[QE-users] DFPT of thin metal layer: convergence has not been achieved

[Christoph Wolf]

Dear Prof. Mazari and colleagues,

thank you again for your input. I have now run the calcs for a series of
smearings (degauss=0.2, 0.1, 0.01, 0.001) and alpha_mix(1...4) from 0.7
down to 0.0001) but there does not seem to be a trend emerging. If you take
a look at the attached plot it seems that for all cases an alpha_mix=0.1
gives the lowest ddv_scf but even in that case it plateaus around 1e-6 (for
the bulk I get good phonons with a threshold of ~1e-12...1e-14). I attached
the run script I used as I am not confident I did not make a really stupid
and obvious mistake.... I could not increase nmix_ph beyond 4 (there is a
limitation saying is has to be less than 5 hardwired in the code).

Thank you all for your input and patience, this seems to be a quite hard
nut to crack...

Best,
Chris


-- 
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/20181117/21813744/attachment.html>
-------------- next part --------------

PREFIX=Ag
PSEUDO='Ag.pbe-n-kjpaw_psl.1.0.0.UPF'
AMASS=107.8682
A0=7.8502
NK=21
Q=11
ECUTWFC=45
DUAL=8
NLAYER=9

for smear in 0.2  0.1 0.01 0.001; do
 mkdir $smear
 cd $smear

cat >gener_slab.in <<EOF
2                           ! Bravais lattice index ibrav
$A0    0.0 0.0 0.0 0.0 0.0  ! celldm of the bulk unit cell
1                           ! 3 cartesian coordinates (1 crystal, 2 ineq. at.)
1                           ! number of atoms in the unit cell
'$PREFIX' 0.0 0.0 0.0       ! Label and coordinates of the atoms
1 0 0                       ! m_1, m_2, m_3 of the G vector
$NLAYER                     ! number of atoms in the unit cell
1 0 0 1                     ! supercell on the surface (t11, t12, t21, t22)
.True.                      ! if .TRUE. the vacuum distance is given below
25.0                        ! the vacuum distance in a.u.
0                           ! origin shift 
.TRUE.                      ! do the origin shift for slab with even layers
'slab.out'                  ! name of the file with the output coordinates
EOF

/qe-6.3/thermo_pw/tools/gener_3d_slab.x <gener_slab.in >gener_slab.out


let ECUTRHO=$DUAL*$ECUTWFC

cat >scf.in <<EOF
&CONTROL
 calculation = 'scf'
 prefix='$PREFIX',
 pseudo_dir = '/scratch/chwolf/00_pseudo/',
 outdir='./tmp'
 wf_collect=.true.
 verbosity='high'
 nstep = 500
 tprnfor=.true.
/
&SYSTEM
  $(head -4 slab.out)
  ntyp = 1
  occupations = 'smearing'
  smearing    = 'gaussian'
  degauss     =  $smear
  ecutwfc     = $ECUTWFC
  ecutrho     = $ECUTRHO
/
&ELECTRONS
 electron_maxstep = 300
 conv_thr=1e-16
 diago_thr_init=1e-2
/

ATOMIC_SPECIES
 $PREFIX $AMASS $PSEUDO

K_POINTS automatic
$NK $NK 1 0 0 0

ATOMIC_POSITIONS {crystal}
$(tail -$NLAYER slab.out)

EOF

pw.x -ndiag 1  <scf.in >scf.out

for mix in $(seq 0.7 -0.1 0.1) 0.01 0.001 0.0001; do

cat >ph.in <<EOF
--
&inputph
  prefix   = '$PREFIX',
  outdir   ='tmp'
  amass(1) = $AMASS
  epsil    = .false.,
  fildyn   = '$PREFIX.dyn',
  ldisp    = .false.
  fildvscf = 'dvscf'
  nq1=$Q,
  nq2=$Q,
  nq3=1,
  tr2_ph   =  1.0d-9,
  alpha_mix(1)=$mix, alpha_mix(2)=$mix, alpha_mix(3)=$mix,  alpha_mix(4)=$mix
 /
 0.0 0.0 0.0
EOF


ph.x -ndiag 1 <ph.in >ph-$mix-$Q.out

done

rm -r tmp
rm -r _ph0
cd ..
done