## full GW+BSE calculation

Run-time issues concerning Yambo that are not covered in the above forums.

Moderators: Daniele Varsano, andrea marini, Conor Hogan, myrta gruning

### Re: full GW+BSE calculation

Dear Khaled,
yes such a huge matrix is a very time consuming for a direct diagonalization. You can use iterative procedure (Haydock) by setting:
Code: Select all
 BSSmod= "h"

Next you can try to reduce the dimension of the matrix considering less band in order to verify if really you need such huge matrix.
Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Daniele Varsano

Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

### Re: full GW+BSE calculation

Dear Dr. Daniele Varsano,

I would like to thank you for your great help and your effort. I have some questions, I hope to answer them one by one. sorry for that Dr. Daniele Varsano.

1) I tried to reduce the paramter BSEBands until I set 6 bands up and 6 bands down the fermi level in this parameter, the calculation went well. I am little bit worry about the accuracy of my result according to that parameter(I think that is few stats). Is that true?

2)The phase of the material I am dealing with occurs at a certain temperature, How can I change the temperature in both GW and BSE stage?

I see two different parameters and I feel confused. should I change ElecTemp or BoseTemp and what is the difference ?.

3) Is there an option to do my calculation with adding different values of external electric field because I am going to study, if there is an effect on the band gap with change the external electric field and study it?

4)I am concerned with studying the excitons in my work. First of all I have selected the exciton which have the highest oscilator strength via ypp -e s and then I have drawed exciton amplitud with respect to energy for the same exciton I am interested in, I found strange behaviour, the exciton energy from ypp -e s is 0.78 ev but when I draw Amplitude versus E for this exciton I found maximum peak at different energy (in more details max- peak correspondes to 0.45 which is my DFT band gap in this graph) and at the exciton energy itself the curve goes to zero. Can you clarify this point to me please?

5)Adding to this point. the second file of the results come from ypp -e a, I meant weight file. The last coulum which is titled by Energy, for which exactly ?. The values of this coulum are DFT band gap. If for DFT gap, why do you include it?

I am waiting Dr. Daniele Varsano for your avaluable suggestions. Thank you in advance.

Best regards,

Khaled.
Khaled Shehata,
PhD Student,
School of Physics,
Witwatersrand University
Khaled

Posts: 21
Joined: Fri Apr 10, 2015 10:11 am
Location: South Africa

### Re: full GW+BSE calculation

Dear Khaled,
1) Well, few or not this has to be checked, increase to 7+7 and see how much your spectra changes, if more or less than the accuracy you need.
2) Caution!!!! ElecTemp and BoseTemp are variable controlling the electronic temperature, or phonons, ie they do affect the occupation number of your electronic structure, they do not have nothing to do with a phase change in your material. In order to study the optical response of different phases you need to provide the ground state of the different phases. Remember, with yambo, you are doing perturbation theory.
3) As before, you are doing perturbation theory, you need to evaluate your ground state under the effect of an external electronic field (if it exist!) and then use yambo to calculate GW corrections or BSE spectra. So, have a look to the option of your ground state engine (QE or Abinit, do not know which one you are using).
4) Yes, the amplitude are given in terms of KS transitions, ie they tell you which and how much do the KS transitions do participate in the building up of the exciton.
5) As before, you have the weight of a certain transition calculated at KS level, the energy as you say it is not so crucial in the analysis of the exciton but gives you the energy range of the transition participating in the exciton formation, in some cases can be useful.

Best,

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Daniele Varsano

Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

### Re: full GW+BSE calculation

Dear Dr. Daniele Varsano,

I hope you are fine Dr. Daniele. BSE and GW calculation went very well with selecting k-point at 4 with offset 0 . but by increasing k-point to 5 with offset 1(selecting offset in order to catch the k-point at which the band gap is set), the GW calculation stoped at the steps [07] Dynamic Dielectric Matrix (PPA).Q I have tried to modify the parameters FFTGvecs, EXXRLvcs and QPkrangea, especially, the last one, but I cant decrease further in order to do BSE calculation. I have enclosed with this forum the required files. Please, have a look, I am waiting for your valuable suggestions.

Best regards,

Khaled.
You do not have the required permissions to view the files attached to this post.
Khaled Shehata,
PhD Student,
School of Physics,
Witwatersrand University
Khaled

Posts: 21
Joined: Fri Apr 10, 2015 10:11 am
Location: South Africa

### Re: full GW+BSE calculation

Dear Khaled,
yes ut stops at the ppa level, probably because of memory reason.
Variable controlling the ppa convergence and memory usage are BndsRnXp and NGsBlkXp which is already a small number.
You can try to reduce the number of bands in BndsRnXp and see if at least it runs. Next you will need to check the convergence.
You can also think to switch to yambo 4.0 as the new release has a better control on the memory as it is possible to distribute the load.
Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Daniele Varsano

Posts: 1985
Joined: Tue Mar 17, 2009 2:23 pm

### Re: full GW+BSE calculation

Dear Dr. Daniele Varsano,

Thanks a lot for your help, it works now. I am currently doing some calculation related to the light-matter interaction, and I have found in yambo 4.0.2 revision 90 "Reflectivity # [BSS] Compute reflectivity at normal incidence" flag. I read about Reflectivity at a normal incidence and I found the governed equation is

\frac{E_{ref}}{E_{inc}}=[\frac{n_{m}-n_{inc}}{n_{m}+n_{inc}}]^{2},

where E_{ref} and E_{inc} are reflected and incident electric field respectivly, n_{m} and n_{inc} are the refractiv index of light through incident and refractive medium. Is this equation right ?.
I am looking to calculate the medium refractive index. I got the calculation result of the reflectivity, but I am getting confussed somehow
Code: Select all
#  E/ev[1]    refl /Refl refl0/Refl refl/Refl#  0.000000   0.344289   0.325416   0.344289   0.01202    0.34429    0.32542    0.34429`

let me formulate what I want to understand in questions
1) what is the title of every columes mean?
2)How can I get the refractive index of the medium from this result ?
3) can you provide me with any refrence or paper shows that the way of calculating reflectivity in yambo?

Best regards,
Khaled Shehata,
PhD Student,
School of Physics,
Witwatersrand University
Khaled

Posts: 21
Joined: Fri Apr 10, 2015 10:11 am
Location: South Africa

### Re: full GW+BSE calculation

Dear Khaled,
your equation for the reflectivity seems correct. In this case one of the two medium is the air (thus n=1) and the other (n_bulk) the one of the simulation (yambo assumes you are computing the dielectric function of a bulk material). Thus it should be:
R=|(n_bulk-1)/(n_bulk+1)|^2

1) Then the three columns refers to the reflectivity computed at (i) the BSE level, (ii) the IP-RPA level, and (iii) the BSE level from the n-1 iteration of the Haydock solver
2) You cannot extract the refractive index from the reflectivity, but you can from the complex dielectric function. n=sqrt(epsilon)
3) There is nothing published yet

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
http://www.ism.cnr.it/en/davide-sangalli-cv/
http://www.max-centre.eu/

Davide Sangalli

Posts: 314
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy

### Re: full GW+BSE calculation

Dear developers,
Happy new year. I wish this year will be full of happiness and success. While I am trying to check how the system dependent parameters in BSE calculation stage change the exciton binding energy for the bound exciton (I meant that the exciton that have energy below the band gap "1st peak"), I have found with increasing occupied and unoccupied bands (especially, unoccupied bands) in BSEBands variable the peaks are shifted to the higher energy. The thing that I got confused about, the first peak moved after the Band gap value "Eg=1.457 eV" and that means no bounded exciton, Although I am working with Semiconductor. I thought, I will get the converged
value of BSEBands variable and the corresponding value of bounded exciton energy as well. Can you help me in clarifying this point, may be I am wrong?
Let me give you some information, to allow you analyze this point, first, I am using yambo 4.0.2 Revision 90, I have 80 occupied bands, and the structure is a semiconductor. I have enclosed with this post the figure that includes the absorption spectrum at different value of BSBands, and the report file. I am waiting for your valuable suggestion. Thank you in advance.

Best regards,
Khaled.
You do not have the required permissions to view the files attached to this post.
Khaled Shehata,
PhD Student,
School of Physics,
Witwatersrand University
Khaled

Posts: 21
Joined: Fri Apr 10, 2015 10:11 am
Location: South Africa

### Re: full GW+BSE calculation

Dear Khaled,

Best wishes for the New Year to you too.

Regarding your question, if you plot the corresponding independent particle spectrum (column 4 of the same file that you used for the BS spectra plots) is the onset at lower or higher energies?
I am asking that since you are looking at a system with spin-orbit, and as the dipole operator does not flip the spin, it may be that the transition corresponding to the minimum band gap (1.457eV) is not allowed. You can see that by plotting the independent particle spectrum and see what is the onset.
This is something to check, of course there can be other explanations for what you observed.

Let us know.
Best,
Myrta
Dr Myrta Grüning
Atomistic Simulation Centre
Queen's University Belfast - Northern Ireland

http://www.researcherid.com/rid/B-1515-2009

myrta gruning

Posts: 303
Joined: Tue Mar 17, 2009 11:38 am

### Re: full GW+BSE calculation

Dear developers,

I would like to thank you for making such a very good code. I need your help in the following :

1) Is Yambo able to give me the complex value of exciton energy (Real and Imaginary part), if not, Is there any way to extract it from the code ?.

2) (Important for my research). I need to extract the transition dipole matrix , which is located in the numerator of the Macroscopic dielectric function $| \sum_{vck} \langle vq-k | e^{-iqr} | ck \rangle A_{\lambda} |^{2}$. I had a look in K_diagonalization.F file in yambo files directory, I found this matrix has a name (BS_R(k)) . Please can you help or guide me, how can I extract this part from yambo. Thank you in advance.

Best regards,
Khaled Shehata,
PhD Student,
School of Physics,
Witwatersrand University
Khaled

Posts: 21
Joined: Fri Apr 10, 2015 10:11 am
Location: South Africa

PreviousNext