Coulomb truncation for TDDFT

[Alesster]

Dear All,

I studied the use of Coulomb truncation here http://www.yambo-code.org/wiki/index.ph ... al_systems
I repeated the example and it works perfectly. But there is example only for the BSE Absorption spectra calculation.
The question is: Does this method work for the TDDFT calculations like RPA, ALDA ... ?

Thank you,
Sasha.

[Daniele Varsano]

Dear Sasha,
the coulomb truncation once activated substitutes the coulomb potential with its truncated version along all the code,
so it is effective also for TDDFT7RPA calculations.

Best,
Daniele

PS: when posting in the forum please fill your signature with your affiliation, you can do once for all filling the signature field in your profile.

[Alesster]

Dear Sasha,
the coulomb truncation once activated substitutes the coulomb potential with its truncated version along all the code,
so it is effective also for TDDFT7RPA calculations.

Best,
Daniele

PS: when posting in the forum please fill your signature with your affiliation, you can do once for all filling the signature field in your profile.

Thank you for your reply, Daniele.

Yes, I suggested that it should work also for TDDFT. But trying to do TDRPA calculation for Black Phosphorus (1 layer) with truncated potential I have got strange results (completely different from results without cut).
So, may be I did something wrong ...

Sasha.

[Daniele Varsano]

Dear Sasha,
there could be some problem in the definition of the absorption and eels spectra when using the cutoff coulomb potential that has to be addressed and eventually fixed.
Are you calculating it in G space or in transition space? If you are doing in G space you can try to repeat the calculation in transition space:
yambo -o b -K alda for TDDFT
yambo -o b -K hartree for TDRPA

Best,
Daniele

[Alesster]

Dear Sasha,
there could be some problem in the definition of the absorption and eels spectra when using the cutoff coulomb potential that has to be addressed and eventually fixed.
Are you calculating it in G space or in transition space? If you are doing in G space you can try to repeat the calculation in transition space:
yambo -o b -K alda for TDDFT
yambo -o b -K hartree for TDRPA

I use yambo v.3.4.2 and for doing TDDFT I use the command: yambo -o c -k alda
If I use
yambo -o b -k alda
then it creates the input for BSE. So, this is not I want.

[Daniele Varsano]

Dear Sasha,
I strongly suggest you to switch to a more recent version. Many bugs have been fixed since then and parallelization has been largely improved.
Anyway you can do alda in transition space also with 3.4. In this way all the G vector in the alda kernel are automatically included and it is much more robust as the kernel is evaluated in real space.
In order to activate it type yambo -H and see the options for the command line. Syntax in changed and now I do not have the possibility to look at it as I'm traveling connected with a mobile phone.
Finally if you do not want use the modified coulomb cutoff RPA/LDA should converge in a reasonable way with respect the volume of the supercell at difference of a GW calculation.
Best,

Daniele

[Alesster]

Dear Daniele,
Thank you very much for your explanation.

I strongly suggest you to switch to a more recent version. Many bugs have been fixed since then and parallelization has been largely improved.

For the moment I cannot use the latest versions of yambo because I use modified code for the kernel (instead of alda) that works in old version. To do the same modification in the resent version takes more efforts and time. I will do this but later. But now I can use only old worked version.

Anyway you can do alda in transition space also with 3.4.

Do you mean under transition space real space?

In this way all the G vector in the alda kernel are automatically included and it is much more robust as the kernel is evaluated in real space.

As I understand for G-space it corresponds subroutine tddft_alda_g_space.F and for real space tddft_alda_r_space.F.
I did modifications for the alda kernel in subroutine tddft_alda_g_space.F, so it is realized in G space. Will it work in real space?

In order to activate it type yambo -H and see the options for the command line.

There exists these options for v3.4:
-o <opt> :Optics [opt=(c)hi is (G)-space / (b)se is (eh)-space ]
-k <opt> :Kernel [opt=hartree/alda/lrc/hf/sex]
(hf/sex only eh-space; lrc only G-space)

So, I don't see here transition or real space.

Best,
Sasha

[Daniele Varsano]

Dear Sasha,

Do you mean under transition space real space?

Yes when calculating the response in transition space, the ALDA kenel is evaluated in real space.

I did modifications for the alda kernel in subroutine tddft_alda_g_space.F, so it is realized in G space. Will it work in real space?

I do not think so, I presume you should modify the tddft_alda_r_space.F accordingly.

So, I don't see here transition or real space.

yambo -o b -k alda

Anyway as I told you in the last post there could be problem in the definition of the epsilon when calculating it in G space in presence of the coulomb cutoff.
Actually it is possible that the o.eels output contains the right response, but I'm not sure about that as you are using an old unsupported version, just take a look if it makes sense.
Next I also suggest you to make some supercell volume convergence without using the modified coulomb cutoff just to have an idea if you can avoid to use the modified potential.

Best,
Daniele