Triplet exciton wavefunction plot

[DavidPolito93]

Dear developers,

I am studying excitonic effects on polymers and I wanted to plot the wavefunctions of both the singlet and triplet excitons.
I have used yambo-5.0 and I have performed BSE computations on top of the self consistent GW eigenvalues.

Concerning the triplet state, in order to solve the BSE equation I set BSENGexx = 0 and I truncated the Coulomb potential via the ws option. My polymer is along the x axis.
I also turned on the option WRbsWF.

The issue with the triplet state is the following: I followed the tutorial on plotting excitonic properties and more precisely

- I sorted excitons via ypp -J 2D_WR_WC -e s 1 and I found the first two peaks to be equal to states 1 and 5
- Then I computed the weights via ypp -F ypp_AMPL.in -e a 1 for both of them and I saw direct contribution at the border zone
- Then I plotted the wavefunctions of these two states placing the hole in a point where I had the electron density of valence bands associated to states 1 and 5

My problem is that the shape of such wavefunctions is identical. I would have expected the second exciton to be more extended due to the fact that the second exciton peak of the triplet is closer to the QP peak obtained via GW or at least a different shape.

I am attaching both the alpha and the wavefunctions.

Have I did something wrong or am I missing something?

Best,

Davide Romanin
-----------------------------------------------------
Post-doctoral fellow
Institut des NanoSciences de Paris
Sorbonne Université, CNRS
4 place Jussieu,
75252 PARIS Cedex 05

[Daniele Varsano]

Dear Davide,

in the calculation of the triplet there is a little problem,
BSENGexx should be set to

Code: Select all

BSENGexx= 0 eV

otherwise the default is RL and yambo wants to fill the G vector shells and as a results as you can see at the end of your output you have:

Code: Select all

BSENGexx= 1                RL 

Besides that, the procedure for the excitonic wavefucntions seems to be correct and the results sounds strange also to me.
Can you post the output of the ypp_AMPL.in for the two excitons?

next,

can you also make this simple change to the code in:

Code: Select all

ypp//modules/YPP_SET_defaults.F: l_norm_to_one=.true.

setting the variable to .false. , recompile and repeat your calculation?

Best,

Daniele

[DavidPolito93]

Dear Daniele,

I checked: before in the input I put BSENGexx= 0 Ry.

I re-did the procedure by putting BSENGexx= 0 eV. Anyway the code still change it to 1 RL as you can see in the attachments.

Davide

[DavidPolito93]

By the way,

Here there are both weights and amplitudes

Davide

[Daniele Varsano]

Dear Davide,

about the BSENGexx I think that all is safe. Yambo takes into account the component G=0, but then it is ignored int he kernel building.
About the excwf, it seems that the transition participating in the two excitons are the same, at least the most important ones, even if with different weights, this could explain the similar shape you are finding.
Yambo by default normalise the exc-wf to its maximum, and this could be the reason why you found also very similar localisation. But this needs to be verified.
Try to change the normalisation default as I indicated in the previous post and see what happen.

Best,
Daniele

[DavidPolito93]

Dear Daniele,

Ok I have modified the file you told me and then in the main folder i typed:

make -j 2 ypp

(Should I have made some clean?)

Now I have re-submitted the job and I will send you the results as soon as I have them.

By the way, I have looked at the results of the singlet and again the shape is the same as the triplet one (see attachment).
But probably is again related to the different weights, also because the states are exactly the same in both cases.

Davide

[Daniele Varsano]

Dear Davide,

I start to suspect there is something not working properly.
The weights and amplitudes of exciton 5 seems to me they are not compatible.
Can you do the calculation setting the weight threshold to zero for both excitons?

Let's have anyway a look at the difference without normalisation, but it seems to me that the problem is elsewhere.

Then, may ask you the following?
Update to the last snapshot, yambo 5.0.1, or 5.0.2?
if the problem persists, can you try to do a 2D or 3D plot e.g. xsf format.

Best,
Daniele

[DavidPolito93]

Dear Daniele,

I am using yambo-5.0.0. Should I update to the most recent one?

Here attached the amplitudes and weights setting the threshold to zero

Davide

[Daniele Varsano]

Dear Davide,

from the files you updated now at least it is clear that weights and amplitudes are compatibles, so it seems they are calculated correctly.
Have a look to the wave functions without normalisation, and I would also look at 2D/3D plots.

Best,
Daniele

[DavidPolito93]

Dear Daniele,

Here there are both the 1D and the 3D plots.

From the 1D you can see that, as you said, the weights are different.
However the shape is the same (both in 1D and in 3D)

Moreover, they are also identical in shape to the singlet.

Davide