Blank output file during Static screening calculation

[nsingh3]

Dear Daniele,

The psps very heavy files so I dint upload them. The tarball also did not work as it could minimize the size only upto 783Kb.
You can find them here: https://www.abinit.org/psp-tables. In this the PAW-Datasets and PBE functional.

I do not understand what you mean by this: "Note that the r_setup is from a run without NodiagSC, actually no input file is read, and with a previous generated ndb.kindx database. The database indeed it is read and not generated."

I generate the yambo.in file using yambo -i and then uncomment NoDiagSC there. Next I run $ yambo yambo.in which makes the r_setup file.
What am I doing wrong here.

-Nisha
Postdoctoral Researcher,
TU Delft, NL

[Daniele Varsano]

Dear Nisha,
The correct syntax is yambo -F input
If not indicadet yambo reads by default yanbo.in

So it is correct.

From the mirror file at the end of the report I can see:

Code: Select all

 .-Input file  yambo.in
 | setup                            # [R] Initialization
 | BoseTemp= 0.000000         eV    # Bosonic Temperature
 | K_grids= "X"                     # [KPT] Select the grids (X=response, S=sigma, C=collisions, B=bse) [default X S C B]
 | NoDiagSC                       # New setup for non-diagonal supercells
 | %QptCoord                        # [KPT] [iku] Q-points coordinates (compatibility)
 | %

it should be correct, I just wonder if the QPtCoord namelist was present in the input or not. If so, try to remove it.
The ndb.kindx was indeed generated and not read, probably I get confuse with another report file:

Code: Select all

 [WR./SAVE//ndb.kindx]-

Finally note that PAW pseudo are not supported, only norm conserving pseudo are supported.

Best,
Daniele

[nsingh3]

Dear Daniele,

Thank you for the reply but I still have one question.
I did 1T-HfS2 and 2H-HfS2 using PAW psp with SOC and it went well without any error in yambo.
Now I am wondering why it is creating problem in 1T-HfSSe? As you pointed out only norm conserving PSPs have to be used then I should have faced the same issue even with 1T-HfS2 and 2H-HfS2.
Where am I wrong.

-Nisha
Postdoctoral Researcher,
TU Delft, NL

[Daniele Varsano]

Dear Nisha,

the problem you are facing is not related at all with the PSP as it deal with the k point sampling.
The problem with PAW is that the results you obtained can be inaccurate/unreliable.

Best,
Daniele

[nsingh3]

Dear Daniele,

I am still getting the same blank screening file with norm conserving PSP and my k-point grid is also 6 6 6.
So where is the non-uniform grid being formed? At the abinit level?

-Nisha

[Daniele Varsano]

Dear Nisha,
the problem you are facing is totally independent from the PSP you are using.
The problem is due to the fact that Yambo sees a mismatch between the k point and q point grid:

Code: Select all

[RL indx] Q BZ pts are /= from X grid BS pts. Gamma point only.

I do not know why this happens, and this is the reason why I suggested you to provide a blank setup input file (besides the NoDiagSC).
As this do not solve the problem, we need to reproduce the issue and inspect it, but as I told you I'm out of office so I cannot do that.
I hope someone else will take care of this in the next days.

Best,
Daniele

[nsingh3]

Dear Daniele,

I hope you have enjoyed your holidays.
I am waiting for the update on this thread as I am facing this issue alot.
Any update will be helpful.

Best Regards,
Nisha Singh
Postdoctoral Researcher,
TU Delft

[Daniele Varsano]

Dear Nisha,

I can see that you are using an old method to convert abinit ground state into yambo databases. KSS files are not needed anymore and you can convert directly the .WFK files.
You can follow this post where it is explained how to do that.

Second, please note that PAW pseudopotential at the moment are not supported.

Best,
Daniele

[Davide Sangalli]

Dear Nisha,
as Daniele pointed out the PAW pseudo are not supported.

This means that, using the new procedure Daniele suggested you cannot use the variable prtkbff in abinit.
Also all parts of the code where PAW augmentation is required will not be coded.

Anyway, you can still ignore this (do not add prtkbff in abinit input) and go on with the simulations.
Now The problem you are experiencing is because you set

Code: Select all

pawspnorb 1

and abinit disables the time-reversal symmetry accordingly.
However a2y does not know such variables, and it sets by default the time reversal symmetry to true.
(I'm not sure how much the reading of SOC parameters from abinit is coded, also in the standard case of NC psps)

To circumvent this, you can open the yambo source file

Code: Select all

interfaces/int_modules/mod_com2y.F

change in line 107

Code: Select all

i_time_rev=0

and recompile

Code: Select all

make a2y

After that, you should be able to generate a SAVE which works (but without the info on the pseudo as long as you work with PAW ones!)

Best,
D.

[nsingh3]

Dear Nisha,

I can see that you are using an old method to convert abinit ground state into yambo databases. KSS files are not needed anymore and you can convert directly the .WFK files.
You can follow this post where it is explained how to do that.

Second, please note that PAW pseudopotential at the moment are not supported.

Best,
Daniele

Dear Daniele,
I did the calculations again with norm conserving PSPs after you had pointed out that PAW PSPs are not there in Yambo in one of the previous posts. And I am also aware of the fact that KSS files are no longer generated by abinit so I use the new method of generating them from the .WFK.nc files.
I am attaching the input file again to avoid any confusion.

Dear Davide,

I am not using prtkbff variable in abinit in the input file.
pawspnorb was used when PAW PSPs were used but when I used Norm Conserving PSPs I used nspinor=2.
I am still facing the problem of having a blank screen of screening. I tried with MoWS2 as well but the same problem comes up. I am attaching the input file with spin orbit commented out.
So are you suggesting that I can not do spin orbit in BSE with NC PSPs?
r_setup shows that the grid is non uniform which is the problem.
You will be able to reproduce the problem with the input files and r_setup is also attached.

-Best Regards,
Nisha Singh
Post doctoral Researcher,
TU Delft