[Pw_forum] problem on k points and ecutwfc test

[Bo Peng]

Hi Dear user,

I have tried to test the k-points and ecutwfc convergence for
a Mg(002) slab which contains 7 layers and the middle 5 layers
have been fixed.

PROBLEM I:
I have tested (1x1x1, 2x2x1, 4x4x1, 5x5x1, 7x7x1, 8x8x1, 10x10x1) M-P
k-points grids for one input file, other parameters are the same(see below).

For 2x2x1, 5x5x1 and 8x8x1, I have got
"The maximum number of steps has been reached.
 End of BFGS Geometry Optimization"
However, for others, say, (1x1x1, 4x4x1, 7x7x1, 10x10x1)I have got
"bfgs converged in xx scf cycles and xx bfgs steps
 End of BFGS Geometry Optimization"

For the latter cases, the convergences have completed successfully,
but for the former cases, they seemed not finished. I wonder, if just
for k-points test, whether it is worth to restart the jobs to make the
optimization converged? Or, choosing new K sampling, such as (3x3x1,
6x6x1, 9x9x1, etc.)?

PROBLEM II
when testing ecutwfc convergence, I have chosen 10, 14, 18, 22, 26, 30 Ry
as the value of the ecutwfc. It works well for ecutwfc=18, 26 and 30. When
ecutwfc=14 or 22, the following errors occured:
...
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     from bfgs : error #         1
     bfgs history already reset at previous step
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

     stopping ...

I have checked the mailing list archives, like:
http://www.democritos.it/pipermail/pw_forum/2006-November/005331.html
suggested that restarting the jobs to make the geometry optimization
converged.
However, I wonder why the optimizaition is failed for certain lower values
of
ecutwfc which seems would be more easily to be completed.

In PROBLEM I, ecutwfc=24Ry; In PROBLEM II, K_POINTS are 5 5 1 0 0 0

the following is the input file:
&CONTROL
calculation  = 'relax',
prefix       = 'Mgrelax',
pseudo_dir   = '/....../pseudo/',
outdir       = '/....../tmp/',
tstress=.true.
tprnfor=.true.
/
&SYSTEM
ibrav       = 14,
a           = 3.149307D0,
b           = 3.149307D0,
c           = 32.624496,
cosab       = 0.5D0,
cosac       = 0.D0,
cosbc       = 0.D0,
nat         = 7,
ntyp        = 1,
ecutwfc     = xx.D0,
occupations = "smearing",
smearing    = "methfessel-paxton",
degauss     = 0.05D0,
/
&ELECTRONS
/
&IONS
ion_dynamics='bfgs',
pot_extrapolation = "second_order",
wfc_extrapolation = "second_order",
/
ATOMIC_SPECIES
Mg  24.305  Mg.pw91-np-van.UPF
ATOMIC_POSITIONS {angstrom}
Mg   2.099538   2.099538   0.000000
Mg   1.049769   1.049769   2.718708   0 0 0
Mg   2.099538   2.099538   5.437416   0 0 0
Mg   1.049769   1.049769   8.156124   0 0 0
Mg   2.099538   2.099538  10.874832   0 0 0
Mg   1.049769   1.049769  13.59354    0 0 0
Mg   2.099538   2.099538  16.312248
K_POINTS {automatic}
nk1 nk2 nk3 0 0 0
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20071012/56fd0019/attachment.html>