In my opinion, I think the approach you chose depends strictly from the 
question you want to answer and after that from the the size (number electrons, 
Number plane wave thou) you have. 
Do you need a full excitonic optical spectrum ? or just a RPA spectrum is 
sufficient for your study? 

If it is silicon-like and then maybe not too much computationally demanding, I 
think you can use a linear response TDDFT as a first step and then move to a 
more sophisticated approach as GW (to calculate not the optic, but the 
screening of your material) + Bethe-Salpeter equation BSE to have correctly 
acces to the excitonic effects in your absorption spectra. 

Of course all the calculations have to be converged wrt kpoints, number of 
Gvectors, number of bands. When increasing these parameters wrt their pristine 
value for your converged density in the DFT framework the computational cost 
will increase not linearly. 

I would suggest you to have a look and to do the tutorial with the Yambo code 
on a small system as they proposed in order to correctly understand what are 
you doing. 

Moreover there is the theory to have a look at in order to understand the 
meaning and the effect of the quantities you are calculating . 
I would reccomend especially 
Onida Reining Rubio Rev Mod Phys 2002 
as all the others references in the yambo website as well. 

Moreover, since I notice you are in Bilkent UNiv, try to ask to Seymour 
Cahangirov from the same univ. I think he can address you properly. 

I hope this could help a bit. 

With the best 

Federico IORI 
Marie Curie Fellow 
Laboratoire de Physique des Solides 
Bâtiment 510 - Rue André Rivière 
91400 Orsay 

----- Mensaje original -----

De: "Muhammad Adnan Saqlain" <saql...@bilkent.edu.tr> 
Para: "pw forum" <pw_forum@pwscf.org> 
Enviados: Sábado, 17 de Septiembre 2016 13:25:12 
Asunto: [Pw_forum] calculation of absorption spectrum 

Dear users 

I have a system consisting of 94 atoms (periodic with cell lengths of 
16.5, 8.5 and 8.7). I want to calculate its absorption spectrum. Can 
anyone guide me which technique would be easy to run. Actually, we have 
16 core machine and we can run a job for a maximum of 15 days only. 
Based upon this time limit, I would like to know which technique would 
be suitable. TDDFT or GW? 

any help in this regards would be highly appreciated. 



Bilknet U. Ankara 

Pw_forum mailing list 

Pw_forum mailing list

Reply via email to