Some questions about using Yambo.

[lijin18]

Dear Administrators,
I am a new user of YAMBO. I have read your paper, Computer Physics Communications 180 (2009) 1392–1403. I found that yambo is a useful tool to study the quasiparticle energies and optical properties within the framework of many-body perturbation theory and TDDFT and I want to do some research with yambo and learn it now. But I encounter some problems and ask you for help:
1. QP_E: QP corrections using a scissor operator and valence/conduction stretching.
scissor | cond. stretching| val.stretching.
What's the meaning of cond. stretching and val.stretching? And how to determine the values of the three parameters?
2. In your paper, you have shown the Optical spectrum of bulk silicon. I tried to reproduce the results but I failed.
Would you please show me the input files of bulk silicon just as the example of 3D LiF (Abinit and yambo input
files)?
3. How to get the GW bandstructure of bulk silicon by yambo?
4. How to get and show the excitonic wave function and the excitation energies?
I would be grateful if you can give me some advice and some example (just as bulk silicon) about the above questions.
Lijin
Xiangtan University, Xiangtan, Hunan province, China
lijin@xtu.edu.cn

[myrta gruning]

1. QP_E: QP corrections using a scissor operator and valence/conduction stretching.
scissor | cond. stretching| val.stretching. What's the meaning of cond. stretching and val.stretching? And how to determine the values of the three parameters?

E_{QP}^{cond} = cond. stretching*E_{LDA}^{cond} + scissor
E_{QP}^{val} = val. stretching*E_{LDA}^{val}

2. In your paper, you have shown the Optical spectrum of bulk silicon. I tried to reproduce the results but I failed. Would you please show me the input files of bulk silicon just as the example of 3D LiF (Abinit and yambo input
files)?

Do you refer to the converged or the example spectrum. What is the problem exactly?
For the unconverged spectrum the instruction on the paper should be sufficient. For the converged caclculation you can download all necessary info here.

3. How to get the GW bandstructure of bulk silicon by yambo?

This issue has been already discussed in this post

4. How to get and show the excitonic wave function and the excitation energies?

To get the excitonic wave function you need to use the diagonalization to solve the BSE, (BSSmod="d") and to have the key WRbsWF. (to get the corresponding input to edit use yambo -b -o b -y d -V resp)
Then you should use the ypp tool (ypp -H to get the help)
ypp -e s to generate the input to sort all excitation energies by energy and intensity
ypp -e a to generate the input to get the "composition" of the exciton in terms of KS transitions
ypp -e w to generate the input to plot the exciton wf

The comments should help you to edit the inputs. Anyway ypp runs quite fast on a PC so you can "play" with it.

[Daniele Varsano]

Dear Lijin,
welcome to yambo,

But I encounter some problems and ask you for help:
1. QP_E: QP corrections using a scissor operator and valence/conduction stretching.
scissor | cond. stretching| val.stretching.
What's the meaning of cond. stretching and val.stretching? And how to determine the values of the three parameters?

These parameter control the stretching of the QP conduction/valence bands with respect the LDA (or other
DFT) previously calculated energies. In many case the QP energies are not only rigidly shifted (scissor) with respect
the LDA but there is also a stretching. You can find the explanation in this post of the forum by Myrta.
Hoe to determine them, well you should first perform a GW calculation and then evaluate this parameters,
alternatively directly from the experiments.

2. In your paper, you have shown the Optical spectrum of bulk silicon. I tried to reproduce the results but I failed.
Would you please show me the input files of bulk silicon just as the example of 3D LiF (Abinit and yambo input
files)?

In the paper there is the input file (pag. 1401, Table 2), try to reproduce the result with that input, if it fails let
us know where you encountered problems.

3. How to get the GW bandstructure of bulk silicon by yambo?

Calculating the GW energies corrections for all the bands and k-points you need, these are determined
by the variable: QPerange, see here the documentation.

4. How to get and show the excitonic wave function and the excitation energies?

To do that there is a post-processing tool, that you should compile typing make ypp, then you can find hoe to use it
in the tutorial, or typing ypp -h for a short list of the options.

Hope this helps,

Daniele

Lijin
Xiangtan University, Xiangtan, Hunan province, China

You can update your signature directly in your profile, avoiding to write it any time you send a post,
and, in order to avoid double account, may ask you if you previously registered in the forum with the nickname lijin?

[lijin18]

Dear myrta gruning and Daniele Varsano ,
Thank you very much for your useful help!

E_{QP}^{cond} = cond. stretching*E_{LDA}^{cond} + scissor
E_{QP}^{val} = val. stretching*E_{LDA}^{val}

I found two parameters for E_QP, XfnQP_E and KfnQP_E, what is the difference?

From the formula of E_QP, cond. stretching and val. stretching control the difference of the bandstructure shape between LDA and QP, and the scissor controls the difference of LDA and QP gap. For the simplest case, as cond. stretching and val. stretching are set to 1, it is considered that the LDA and QP have the same bandstructure shap, and the scissor may be equal to the Gap(QP)-Gap(LDA). So, we can easily obtain the QP bandstructure when scissor is determined. Is this right?


Do you refer to the converged or the example spectrum. What is the problem exactly?
For the unconverged spectrum the instruction on the paper should be sufficient. For the converged caclculation you can download all necessary info here.
I agree with myrta gruning's advice that I did an unconverged calculation. As Yambo is besed on Abinit or Pwscf, what parameters are needed to test before as for the converged calculation(For Abinit: ecut, number of kpoints, nband; for yambo: QpntsRXs, BndsRnXs, NGsBlkXs, BSEBands, BSENGBlk, BSENGexx )? Would you please give some instructions about how to converge the parameters?
Thanks a lot!
lijin

[lijin18]

You can update your signature directly in your profile, avoiding to write it any time you send a post,
and, in order to avoid double account, may ask you if you previously registered in the forum with the nickname lijin?

Yes, I registered in the forum with the nickname lijin, but I didn't receive the e-mail. So I did another register with lijin18.
You can cancel the lijin account.

[Daniele Varsano]

I found two parameters for E_QP, XfnQP_E and KfnQP_E, what is the difference?

The difference rely in the quantity you want to apply your correction:
E_QP --> enter in the Quasiparticle band structure
XfnQP_E --> response function X
KfnQP_E --> Diagonal of the BSE kernel

From the formula of E_QP, cond. stretching and val. stretching control the difference of the bandstructure shape between LDA and QP, and the scissor controls the difference of LDA and QP gap. For the simplest case, as cond. stretching and val. stretching are set to 1, it is considered that the LDA and QP have the same bandstructure shap, and the scissor may be equal to the Gap(QP)-Gap(LDA). So, we can easily obtain the QP bandstructure when scissor is determined. Is this right?

Yes, that's right

Would you please give some instructions about how to converge the parameters?

Well, you should perform different calculations increasing the values of all the parameters until the quantities you want to calculate do not change anymore according to the desired accuracy.

Hope it helps,

Daniele

[lijin18]

[quote="Daniele Varsano
PS: Lijin, Please put your affiliation in the signature of your profile, it is a rule of the forum!!
[/quote]

I am sorry. Is it OK now?

[Daniele Varsano]

I am sorry. Is it OK now?

Yes thanks, about the double account, now it is solved,
sorry for that, but we have problem with spam accounts, it always better to register with academic
mail adress.

Cheers,

Daniele

[lijin18]

Yes thanks, about the double account, now it is solved,
sorry for that, but we have problem with spam accounts, it always better to register with academic
mail adress.

Cheers,

Daniele

My academic mail may fail to receive e-mail sometimes, so I wanted to register with other mail.
Now, everything is ok. Thank you for your help. I love yambo people.
lijin