hubbard U parameter diverges

Dear all,

I have tried to calculate the Hubbard U parameter self-consistently using hp.x code distributed with quantum-espresso software. In my calculations, the estimate of the U parameter constantly increases with each iteration. I started with a value of 0.0001 in the 0th iteration, relaxed the structure, and ran scf calculation followed by hp.x calculation. The first estimate of the U parameter is 25.6456 eV. I ran the second iteration with this value (relaxed the structure and ran scf.x/hp.x) and got the second estimate of 116.1282 eV. On the third iteration, U was 1073.3687 eV. I tried calculations with 'ortho-atomic' and 'atomic' projections. I tried calculations with the different starting U. I always get the divergence of the U parameter. I am using v.6.8 version of quantum-espresso. I attached pw.in and hp.in. Am I doing something wrong? If not, what could be the reason for this behavior?

pw.in:

&control
    prefix = 'LuH'
    calculation = "scf"
    tprnfor = .true.
    tstress = .true.
    outdir='./'
    pseudo_dir='/scratch/djordjedangic/LuNH/'
&end
&system
    ibrav             = 0
    nat               = 4
    ntyp              = 2
    ecutwfc           = 100.0
    occupations       = 'smearing'
    smearing          = 'gaussian'
        degauss = 0.015
    input_dft         = 'pbe'
    lda_plus_u        = .true.
    lda_plus_u_kind   = 0
    U_projection_type = 'ortho-atomic'
    Hubbard_U(1)      = 0.0001
&end
&electrons
        conv_thr = 1e-14
&end

ATOMIC_SPECIES
Lu      174.9668   Lu.upf
H       1.00784    H.upf

K_POINTS automatic
12 12 12 0 0 0

CELL_PARAMETERS (angstrom)
  -2.494212191  -0.000000000   2.494212191
   0.000000000   2.494212191   2.494212191
  -2.494212191   2.494212191  -0.000000000

ATOMIC_POSITIONS (crystal)
Lu           -0.0000000000        0.0000000000       -0.0000000000
H             0.2500000000        0.2500000000        0.2500000000
H             0.7500000000        0.7500000000        0.7500000000
H             0.5000000000        0.5000000000        0.5000000000

hp.in:

&inputhp
    prefix = 'LuH'
    outdir='./'
    nq1=6
    nq2=6
    nq3=6
    conv_thr_chi=1.0d-12
/

Kind regards,

Dorde

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20231011/ad0baa49/attachment.html>