Yambo Community Forum

Hello,

I'm interested in study exciton fine structure on a specific system, which requires a very very dense grid. I would like to know if there is some way to use such dense grid just on a specific area of the absorption spectra, since the excitonic feature is present on the low energy area of the spectrum. Is it possible to do it someway? Maybe also using wannier interpolation?

Best regards,

Dear Bruno

at present there is not a tool to generate k-points around a give k-point in Yambo.
In general if you need many k-points consider that you can solve big BSE equations using Haydock for the spectra
and Slepc to get exciton wave-function for the lowest excitons.

This allows you to solve BSE with a size up to 200.000 or more.

Then there is the possibility to use inversion and the double-grid

http://www.attaccalite.com/speed-up-die ... ith-yambo/

but at present this approach has different limitation.
If the size of your BSE is not too large (no too many bands),
then you can use a double k-points grid with a huge number of points to converge the spectra.
However since this approach use Lapack Inversion, if you initial BSE is too large it can fail.... try to test it.

We are working to more advanced approaches to converge BSE with large k-point sampling, that will be available soon

best
Claudio

Dear Claudio,

Thank you for the fast reply. Both are very interesting options indeed, I'll take a look. Thank you for your work also!

Best regards,