[Pw_forum] Si diamond-cubic to beta-Sn transformation

[Paolo Giannozzi]

Daryl Chrzan wrote:

> smearing = 'gaussian',
> degauss = 0.02,

you need occupations='smearing' as well, or else the code
will silently default to fixed occupations. There should
be a check and a warning, but there isn't any. Using QE in
teaching is a great way to find all kinds of problems.
I noticed it because this statement looked weird to me:
      highest occupied, lowest unoccupied level (ev):    11.4054   5.2182
and no, the numbers were not printed in revers order. It doesn't make
any difference for diamond Si, but it does for beta-Sn Si. I got
   betatin.out:!    total energy              =     -19.20675112 Ry
   diamond.out:!    total energy              =     -19.25371607 Ry
i.e. 0.32eV/atoms

> Beta-Sn:
> nat=2

I was convinced that beta-Sn has 4 atoms per unit cell, not 2.
I am interested because I also have used QE for teaching, and
the diamond to beta-tin transition is a good exercise

Paolo
-- 
Paolo Giannozzi, Democritos and University of Udine, Italy