[QE-users] Phonon SCF does not converge with semicore pseudopotentials
Antonio Crepaldi
antonio.crepaldi.lanza at gmail.com
Wed Feb 9 08:50:11 CET 2022
Dear QE developers and users,
I'm currently trying to compute the phonon dispersion of a MoTe2 bilayer
with QE 6.8. I'm using the stringent accuracy, FR pseudopotentials from
PseudoDojo, which contain semicore states for Te as well.
I'm using vdW-DFT as vdw-df2-b86r.
I've carefully optimized the cell and the atomic position, however the scf
in the phonon calculation does not converge. In fact, it seems to diverge
straight away:
Electric Fields Calculation
> iter # 1 total cpu time : 118.9 secs av.it.: 10.2
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 4.707E-01
> iter # 2 total cpu time : 150.1 secs av.it.: 9.8
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 1.742E+01
> iter # 3 total cpu time : 170.5 secs av.it.: 4.8
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 1.238E+01
> iter # 4 total cpu time : 190.2 secs av.it.: 4.5
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 1.133E+01
> iter # 5 total cpu time : 210.5 secs av.it.: 4.8
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 3.608E+00
> iter # 6 total cpu time : 227.5 secs av.it.: 3.3
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 4.009E+00
> iter # 7 total cpu time : 256.9 secs av.it.: 9.3
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 6.845E+01
> iter # 8 total cpu time : 277.9 secs av.it.: 5.2
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 4.813E+01
> iter # 9 total cpu time : 297.8 secs av.it.: 4.8
> thresh= 1.000E-02 alpha_mix = 0.400 |ddv_scf|^2 = 1.080E+02
I've tried to reduce the alpha_nmix to 0.4 and 0.2 but it did not help.
The same calculation, with proper parameters but with pseudopotentials
without semicore states does not give any problem.
You can download the input and output file, as well as the pseudopotentials
at the following link
https://www.dropbox.com/sh/zw2db8q8loczz02/AACmax-Dby-9JqZjnm1QW_LWa?dl=0
Do you have any suggestions?
Thank you very much for your kind help!
Best,
Antonio Crepaldi-Lanza
-- Department of Physics "Aldo Moro" University of Bari
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20220209/e8e61a96/attachment.html>
More information about the users
mailing list