[QE-users] Re2: pw.x Bands Calculation not Converging for Ba4Sn9Bi3

[Lucas Nicolás Lodeiro Moraga]

Hi,

As Karpati says, you have to ensure that the parameters like ecutwfc,
number of kpoints and smearing are ok, with a good system behavior and
convergence, for the scf calculation. (A smaller degauss, 0.01 or less, and
higher ecutwfc, 45, maybe)
If your calculation is converging in the scf, you do not need to increase
more the number of bands, the default increasing due to smearing it is
fine.
Your problem in the nscf bands calculation is probably because you are
using a very high convergence threshold.  conv_thr in scf calculation its a
total energy like criterion... but for nscf calculation it defines
indirectly, the diago_thr_init, which is the convergence criteria for
eigenvalues.

For non-scf calculations: default is (conv_thr
<https://www.quantum-espresso.org/Doc/INPUT_PW.html#conv_thr>/N
elec)/10


With your input this values is 10⁻¹².... too low. For bands calculation it
is not necessary a diago_thr_init lower than 10⁻ ⁵, if you need only the
band energies. If you use a conv_thr=1e-2 or 1e-3, the problem should
disappear.

Regards!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200725/f1c9720d/attachment.html>