Yambo Community Forum

Dear Developers,

Can someone please help me with doing BSE calculation using a very dense k-grid, like 51x51x1 while keeping the DFT k-grid low (12x12x1)? There is a paper that has mentioned YAMBO for this purpose (PHYSICAL REVIEW B 88, 045412 (2013).

Wishes,
Pushpendra

Dear Pushpendra,
I may be wrong, but in the paper you mention a large grid (51x51x1) is used.
Can you explain better what you mean by keeping the DFT k-grid low (12x12x1)?

Best,
Daniele

Dear Dr. Daniele,

Maybe I misunderstood the paper mentioned above but I wanted to understand that suppose we do a ground-state calculation (scf/nscf) for k-grid 12x12x1 and use these ground state wave-functions to do the G0W0 calculation. Now if I want to calculate BSE spectra for a higher k-grid (say 51x51x1). For this hogher k-grid BSE calculation do I need to go back to DFT calculation with higher k-grid or is it possible to implement higher k-grid in YAMBO?

Thank you!

Dear Pushpendra,
yes, you need to recalculate the electronic structure for a denser grid, as you need it to build your BSE matrix.

Best,
Daniele

Thank you, Dr. Daniele, for the clarification.

pyadav wrote: ↑Thu Jan 14, 2021 12:38 pm Dear Dr. Daniele,

Maybe I misunderstood the paper mentioned above but I wanted to understand that suppose we do a ground-state calculation (scf/nscf) for k-grid 12x12x1 and use these ground state wave-functions to do the G0W0 calculation. Now if I want to calculate BSE spectra for a higher k-grid (say 51x51x1). For this hogher k-grid BSE calculation do I need to go back to DFT calculation with higher k-grid or is it possible to implement higher k-grid in YAMBO?

Thank you!

Dear Developers,

Is this calculation possible for the new version of the Yambo program?

Best regards,
Zafer

Dear Zafer,

can you be more explicit on what kind of calculation you are referring?
In the previous post, the user was asking about BSE calculations using a dense grid without calculating the ground state on such grid. This is not possible.
What is possible is to speed up the calculation using a double gri method as explained here:
http://www.attaccalite.com/speed-up-die ... ith-yambo/

or via Haydock method as described here:

https://www.frontiersin.org/articles/10 ... 63946/full


Best,
Daniele

Dear Daniele,

Thank you for replying, my question is like this: For k-grid 18x18x8, a ground state calculation of a bulk structure was made, and then GW calculation was made and ndb.QP was obtained. Now I want to calculate BSE spectra for k-grid 26x26x10. This calculation is not possible, ok.
If I want to use the double grid method, I should choose k-grid:18x18x8 for scf calc. and k-grid:26x26x10 for ncsf calc. Is it possible to calculate BSE spectra by using the ndb.QP file obtained from the previous GW calculation?

Best regards,

Zafer

Dear Zafer,

It is not possible to use the ndb.QP directly. You can use a scissor/stretching operator with parameters extracted from the previous calculation and apply them to the calculation with denser grid.

Best,
Daniele