Open Shell Systems

[Daniele Varsano]

Dear Vito,
it is documented in the yambo page, see here.
Did you use the flag wf_collect= .true. in your nscf calculation? The data-file.xml it is not useful, better to post the input file.
Best,
Daniele

[vitoversace]

Dear Daniele,

wf_collect= .true.

That solved the problem. Thanks a lot.

I have some Questions :
1) How does yambo calculate exciton amplitudes ? are exciton amplitudes the eigen vectors of bse matrix ? what is the physical meaning of these amplitudes ? How does this energy dependency come into exciton amplitudes ?
2) What does this output mean :

Code: Select all

 
 <01s> [05.01] Excitonic Amplitude
 <01s> Processing 12 states
 <01s> [05.02] Reporting amplitude and weights
 <02s> :: State 3 Merged with states 3 -> 5
 <02s> :: State 7 Merged with states 7 -> 8
 <02s> :: State 9 Merged with states 9 -> 10
 <03s> :: State 13 Merged with states 13 -> 14
 <03s> :: State 15 Merged with states 15 -> 16

3) The oscillator strengths of different excitation energies

#E [ev] Strength Index
7.985 0.1337E-3 1.000
8.001 0.8257E-4 2.000
8.24455 0.04137 3.000

have been calculated for any excited state from the ground state of the system ? Right?
4) As much as i know an excitonic wave function is a coupled superposition of quasi hole and quasi electron states. And one gets this coupling directly from diagonalization of bse matrix. I hope, this does the code, too.

Bests
Vito

[Daniele Varsano]

Dear Vito,
wf_collect=.true. solve the problem: of course, it is mandatory as written in the documentation.
1) Yes. It gives you the weight a KS transition participate in the excitation. The energy there is just the KS energy transition.
2) It means that you have different excitations very closed in energy, the code consider them "degenerate" and merge them. I usually do not like to have merged states, and yo can avoid it by reducing Degen_Step variable.
3) Yes
4) Yes, it is written in the Yambo paper.

Best,
Daniele

[vitoversace]

Dear Daniele,

many many thanks.
Does yambo do a symmetry analyse of the excited states of molecules ? Just to know which excited state of a molecule belongs to which irreducible representation !!!

Bests
Vito

[Daniele Varsano]

Dear Vito,
No, it does not. You can anyway look the weight of each KS transition. Next looking at the KS orbitals participating you can deduce the symmetry (if any).

Best,
Daniele

[vitoversace]

Dear Daniele,

I am sorry for asking it again.

1) Could you please explain it in more detail how to extract irreducible representation of a molecule's excited state from the weight of each KS transition and looking at the KS orbitals participating??? I really have no idea how to do it.

2) Up to how many atoms in a super cell approximation is yambo capable of calculating ? Is there any limitation to the number of atoms in yambo ?
3) Is there a way to get oscillator strength components, too (xx yy zz xy xz yz) ? Is this oscillator strength in the output averaged ?

4) What about the non-equilibrium green's function theory for calculating photoluminescence spectra? Is this going to be implemented in yambo in near future ?

Bests
Vito

[Daniele Varsano]

Dear Vito,
1) If your system has some symmetries that can be strictly defined excited states as B1u, E1u etc., these can be deduced from the symmetries of the KS states participating on that excitations. For instances in polyene chains you have Homo-->Lumo to be Bu and H-1-> Lumo and H--> Lumo+1 to form and A excited states. The weights in the post processing tells you which orbitals transition are participating in the excited state.
2) Limitations depends on you computational power. Some parts of the code are memory consuming so I would say that the order of magnitude is hundreds (and not thousands).
3) No!!, careful, the variable BLongDir, LongDrXd etc. gives you the direction of the q-->0 limit, so you have the xx,yy,zz, according to that variable. For instances if you have a molecule (without symmetry) and you want to calculate isotropic average in the absorption you need to perform 3 calculations:
100
010
001
and average them (in this way the results does not depend on how the molecule is oriented in space).
About non-diagonal component, Davide implemented something for the calculations of Kerr effect, may be he can add something here.
4) Yes, there is a lot of effort on that right now, but all the features are experimental and not released.

Best,
Daniele

[Davide Sangalli]

About non-diagonal component, Davide implemented something for the calculations of Kerr effect, may be he can add something here.

Dear Vito,
with yambo_kerr (make yambo_kerr) you should be able to compute also the off diagonal components of the dielectri tensor. Usually these are much smaller than the diagonal ones.

You can have a look to the tutorial.
http://www.yambo-code.org/tutorials/KER ... /index.php

Beste regards,
Davide

[vitoversace]

Dear Daniele and Davide,
Thank you very much.

1) Okay, i see. I was just wondering why i do not have in the weight output file any information about which KS states are participating in that specific excitonic state with which weights, as you can see it in my attached file.
But the output file is calculated for solid. I still did not test it for a molecule, like benzene. But i will very soon. Maybe then i get the full information.

Assume i have this information :

Code: Select all

Band_V     Band_C     K  ibz     Symm.      Weight     Energy

so the irreducible representation of the excitonic state will be then equal to the irreducible representation of that specific unoccupied KS state (here Band_C ) with the most contribution ( with the largest weight ) ? Right?

2) I am running the code on a cluster, and maybe later on supercomputers in Jülich. computational power and memory will not be a big issue for me.
I was here thinking of a molecule consisting of 400 atoms in big a super cell. Hopefully yambo will not crash.

3) So if i want to calculate these xz, xy, yz components of strength oscillator i just adjust it like this : 101, 110, 011 for each calculation. Right?

4) Can you tell me when we can have these non-equilibrium green's function features for photoluminescence spectra in Devel version ? in 6 month, one year or later ???

5) Is there any possibility to get solvent effects calculated within the frame work of MBPT ?

6)

When you want to use coulomb cutoff technique there is a relation between the size of your system, cutoff range and supercell size.

Can you please tell me this relation for a cubic box of a system of size (say) 18 atoms (naphthalene molecule)?
The keyword for coulomb cutoff technique in yambo 3.3.0 and generally is CUTCol_test, right ?
7) Thank you very much Davide for the hint.

With best wishes
Vito

[Daniele Varsano]

Dear Vito,

1) Because you do not have a single k-point and transition with a weight more than 5%, as stated in the output
2) 400 atoms is a lot, just try and check if you can reach converged results
3) No, in this way you are orienting the e-field in that direction, and you won't get the off-diagonal term, please see the yambo_kerr tutorial as said by Davide to get off-diagonal term
4) We cannot do any prevision, surely not before one year
5) Solvent effects: it is possible only in a very approximate way, considering the solvent as point charges, calculating the ground state in QM/MM framework using the cpmd code. Next yambo is interfaced with cpmd. If you plan to do that, you need to know few things that are not documented.
6) The cutoff coulomb parameter is extensively discussed in the forum: there are different threads. Have also a look to this paper:Phys. Rev. B 73, 205119 (2006).
CUTCol_test is just an optional flag used for testing purpose: the only effect is to give a real space representation of the interaction in the ouptut. The relevant keywords are documented here:
http://www.yambo-code.org/input_file/ya ... rimcut.php

Best,
Daniele