The ABINIT Users Mailing List ( CLOSED )

["matthieu verstraete" <matthieu.jean.verstraete@gmail.com>]

Hi Aldo,

as you shrewdly detected the sciss is only used for the resp function part. The output band structure comes from the ground state bits of abinit, and does not depend on sciss. This might be changed to include a nice aesthetic shift, but it wouldn't affect the calculation itself. In the calculation of phonons or other perturbations, sciss comes in every time the GS eigenvalues are reused, in particular in:

abinit-5.6.x/src/16response/cgwf3.F90
abinit-5.6.x/src/16response/eig2tot.F90
abinit-5.6.x/src/16response/vtowfk3.F90

if you care to look.

Basically they intervene in the  Sternheimer equation (eq 16 of Xavier's prb 55 10337) and the variational expressions for the 2nd order energies (eq 15 of Xavier's PRB 55 10355), and that is where the sciss is added in. It works, and you have to pass the sciss on to anaddb as well for the optical properties (as I remember this is automatic through the ddb files). I think it should only be used for q=Gamma for rfelfd 3, ie phonons with LOTO splitting.

Has anyone used sciss for third order responses (Raman etc)?

Ciao

Matthieu