[QE-users] Non-convergence

[Matthew Marcus]

I'm trying to optimize the structure of Ni(OH)2 before doing a phonon 
calculation.  This material is an antiferromagnet, and the structure is 
well-known except for the positions of the H atoms.  Therefore, I want 
to relax the unit cell and the H positions.  The problem I have is that 
the first SCF step doesn't converge after 500 electron iterations.  The 
'estimated SCF accuracy' bounces between 0.005 and 0.03 Ry.  What's 
going on?  The input file is:

&CONTROL
   calculation = 'vc-relax'
   etot_conv_thr =   1.0000000000d-04
   forc_conv_thr =   1.0000000000d-04
   outdir = './out/'
   prefix = 'aiida'
   pseudo_dir = './pseudo/'
   tprnfor = .true.
   tstress = .true.
   verbosity = 'low'
/
&SYSTEM
   degauss =   1.4699723600d-02
   ecutrho =   6.0000000000d+02
   ecutwfc =   100
   ibrav = 0
   nat = 10
   nspin = 2
   ntyp = 4
   occupations = 'smearing'
   smearing = 'cold'
   starting_magnetization(1) =   1.0000000000d-01
   starting_magnetization(2) =   -.1
/
&ELECTRONS
   conv_thr =   2.0000000000d-09
   electron_maxstep = 500
   mixing_beta =   2.0000000000d-01
/
&IONS
ion_dynamics='bfgs'
/
&CELL
cell_dynamics='bfgs',
press=0.d0,
press_conv_thr=0.5d0,
/
ATOMIC_SPECIES
Ni1     58.6934 ni_pbesol_v1.4.uspp.F.UPF
Ni2     58.6934 ni_pbesol_v1.4.uspp.F.UPF
H      1.00794 H_ONCV_PBEsol-1.0.upf
O      15.9994 O.pbesol-n-kjpaw_psl.0.1.UPF
ATOMIC_POSITIONS crystal
Ni1           0.0000000000       0.0000000000       0.0000000000 0 0 0
Ni2          -0.0000000000      -0.0000000000       0.5000000000 0 0 0
H            0.3333333300       0.6666666700       0.2311929860 0 0 1
H            0.3333333300       0.6666666700       0.7311931750 0 0 1
H            0.6666666700       0.3333333300       0.2688069800 0 0 1
H            0.6666666700       0.3333333300       0.7688069630 0 0 1
O            0.3333000000       0.6667000000       0.1250000000 0 0 0
O            0.3333000000       0.6667000000       0.6250000000 0 0 0
O            0.6667000000       0.3333000000       0.3750000000 0 0 0
O            0.6667000000       0.3333000000       0.8750000000 0 0 0
K_POINTS automatic
12 12 4 0 0 0
CELL_PARAMETERS angstrom
       3.1140000000       0.0000000000       0.0000000000
      -1.5570000000       2.6968031074       0.0000000000
       0.0000000000       0.0000000000       9.2340000000

	Matthew Marcus