[QE-users] vc-relax Convergence problem with 'TS' VdW correction

Paolo Giannozzi p.giannozzi at gmail.com
Sun Oct 14 13:40:13 CEST 2018


Answering myself: it's the combination "variable cell +
Tkatchenko-Scheffler" that doesn't work properly

PG

On Sat, Oct 13, 2018 at 6:46 PM Paolo Giannozzi <p.giannozzi at gmail.com>
wrote:

> I don't think a tighter convergence threshold will fix the problem here:
> there is a mismatch between forces and energies. This calculation uses
> - Tkatchenko-Scheffler corrections
> - in a spin-polarized case
> - in a variable-cell calculation
> Assuming that there is no problem without TS, it would be helpful to know
> whether the problem is due to spin polarization, to variable cell, to both,
> or to none.
>
> PG
>
>
> On Fri, Oct 12, 2018 at 9:32 PM Nicola Marzari <nicola.marzari at epfl.ch>
> wrote:
>
>>
>> First I would try to have a much, much tighter convergence threshold
>> than conv_thr = 1.0d-6; I think we use as a standard 0.2d-9 times
>> the number of atoms - so in your case 4.0d-9, rather than
>> 1.0d-6. If your selfconsistency is poor, your forces are noisy, and the
>> relaxation slow.
>>
>> I also note that there is a variable called
>> ts_vdw_econv_thr ; not sure what it does, but seems relevant.
>>
>> nicola
>>
>>
>>
>>
>> On 12/10/2018 16:36, Hooman Yaghoobnejad Asl wrote:
>> > Dear all,
>> > I'm experiencing a problem with total force minimization when
>> > the vdw_corr = 'TS' is used. Without the vdw_corr option, force
>> > minimization completes smoothly, as shown below:
>> >
>> >       Total force =     0.144455     Total SCF correction =     0.002938
>> >       Total force =     0.096620     Total SCF correction =     0.004258
>> >       Total force =     0.023833     Total SCF correction =     0.000694
>> >       Total force =     0.013572     Total SCF correction =     0.000196
>> >       Total force =     0.013176     Total SCF correction =     0.000775
>> >       Total force =     0.013520     Total SCF correction =     0.001316
>> >       Total force =     0.013322     Total SCF correction =     0.000670
>> >       Total force =     0.013164     Total SCF correction =     0.000897
>> >       Total force =     0.013634     Total SCF correction =     0.000800
>> >       Total force =     0.016822     Total SCF correction =     0.000639
>> >       Total force =     0.015558     Total SCF correction =     0.000354
>> >       Total force =     0.012359     Total SCF correction =     0.000858
>> >       Total force =     0.009014     Total SCF correction =     0.000322
>> >       Total force =     0.007140     Total SCF correction =     0.000338
>> >       Total force =     0.006073     Total SCF correction =     0.000556
>> >       Total force =     0.005047     Total SCF correction =     0.000465
>> >       Total force =     0.004639     Total SCF correction =     0.000050
>> >       Total force =     0.003113     Total SCF correction =     0.000382
>> >       Total force =     0.002225     Total SCF correction =     0.000268
>> >       Total force =     0.001698     Total SCF correction =     0.000156
>> >       Total force =     0.001505     Total SCF correction =     0.000372
>> >       Total force =     0.001422     Total SCF correction =     0.000244
>> >       Total force =     0.001442     Total SCF correction =     0.000420
>> > However, after implementing the VdW correction, it seems that the
>> forces
>> > cannot be minimized further even after 200 ionic steps:
>> >
>> >       Total force =     0.146620     Total SCF correction =     0.003488
>> >       Total force =     0.108842     Total SCF correction =     0.004843
>> >       Total force =     0.069519     Total SCF correction =     0.000923
>> >       Total force =     0.016339     Total SCF correction =     0.002655
>> >       Total force =     0.015353     Total SCF correction =     0.001041
>> >       Total force =     0.016188     Total SCF correction =     0.000361
>> >       Total force =     0.016350     Total SCF correction =     0.000832
>> >       Total force =     0.015878     Total SCF correction =     0.001654
>> >       Total force =     0.016273     Total SCF correction =     0.000560
>> >       Total force =     0.016349     Total SCF correction =     0.000406
>> >       Total force =     0.015879     Total SCF correction =     0.001403
>> >       .
>> >       .
>> >       .
>> >       .
>> >       Total force =     0.015866     Total SCF correction =     0.001214
>> >       Total force =     0.016279     Total SCF correction =     0.000263
>> >       Total force =     0.016351     Total SCF correction =     0.000394
>> >       Total force =     0.015879     Total SCF correction =     0.000394
>> >       Total force =     0.016277     Total SCF correction =     0.001323
>> >       Total force =     0.016347     Total SCF correction =     0.000194
>> >       Total force =     0.015878     Total SCF correction =     0.000606
>> > I was wondering if anyone can make a comment on this? I'm using QE 6.2
>> > on a cluster. Input copied below:
>> >   &CONTROL
>> >                         title = 'Li0.5NiO2 NCPP FM VdW ' ,
>> >                   calculation = 'vc-relax' ,
>> >                   max_seconds = 86300 ,
>> >                !    wf_collect = .true. ,
>> >                        outdir = '/' ,
>> >                        wfcdir = '/' ,
>> >                    pseudo_dir = '/scratch/05246/pseudo/' ,
>> > !restart_mode = 'restart' ,
>> >                        prefix = 'LNO' ,
>> >                       disk_io = 'low' ,
>> >                     verbosity = 'high' ,
>> > nstep = 200 ,
>> >   /
>> >   &SYSTEM
>> >                         ibrav = 0,
>> >                     celldm(1) = 1.889726,
>> >                           nat = 21,
>> >                          ntyp = 3,
>> >                       ecutwfc = 80 ,
>> >                       ecutrho = 320 ,
>> >                  occupations = 'smearing' ,
>> >                       degauss = 0.001 ,
>> >                      smearing = 'gaussian' ,
>> >                         nspin = 2 ,
>> >     starting_magnetization(1) = 0.25,
>> >               vdw_corr = 'TS' ,
>> >   /
>> >   &ELECTRONS
>> >              electron_maxstep = 500,
>> >                      conv_thr = 1.0d-6 ,
>> > /
>> >   &IONS
>> > /
>> >   &CELL
>> > /
>> > CELL_PARAMETERS {alat}
>> >    5.790000   0.000000   0.000000
>> >   -1.447500   2.507144   0.000000
>> >    0.000000   0.000000  14.221000
>> > ATOMIC_SPECIES
>> >     Ni   58.69000   Ni_ONCV_PBE-1.0.upf
>> >      O    15.99000  O_ONCV_PBE-1.0.upf
>> >     Li    6.94000   Li_ONCV_PBE-1.0.upf
>> > ATOMIC_POSITIONS crystal
>> > O       0.000000   0.000000   0.232000
>> > O       0.000000   0.000000   0.768000
>> > O       0.333333   0.333333   0.565333
>> > O       0.333333   0.333333   0.101333
>> > O       0.166667   0.666667   0.898667
>> > O       0.166667   0.666667   0.434667
>> > Ni     0.000000   0.000000   0.500000
>> > Ni     0.333333   0.333333   0.833333
>> > Ni     0.166667   0.666667   0.166667
>> > Li      0.166667   0.666667   0.666667
>> > O       0.500000   0.000000   0.232000
>> > O       0.500000   0.000000   0.768000
>> > O       0.833333   0.333333   0.565333
>> > O       0.833333   0.333333   0.101333
>> > O       0.666667   0.666667   0.898667
>> > O       0.666667   0.666667   0.434667
>> > Ni     0.500000   0.000000   0.500000
>> > Ni     0.833333   0.333333   0.833333
>> > Ni     0.666667   0.666667   0.166667
>> > Li      0.500000   0.000000   0.000000
>> > Li      0.833333   0.333333   0.333333
>> > K_POINTS automatic
>> > 4 8 2   0 0 0
>> >
>> > Thank you
>> > P.S. I benchmarked the wf cut-off energy vs binary oxides for each
>> metal
>> > and 80 Ry gave values comparable to 320 Ry.
>> >
>> > --
>> > *Hooman Yaghoobnejad
>> > *
>> > *PhD, Department of Chemistry
>> > *
>> > *Missouri University of Science and Technology
>> > *
>> > *Rolla, MO 65409
>> > *
>> > *USA*
>> >
>> >
>> > _______________________________________________
>> > users mailing list
>> > users at lists.quantum-espresso.org
>> > https://lists.quantum-espresso.org/mailman/listinfo/users
>> >
>>
>>
>> --
>> ----------------------------------------------------------------------
>> Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
>> Director, National Centre for Competence in Research NCCR MARVEL, EPFL
>> http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project
>> _______________________________________________
>> users mailing list
>> users at lists.quantum-espresso.org
>> https://lists.quantum-espresso.org/mailman/listinfo/users
>
>
>
> --
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
>
>

-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181014/40709f81/attachment.html>


More information about the users mailing list