Unfolding dielectric function in Coulomb matrix element calculation?

[Franz Fischer]

Dear developers,

the QE wave functions and the inverse dielectric function - computed within yambo - are saved on the irreducible part of the Brillouin zone.
When computing Coulomb and Exchange matrix elements you can certainly drop out of the irreducible wedge and I am wondering how this is taken care of by yambo.

1) Are you unfolding the two quantities to the full first Brillouin zone? Is there a smarter way, because I feel that will take a lot of memory if you are on dense k-grids. Maybe you could guide me to the relevant source code files so I can have a look by myself to see how you compute these matrix elements.

2) Are you including k-point weights in order to compensate for using the irreducible wedge. And if so how and where in the code can I find it?

3) I uploaded a picture writing out the Coulomb kernel for plane-waves. Did you implement it in the code the same way as the last line of the equation? Where can I see this in the code?


Best,
Franz

[Davide Sangalli]

Dear Franz,
in short

a) The WFs are expanded on the fly from the IBZ to the BZ when needed, without storing them in memory.
If instead we can use weights and avoid the expansion we do so.

b) The screening is never expanded. When needed we move back the symmetry operations to the WFs.
For a definition of the correlation part of the kernel see here (slide 17):
https://www.yambo-code.eu/wiki/index.ph ... heatsheets

Best,
D.