[Pw_forum] Bands calculation for Silica

[Paolo Giannozzi]

On Fri, 2014-12-05 at 16:48 +0000, Francesco Pelizza wrote:

> - in the bands plot some strange "crossover" (errant behaviour) appear 
> and i do not know what it means.

it means that the algorithm that tries to align bands doesn't work well.
If you just plot bands as they come out of the calculation, they will
be ordered from lowest to highest state and you will miss the correct
crossings. The code tries to align energies along the true bands using
one of the following two methods (see the documentation of bands.x):
- by maximising band overlap between adjacent k-points
- by finding band irreps (irreducible representations)
The first one often fails, the second is clumsy.

Please also note that the highest bands in a non-scf calculation may
sometimes not be the real ones, due to the way iterative diagonalization
works. In a non-scf calculation you should compute a few more bands than
those you plan to use or plot.


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