[QE-users] No SCF convergence for simple actinide compounds

[Sergei Butorin]

I use QE 6.4.1 and have encountered a problem of pure or no-convergence in scf runs for simple actinide compounds. For example, for PuO2, I get quick convergence for a primitive cell but for a conventional unit cell the convergence to 'poor' conv_thr=1.1d-6 is reached only after 560 iterations with the same parameters and pseudos in the input file (see below):

&control
  calculation = 'scf'
  verbosity='high'
  prefix = 'system'
  pseudo_dir = '/My_input/PuO2/'
  outdir = './Out'
  wfcdir = 'undefined'
  tstress = .false.
  tprnfor = .false.
/
&system
    ibrav = 0
  A =    5.39819
    nat   = 12
    ntyp  = 2
  noncolin = .false.
  lspinorb = .false.
  ecutwfc = 50
  ecutrho = 400
  occupations = 'fixed'
  nspin  = 1
/
&electrons
  conv_thr = 1.1d-6
  mixing_beta = 0.3
  electron_maxstep = 2000
  startingwfc = 'atomic+random'
  diagonalization = 'david'
/
&ions
/
ATOMIC_SPECIES
Pu   239.0522      Pu.pbe-spfn-rrkjus_psl.1.0.0.UPF
O   15.9994        O.pbe-n-rrkjus_psl.1.0.0.UPF
CELL_PARAMETERS alat
  1.000000000000000   0.000000000000000   0.000000000000000
  0.000000000000000   1.000000000000000   0.000000000000000
  0.000000000000000   0.000000000000000   1.000000000000000
ATOMIC_POSITIONS crystal
Pu   0.000000000000000   0.000000000000000   0.000000000000000
Pu   0.000000000000000   0.500000000000000   0.500000000000000
Pu   0.500000000000000   0.000000000000000   0.500000000000000
Pu   0.500000000000000   0.500000000000000   0.000000000000000
 O   0.250000000000000   0.250000000000000   0.250000000000000
 O   0.750000000000000   0.250000000000000   0.750000000000000
 O   0.750000000000000   0.750000000000000   0.750000000000000
 O   0.250000000000000   0.750000000000000   0.750000000000000
 O   0.250000000000000   0.250000000000000   0.750000000000000
 O   0.750000000000000   0.250000000000000   0.250000000000000
 O   0.250000000000000   0.750000000000000   0.250000000000000
 O   0.750000000000000   0.750000000000000   0.250000000000000
K_POINTS automatic
2  2  2  1  1  1

Things I have tried to improve the situation but without success:

1. Significantly increasing ecutwfc and ecutrho values and keeping the 10-12 ratio for US pseudos.
2. Using NC pseudos both from the sg15 collection and harder TM pseudos with smaller rc with appropriate ecutwfc and ecutrho values in my pw.x input.
3. Smaller mixing-beta down to 0.1.
4. Different mixing_mode (e.g. 'local-TF').
5. Different diagonalization ('cg').

No wonder that in case of a supercell, the convergence value goes only down to a few Ry at best. The occupations = 'smearing' option does not help either.

Similar problems I have encoutered for UO2, although I have not investigated that case in detail.

My question is whether QE (pw.x) is generally so bad in handling the f-element systems or something is really wrong on my side.

Dr. Sergei Butorin
Uppsala University
Sweden








N?r du har kontakt med oss p? Uppsala universitet med e-post s? inneb?r det att vi behandlar dina personuppgifter. F?r att l?sa mer om hur vi g?r det kan du l?sa h?r: http://www.uu.se/om-uu/dataskydd-personuppgifter/

E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200730/9b6bfd96/attachment.html>