[Pw_forum] Extracting the kinetic energy from the one-electron contribution

[Lorenzo Paulatto]

On Mon, May 18, 2009 11:56, Passerone, Daniele wrote:
>  Dear users/developers
>
>  I have a very trivial question, but unfortunately I can not find an
>  immediate solution.
>  How to isolate and print the kinetic energy contribution to the
>  one-electron energy (that is printed at the end  of a scf run?).

I'm afraid you question may not be so trivial, the one-electron term (band
energy) is the sm of the band eigenvalues multiplied by the band occupancy
and integrated over the brillouin zone. The problem is that at each kpoint
an hamiltonian (depending on k) is contructed and diagonalized, hence the
kinetic term is mixed with the external KS potential.

Furthermore, the eigenvalues of the crystal hamiltonian are not
eigenstates of the kinetic energy operator, what you can get is the mean
value
  <psi_i|K|psi_j> .

It does not mean that it is impossible to obtain the kinetic contribution,
as it can always be computed from the wavefunctions. Actually, the
meta-gga extension to density functionals uses kinetic energy, expressing
it in terms of kinetic energy density. With a bit of fiddling you can
probably activate the calculation of kinetic energy density without all
the other meta-gga stuff; if you grep dft_is_meta through the code you
should get some hints.

Maybe, if you let us know what you need, instead of what you want, we can
find a simpler solution. But if you are really interested in kinetic
energy I cannot just tell you "print ekin in kinetic.f90".

Best regards, and good luck!



-- 
Lorenzo Paulatto
SISSA  &  DEMOCRITOS (Trieste)
phone: +39 040 3787 511
skype: paulatz
www:   http://people.sissa.it/~paulatto/



----------------------------------------------------------------
  SISSA Webmail https://webmail.sissa.it/
  Powered by SquirrelMail http://www.squirrelmail.org/