[Pw_forum] surface relaxation and non-zero parallel forces.

Gisela Bocan gbocan at gmail.com
Sat Jun 11 21:15:14 CEST 2011


Dear Stefano,
                     Thanks very much for your reply and suggestions. I will
try and check how the forces react to changing degauss (and kpoints). As I
mentioned in a previous reply, I managed to solve the problem by messing
with mixing_beta and mixing_mode. Those tests were done with kpts=10x7x1,
ecutrho=10*ecutwfc and degauss=0.002. The problem persisted for
ecutwfc=45Ry.. but I think ecutwfc=35 Ry is more than enough for me. I might
be getting too picky... Anyway.. thanks again!

Dr. G.A. Bocan
Surface Physics Group
Centro Atómico Bariloche


On Sat, Jun 11, 2011 at 5:03 AM, Stefano de Gironcoli <degironc at sissa.it>wrote:

>  Dear Gisela Bocan,
>
>    the calculated parallel forces are very small. I don't expect any
> significant in-plane relaxation.    perpendicular forces are significant and
> you need to relax the system.
>    you should check converge w.r.t. number of surface layers and vacuum
> size.
>    I would do this test in the 1x1 surface rather than in the 2x2
> .
>    I think your smearing parameter is also very small: 2mRy=27meV which
> would be room temperature for Fermi-Dirac, with "mv" you should be able to
> use significantly larger smearing with little or no effect.  Such a small
> smearing forces you to use a very dense k-point sampling, and it might not
> be enough. This may be the origin of erratic behaviour.
>    I would try some degauss in the 0.01 - 0.03 range and test convergence
> w.r.t. k-points In the bulk... and then take equivalent (ore close to
> equivalent) sets  in the surface case
>
> stefano
> -
> stefano de Gironcoli - SISSA and DEMOCRITOS
>
>
> On 06/11/2011 04:35 AM, Gisela Bocan wrote:
>
> Hello,
>          I assumed surface relaxation for the Ag(110) surface would be an
> easy matter. As it happens, I find non-zero forces parallel to the surface
> and therefore, upon relaxation, I get a non-neglectable parallel
> displacement of the ions. I am working with a 2x2sqrt(2) unit cell. I attach
> my file.in and the forces I get. Note that the forces along one direction
> (x) are much larger that along the other one (y). Note as well the large
> values considered for ecutwfc, ecutrho, kpoints and the small degauss and
> conv_thr. Any ideas on how to correct this behaviour? Am I missing
> something? Thanks in advance and sorry if this is a stupid question. Small
> changes in the input parameters can lower the forces I get but I find
> nothing systematic.
>
> &control
>    prefix        = 't_G0',
>    outdir        = ' ',
>    pseudo_dir    = '/home/qe/pseudo',
>    calculation   = 'scf',
>    tprnfor       = .true.,
>    lkpoint_dir   = .false.,
>    verbosity     = 'high',
> /
> &system
>    ibrav       = 8,
>    celldm(1)   = 11.1298607358762567,
>    celldm(2)   = 1.4142135623730949,
>    celldm(3)   = 2.25,
>    nat         = 16,
>    ntyp        = 1,
>    ecutwfc     = 45,
>    ecutrho     = 450.00,
>    occupations = 'smearing',
>    smearing    = 'mv',
>    degauss     = 0.002,
> /
> &electrons
>    conv_thr = 1.d-12
>    diagonalization = 'cg'
> /
> &ions
> /
> ATOMIC_SPECIES
>    Ag   107.8682  Ag.pbe-d-rrkjus.UPF
> ATOMIC_POSITIONS  crystal
>    Ag   0.0000000000000000   0.0000000000000000   0.3333333333333333
>    Ag   0.5000000000000000   0.0000000000000000   0.3333333333333333
>    Ag   0.0000000000000000   0.5000000000000000   0.3333333333333333
>    Ag   0.5000000000000000   0.5000000000000000   0.3333333333333333
>    Ag   0.2500000000000000   0.2500000000000000   0.2222222222222222
>    Ag   0.7500000000000000   0.2500000000000000   0.2222222222222222
>    Ag   0.2500000000000000   0.7500000000000000   0.2222222222222222
>    Ag   0.7500000000000000   0.7500000000000000   0.2222222222222222
>    Ag   0.0000000000000000   0.0000000000000000   0.1111111111111111
>    Ag   0.5000000000000000   0.0000000000000000   0.1111111111111111
>    Ag   0.0000000000000000   0.5000000000000000   0.1111111111111111
>    Ag   0.5000000000000000   0.5000000000000000   0.1111111111111111
>    Ag   0.2500000000000000   0.2500000000000000   0.0000000000000000
>    Ag   0.7500000000000000   0.2500000000000000   0.0000000000000000
>    Ag   0.2500000000000000   0.7500000000000000   0.0000000000000000
>    Ag   0.7500000000000000   0.7500000000000000   0.0000000000000000
> K_POINTS automatic
>    33 23 1  0 0 0
>
>      atom   1 type  1   force =     0.00000000    0.00000000   -0.01175276
>      atom   2 type  1   force =     0.00000000    0.00000000   -0.01175282
>      atom   3 type  1   force =     0.00000000    0.00000000   -0.01175318
>      atom   4 type  1   force =     0.00000000    0.00000000   -0.01175325
>      atom   5 type  1   force =    -0.00004221    0.00000033    0.01048323
>      atom   6 type  1   force =     0.00004221    0.00000033    0.01048323
>      atom   7 type  1   force =    -0.00004221   -0.00000033    0.01048323
>      atom   8 type  1   force =     0.00004221   -0.00000033    0.01048323
>      atom   9 type  1   force =     0.00000000    0.00000000   -0.01048309
>      atom  10 type  1   force =     0.00000000    0.00000000   -0.01048306
>      atom  11 type  1   force =     0.00000000    0.00000000   -0.01048272
>      atom  12 type  1   force =     0.00000000    0.00000000   -0.01048269
>      atom  13 type  1   force =    -0.00004148    0.00000003    0.01175266
>      atom  14 type  1   force =     0.00004148    0.00000003    0.01175266
>      atom  15 type  1   force =    -0.00004148   -0.00000003    0.01175266
>      atom  16 type  1   force =     0.00004148   -0.00000003    0.01175266
>
> Dr. G. A. Bocan
> Surface Physics Group,
> Centro Atómico Bariloche,
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.orghttp://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110611/02baafed/attachment.html>


More information about the users mailing list