[Pw_forum] problems with pressure calculation

[Paolo Giannozzi]

On Mon, 2013-04-01 at 14:28 -0400, Eduardo Ariel Menendez Proupin wrote:

> I have found strange differences in results for the calculation of
> pressure using pwscf in versions 4.3 and 5.0.N (N=1,2).

it looks like a nonlinear interaction between symmetry and insufficient
convergence of self-consistency. If you reduce the convergence threshold
to conv_thr=1.0e-8 or so (default is 10e-6) the jumps disappear; same if
you use all symmetries, whether their fractional translation is
commensurate or not with the FFT grid (option use_all_frac). I guess
that the discrepancy wrt previous versions has a similar origin. 

The default value for the convergence threshold is good if you are
interested in energies, reasonable if you are interested in forces, 
not so good for stress and pressure

Disclaimer: I just tried two cases

Paolo

> PAW calculation
>    Cd  112.40000  Cd.pbe-dn-kjpaw_psl.0.2.UPF 
>    Se   78.96000  Se.pbe-n-kjpaw_psl.0.2.UPF 
> 
> 
> 
> Ultrasoft PP calculation
>    Cd  112.40000  Cd.pbe-dn-rrkjus_psl.0.2.UPF 
>    Se   78.96000  Se.pbe-n-rrkjus_psl.0.2.UPF 
> 
> 
> 
> 
> For the PAW calculations, the convergence of pressure with the density
> cutoff (euctrho) shows jumps  (see the attached  figure paw-press.eps)
> and no convergence in the case of version 5.0.N, while there is
> convergence with version 4.3. For both versions the total energies are
> the same. 
> 
> The jumps are associated with differences in the number of symmetry
> operations detected. The message 
> 
>  "         (note:  6 additional sym.ops. were found but ignored
>            their fractional translations are incommensurate with FFT
> grid)"
> 
> is associated with higher pressures (33 kbar). Lower pressures ~ 26
> kbar) are found when the FFT grid is commensurate and all 12 sym. ops.
> are used. 
> In the calculations with version 4.3 the pressures are somewhat lower,
> but apparently convergent. Also, there are less symmetry operations
> detected or used than for version 5.0.N. 
> 
> 
> 
> In the ultrasoft calculations, (see figure uspp-press), both versions
> 4.3 and 5.0.N produce convergent pressures vs ecutrho, but the
> pressures differ almost by 10 kbar. (The total energies are almost
> equal, as expected). 
> 
> 
> 
> Thanks for any comment. 
> 
> 
> -- 
> Eduardo Menendez Proupin
> Departamento de Fisica, Facultad de Ciencias, Universidad de Chile 
> URL: http://fisica.ciencias.uchile.cl/~emenendez
> "Lograremos el anhelado desarrollo cuando sea la ciudadanía, y no los
> científicos, la que pida más ciencia para Chile." J. Babul
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum

-- 
 Paolo Giannozzi, Dept. Chemistry&Physics&Environment, 
 Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
 Phone +39-0432-558216, fax +39-0432-558222