Convergence and Bandstructure with GW correction

[rupesh]

Hi all,

I am new to Yambo. I want to plot the bandstructure of Black Phosphorus and Phosphore with the GW corrections. I read the tutorial given at http://www.yambo-code.org/tutorials/GW/ . But I have problem in convergence plot given on this site . How did they get Direct and indirect bandgap from data like
XC HF and DFT [eV] @ K [1] (iku): 0.00 0.00 0.00
<1|HF|1> = -16.98063 -.3073E-7 <1|DFT|1> = -10.50084 -.5556E-8
<2|HF|2> = -12.04911 0.2742E-8 <2|DFT|2> = -11.32918 0.1649E-7
If I have to do convergence study how should I proceed ? Please help me .
Is there any scrpt available which convert the data in file o.qb (File of final GW PPA ) to plot bandstructure or we have to do it manually ?

Thank you !!

Regards,
Rupesh

[Daniele Varsano]

Dear Rupesh,
first please let me ask you to fill your signature with your affiliation. This is a rule of the forum.
Please consider that the part of the report you posted gives you DFT energies and Hartree-Fock self energy so it is not the GW one.
In order to plot the band structures you can either to by hand (after a succesfull GW calculation you have an o.qp file where GW energies are reported), either you can use the post processing executable ypp (compile it with make ypp).
Here some instructions on how to use it:
http://www.yambo-code.org/input_file/ypp/index.php

About convergences, please have a look to the tutorials and the forum.
In general you need to converge: Gvector in the exchange (EXXRLvcs), band in the screening (BndsRnXp), bands in the GW summation (GbndRnge) and Block in G-space of the screening (NGsBlkXp) and K-points by running different ground state calculations.

Best,
Daniele

[rupesh]

Dr. Daniele ,

First of thank you for instant response. The link that I had given in the last mail ,I want to know in the plot of [eV] Vs kpt or Ry (i.e. convergence with respect to EXXRLvcs 3,6,7,15) which value of energy should I choose so that I can see the convergence is happening or not ? This may be very trivial or basic question but please help me in detail. I have calculated for Si (4x4x4 k-grid) with EXXRLvcs 3,6,7,15. I have attached the same files below.

[Daniele Varsano]

Dear Rupesh,
looking at your report files, peaking for instances the first k point and first band you have:
3Ry -16.98063 eV
6Ry -17.08444 eV
7Ry -17.08972 eV

similarly for the other bands. It looks that 3Ry it is not enough while 6Ry it is converged.
Here if you want to plot the HF band structure, there is not output so you need to play with the report file preparing a script.

Best,
Daniele

[rupesh]

Thank you very much !!


Regards,
Rupesh

[rupesh]

Hi all,

I have done the band structure calculation for silicon with slightly modified parameters in Quantum espresso adn I got the band structure for it.Now I started for GW corrections and trying to replot bandstructure for the same. I have converted my pwscf silicon.save file to SAVE database required for YAMBO. But when I start running calculation in Yambo I got error. I have attached error that occures on terminal here . I think k-point in my "silicon_nscf" file may be the problem but it run successfully in Quantum Espresso.
One more doubt If I change the k points in that file to automatic then how would I get GW- corrected band structure plot along the same directions as calculated in quantum espresso so that I can compre both. Please help me.


Thank you!
Rupesh

[Daniele Varsano]

Dear Rupesh,
yambo need uniform k-point grid, so if it works for qe, does not mean that it will work for yambo.
I suggest you to stack in automatic mode of k-point sampling in the nscf qe calculations.
Consider that GW needs integration over the Bz, so you cannot have the sampling along the desired direction only.

Did you run the setup of yambo first?

Once you have your kpoint sampling, you can calculate GW corrections on that point and eventually you can interpolate the results along the desired direction using the ypp utility.

When posting about errors, it is always useful to attach also input and report files.

Best,
Daniele

[rupesh]

Dr. Daniele,

Yes, I run the setup of yambo first ,it was successful. When I give input as GoWo_silicon file (attached below) it had thrown error. I will try again with symmetric k-grid (4x4x4) in nscf file and will see what happen .
Very very thank you for helping me in this problems !

Regards,
Rupesh

[Daniele Varsano]

Yes,
the setup run but if you look at the report here is the problem:

Code: Select all

 [RD./SAVE//ndb.kindx]---------------------------------------
  Polarization last K   :  36
  QP states             :   1   36
  X grid is uniform     :no
  BS scattering         :no
 - S/N 002674 --------------------------- v.03.04.01 r.3187 -

 IBZ Q-points : 1
 BZ  Q-points : 1

 Q [00001] : 0.00      0.00      0.00     (iku) * weight  1.000000

 Q [00001] : 0.00      0.00      0.00     (cc ) * weight  1.000000

The k-grid is not uniform and you have only one point in the q-grid (k-k'). This is the reason why next there are problems in integrate in the Bz for the calculation of the Self Energy.

Best,
Daniele

[rupesh]

Hi Daniele ,

Thank you so much for last reply and it works when I use symmetric k-grid in nscf file. But now I am running ypp to plot band-structure , I am unable to understand how o run it . I tried to search on documenttaion where I got the structure of Input file necessary for it. Is there any initialization process for ypp just like yambo? Because as mention here http://www.yambo-code.org/input_file/ypp/ypp_bands.php . I have written input file and attached the same below. When I run ypp -s b -F input-file it just print of terminal big symbol of Yambo name and <---> [01] Game Over. It will be very helpful to me if you give me detailed execution process for ypp.
Second point is that Is it necessary to run BZ kpoint grid generator (ypp -k k) before moving to Bands Interpolation (ypp -s b) ? If it is then for each point in my Bands Interpolation file (i.e. 0.5,0.5,0.5 ; 0.0,0.0,1.0 ; 0.0,1.0,1.0) should I follow the same procedure and then move to band Interpolation ? But even I typed on terminal ypp -k k -F Input file name it produce no effect just mentioned above print YAMBO name in terminal and Game Over .

Why is this happening ?

Regards,
Rupesh