[Pw_forum] Relaxation doesn't converge

[Pablo Aguado]

Well, conv_thr is the convergence criterion for electronic
self-consistency and seemingly my problem is with dynamics
convergence. In fact the calculation achieves self-consistency in
every dynamics step, so setting conv_thr = 8 shouldn't change
anything. As I pointed out in my first mail, my problem is that all
convergence criteria are satisfied but calculation doesn't stop.

Regards,

Pablo


On Wed, Sep 2, 2009 at 10:26 AM, 程迎春<yccheng.nju at gmail.com> wrote:
> In my opinion, the stricter the criterion, the more ecut and more K points
> sampling.
>
> 2009/9/2 Huiqun Zhou <hqzhou at nju.edu.cn>
>>
>> Pablo,
>>
>> Usually, setting conv_thr = 1d-8 is sufficient, please make sure if you
>> need such strict convergence criterion.
>>
>>
>> Huiqun Zhou
>> @Earth Sciences, Nanjing University, China
>>
>>
>> ----- Original Message -----
>> From: "Pablo Aguado" <paguado at gmail.com>
>> To: <pw_forum at pwscf.org>
>> Sent: Tuesday, September 01, 2009 6:38 PM
>> Subject: [Pw_forum] Relaxation doesn't converge
>>
>>
>> > Dear all,
>> >
>> > I'm doing some tests on the structural optimization with pw. I'm
>> > currently testing with the tetragonal structure of PbTiO3, using the
>> > following input:
>> >
>> > &control
>> >    calculation  = 'vc-relax'
>> >    restart_mode = 'restart'
>> >    pseudo_dir   = '/****/espresso-4.1/pseudo/'
>> >    outdir       = '/***/tmp'
>> >    forc_conv_thr = 1d-3
>> >    nstep = 50
>> > /
>> > &system
>> >    ibrav=6
>> >    celldm(1)=7.3699
>> >    celldm(3)=1.04
>> >    nat=5
>> >    ntyp=3
>> >    nbnd=28
>> >    ecutwfc=30.0
>> >    occupations = 'fixed'
>> > /
>> > &electrons
>> >    conv_thr = 1d-12,
>> >    mixing_beta=0.4,
>> > /
>> > &ions
>> >    ion_dynamics = 'bfgs'
>> >    bfgs_ndim = 5
>> > /
>> > &cell
>> >    cell_dynamics = 'damp-pr'
>> >    press_conv_thr = 1.0d0
>> >    cell_dofree = 'z'
>> > /
>> > ATOMIC_SPECIES
>> >  Pb   207.2      Pb.pz-d-van.UPF
>> >  Ti    47.867    Ti.pz-sp-van_ak.UPF
>> >  O     15.9994   O.pz-van_ak.UPF
>> > ATOMIC_POSITIONS crystal
>> >  Pb    0.000    0.000    0.040   0 0 1
>> >  Ti    0.500    0.500    0.520   0 0 1
>> >  O     0.000    0.500    0.500   0 0 1
>> >  O     0.500    0.500    0.000   0 0 1
>> >  O     0.500    0.000    0.500   0 0 1
>> > K_POINTS automatic
>> >  6 6 6 1 1 1
>> >
>> > The relaxation is taking forever so I've checked the output and I've
>> > found several steps where everything seems to be converged (notice I'm
>> > only relaxing out-of-plane lattice vector) but the code keeps running.
>> > An example:
>> >
>> > - Energy difference = 4e-7
>> > - Forces acting on atoms (Ry/au):
>> >
>> >     atom   1 type  1   force =     0.00000000    0.00000000
>> >  0.00085679
>> >     atom   2 type  2   force =     0.00000000    0.00000000
>> > -0.00001779
>> >     atom   3 type  3   force =     0.00000000    0.00000000
>> > -0.00024617
>> >     atom   4 type  3   force =     0.00000000    0.00000000
>> > -0.00034665
>> >     atom   5 type  3   force =     0.00000000    0.00000000
>> > -0.00024617
>> >
>> >     Total force =     0.000988     Total SCF correction =     0.000002
>> >
>> >          total   stress  (Ry/bohr**3)                   (kbar)
>> >   P=   -0.84
>> >  -0.00001150   0.00000000   0.00000000         -1.69      0.00      0.00
>> >   0.00000000  -0.00001150   0.00000000          0.00     -1.69      0.00
>> >   0.00000000   0.00000000   0.00000591          0.00      0.00      0.87
>> >
>> > Any idea of what's going on?
>> >
>> > Thank you very much
>> >
>> > Pablo
>> > _______________________________________________
>> > Pw_forum mailing list
>> > Pw_forum at pwscf.org
>> > http://www.democritos.it/mailman/listinfo/pw_forum
>> >
>>
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://www.democritos.it/mailman/listinfo/pw_forum
>
>
>
> --
> Y. C. Cheng
> Department of Physics
> Nanjing University
> Nanjing 210093
> P. R. China
> Tel: 86-25-83592907
> Email: yccheng.nju at gmail.com
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://www.democritos.it/mailman/listinfo/pw_forum
>
>



-- 
-----------------------------------------------------------
Pablo Aguado Puente
paguado at gmail.com
------------------------------------------------------------