[QE-users] ph.x: problem with "electric fields calculation" - again

[Christoph Wolf]

Dear all,

I have read several posts regarding this issue but I am still unable to get
the electrical fields calculation in ph.x converged for phonon dispersion
calculation. The behavior, which many experience is as follows:

      iter #  10 total cpu time : 21241.7 secs   av.it.:   1.3
      thresh= 6.227E-05 alpha_mix =  0.400 |ddv_scf|^2 =  *1.388E-08*
      iter #  11 total cpu time : 22225.5 secs   av.it.:   4.4
      thresh= 1.178E-05 alpha_mix =  0.400 |ddv_scf|^2 = * 1.528E-12*
      iter #  12 total cpu time : 24354.9 secs   av.it.:  18.0
      thresh= 1.236E-07 alpha_mix =  0.400 |ddv_scf|^2 =  *3.025E-07*
      iter #  13 total cpu time : 25425.8 secs   av.it.:   5.7
      thresh= 5.500E-05 alpha_mix =  0.400 |ddv_scf|^2 =*  6.748E-08*
      iter #  14 total cpu time : 26243.8 secs   av.it.:   2.9
      thresh= 2.598E-05 alpha_mix =  0.400 |ddv_scf|^2 =*  2.569E-08*
      iter #  15 total cpu time : 27076.0 secs   av.it.:   3.3
      thresh= 1.603E-05 alpha_mix =  0.400 |ddv_scf|^2 =  6.075E-08
      iter #  16 total cpu time : 27797.8 secs   av.it.:   2.3
      thresh= 2.465E-05 alpha_mix =  0.400 |ddv_scf|^2 =  3.101E-07
      iter #  17 total cpu time : 29230.3 secs   av.it.:   9.7
      thresh= 5.569E-05 alpha_mix =  0.400 |ddv_scf|^2 =  2.663E-07
      iter #  18 total cpu time : 30358.8 secs   av.it.:   6.5
      thresh= 5.160E-05 alpha_mix =  0.400 |ddv_scf|^2 =  3.309E-07
      iter #  19 total cpu time : 31460.7 secs   av.it.:   6.0
      thresh= 5.752E-05 alpha_mix =  0.400 |ddv_scf|^2 =  8.663E-08

that is, after originally apparently converging it starts oscillating
between 1E-8 and 1E-7 until 100 iterations. Reducing alpha_mix down to 0.01
does not alleviate this problem.

Interestingly this does not happen for Gamma point only:

     Electric Fields Calculation

      iter #   1 total cpu time :  3714.8 secs   av.it.:   8.3
      thresh= 1.000E-02 alpha_mix =  0.700 |ddv_scf|^2 =  1.982E-08
      iter #   2 total cpu time :  5447.0 secs   av.it.:  19.2
      thresh= 1.408E-05 alpha_mix =  0.700 |ddv_scf|^2 =  1.787E-08
      iter #   3 total cpu time :  7174.1 secs   av.it.:  18.0
      thresh= 1.337E-05 alpha_mix =  0.700 |ddv_scf|^2 =  1.003E-09
      iter #   4 total cpu time :  8856.3 secs   av.it.:  18.0
      thresh= 3.167E-06 alpha_mix =  0.700 |ddv_scf|^2 =  9.881E-11
      iter #   5 total cpu time : 10554.6 secs   av.it.:  18.3
      thresh= 9.940E-07 alpha_mix =  0.700 |ddv_scf|^2 =  1.223E-10
      iter #   6 total cpu time : 12142.6 secs   av.it.:  16.9
      thresh= 1.106E-06 alpha_mix =  0.700 |ddv_scf|^2 =  2.022E-12
      iter #   7 total cpu time : 13919.5 secs   av.it.:  18.7
      thresh= 1.422E-07 alpha_mix =  0.700 |ddv_scf|^2 =  3.477E-12
...
...
      iter #  12 total cpu time : 22432.3 secs   av.it.:  17.7
      thresh= 1.628E-09 alpha_mix =  0.700 |ddv_scf|^2 =  1.601E-16
      iter #  13 total cpu time : 24072.0 secs   av.it.:  17.3
      thresh= 1.265E-09 alpha_mix =  0.700 |ddv_scf|^2 =  *1.580E-17*

even with alpha_mix=0.7 (default value).

The system is a relaxed CsPbBr3 (I attach the input below).

Anyone with more experience an idea what to do here?

Have a good weekend everyone!

Chris

&control
 calculation = 'scf'
 prefix='cspbbr3',
 pseudo_dir = '../../pseudo',
 outdir='./'
 wf_collect=.true.
 verbosity='high'
 nstep = 500
/
&SYSTEM
  ibrav = 0
  A =    8.24400
  nat = 20
  ntyp = 3
  ecutwfc = 100,
  occupations='fixed'
/
&ELECTRONS
 conv_thr=1e-17
 diago_thr_init=1e-3
 diagonalization='cg'
/
ATOMIC_SPECIES
  Pb  207.20000 Pb_ONCV_PBE-Chris.UPF
  Cs  132.90500 Cs_ONCV_PBE-Chris.UPF
  Br   79.90400 Br_ONCV_PBE-Chris.UPF

K_POINTS automatic
9 9 9 0 0 0
CELL_PARAMETERS (alat= 15.57890224)
   1.034550924   0.000000000   0.000000000
   0.000000000   1.442904322   0.000000000
   0.000000000   0.000000000   1.002723182

ATOMIC_POSITIONS (crystal)
Pb       0.000000000   0.500000000   0.000000000
Pb       0.500000000   0.000000000   0.500000000
Pb       0.500000000   0.500000000   0.500000000
Pb       0.000000000   0.000000000   0.000000000
Cs       0.557073579   0.250000000   0.015983762
Cs       0.942926421   0.750000000   0.515983762
Cs       0.442926421   0.750000000   0.984016238
Cs       0.057073579   0.250000000   0.484016238
Br       0.293064750   0.462313289   0.204460039
Br       0.206935250   0.962313289   0.704460039
Br       0.206935250   0.537686711   0.704460039
Br       0.706935250   0.962313289   0.795539961
Br       0.706935250   0.537686711   0.795539961
Br       0.793064750   0.037686711   0.295539961
Br       0.793064750   0.462313289   0.295539961
Br       0.293064750   0.037686711   0.204460039
Br       0.013732263   0.750000000   0.072407914
Br       0.486267737   0.250000000   0.572407914
Br       0.986267737   0.250000000   0.927592086
Br       0.513732263   0.750000000   0.427592086

&inputph
  prefix   = 'cspbbr3',
  recover  = .false.
  amass(1) = 207.20000
  amass(2) = 132.90500
  amass(3) = 79.90400
  epsil    = .true.,
  fildyn   = 'cspbbr3.dyn',
  ldisp    = .true.
  nq1=4, nq2=3, nq3=4
  fildvscf = 'dvscf'
  tr2_ph   =  1.0d-12,
  alpha_mix(1)=0.01
  alpha_mix(2)=0.01
  alpha_mix(3)=0.01
  alpha_mix(4)=0.01
/
-- 
Postdoctoral Researcher
Center for Quantum Nanoscience, Institute for Basic Science
Ewha Womans University, Seoul, South Korea
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180825/97e27206/attachment.html>