Yambo Community Forum

Dear Developers,

I have two questions regarding the ypp command and I was hoping you could help me with.

I am getting semi-metalic behavior in my semiconductor sample (both DFT + GW) when I use the regular not-shifted kpoint grid in NSCF.
However, this is not happening when I shift the grid and use {X X 0 1 1 0} kpoints in my NSCF calculation. Yet, I need to estimate the band gap at Gamma point which is not present in the shifted grid.
So I decided to use ypp -k k to include it in my QP calculation.

1-Unfortunately, ypp gives me an expanded grid such that it has lower symmetry than the one QE sees, and QE usually adds more points to it. I tried to avoid this by adding the term "nosym=.true." in my DFT calculation, which gives me the segmentation fault error.
Apparently, YAMBO has problem with the added kpoint grids and it can not read ndb.pp file to estimate the enregies at them! Is there any way I can fix this?

2- Trying to overcome this problem, I decided to use ypp -s b to interpolate the band structure.
(1)p2y -N
(2)yambo -i -V RL
(3)yambo - N
(4)ypp -n (5)cd FixSymm
(6)yambo -i -V RL
(7)yambo -N
(8)ypp -s b (9)ypp

While it works fine for the case of the bulk GaSb as I tested, it does give me the error :

"ERROR] STOP signal received while in :[06] Interpolation tool
[ERROR]Interpolation coefficients not present!"

I realized it only gives this error when the band range is above 20, (the band range associated to VB/CB of my sample 136-137!)

I'd appreciate if you could kindly help me with these issues.

Best,
Zahra Taghipour,
PhD student, UNM, USA

Dear Zahara,

About route 1: may be you can ask in the PW mailing list what is the reason of the seg fault when considering nosym-.true.
About point 2: I do not know which precision you need, the extrapolation could be not the best tool to evaluate the gap. Moreover (even if totally not related with the problem you are facing in your ypp input you are not adding QP corrections, this is achieved by adding GfnQPdb= "E < SAVE/ndb.QP" in the ypp input file.
I do not have a clear idea on what is going on, it looks the band range it is not read correctly, looking at the source there is constraint about not interpolating more than 20 bands at the same time, but not more than band index 20. Can you please try to update the version of the code and try again?
It could be there was a bug that has been solved, me personally was not aware of that, may be some other people can give you indications.

Best,

Daniele

Dear Daniele,

Thank you for your response.

About route 1: may be you can ask in the PW mailing list what is the reason of the seg fault when considering nosym-.true.

The problem of ypp -k k that I mentioned and its "segmentation fault" error is in Yambo. QE gives the energies at suggested kpoints by Yambo. However, Yambo gives this error when you want to used ndb.PP from your previous calculaion! I found also the post in viewtopic.php?f=9&t=779 reporting a problem similiar to mine, not solved though.

Can you please try to update the version of the code and try again?

The problem with ypp -s b for band interpolation was solved for me by upgrading Yambo's version! Thank you!

It seems the interpolation is the only way to estimate the gap at gamma point for my structure. However, I don't see what is exactly happening during the interpolation:
Let's say I have QP energies for a grid of k-points. Is it correct to use interpolation to get the QP energies for some other k-points which are not in the initial grid? What about the bands that have not been included in the GW calculation before! Does interpolation including ndb.QP work on them, or Yambo would use DFT energies in such a case, instead?

I was wondering if you could kindly suggest any references to get some information on what Yambo exactly do in interpolation.

Thank you so much,
Best,
Zahra Taghipour, PhD Student,
UNM, USA

Dear Zahra,

The problem of ypp -k k that I mentioned and its "segmentation fault" error is in Yambo. QE gives the energies at suggested kpoints by Yambo. However, Yambo gives this error when you want to used ndb.PP from your previous calculaion! I found also the post in viewtopic.php?f=9&t=779 reporting a problem similiar to mine, not solved though.

Sorry I misunderstood this point, in order to understand what's going wrong here we should reproduce the error.

It seems the interpolation is the only way to estimate the gap at gamma point for my structure.

Well, using the extrapolation probably you can have an estimation, but depending on the reliability of the interpolation the result could be not very accurate. Probably the most accurate way to have your gap at gamma, would be to consider the k point grid including gamma and recalculate the screening for this grid.

Let's say I have QP energies for a grid of k-points. Is it correct to use interpolation to get the QP energies for some other k-points which are not in the initial grid?

Yes, it is correct, but as I wrote above, it is not guarantee that the result is accurate.

What about the bands that have not been included in the GW calculation before! Does interpolation including ndb.QP work on them, or Yambo would use DFT energies in such a case, instead?

If all is working correctly the code should linearly interpolate the missing band based on the provided QP energies.

The reference for the band interpolation used in Yambo is this one: Phys. Rev. B 38, 2721

If the interpolation fails, it is possible to perform a more accurate interpolation using the WanT code,
http://www.wannier-transport.org/wiki/i ... /Main_Page
as it can read the QP outputs produced by Yambo.
This way should be more accurate but it has the drawbacks that it is possible that many qp corrections need to be calculated in order
the algorithm works correctly. If you think to pursue in this way let us know, as it is not well documented and Andrea Ferretti can give you some hint.
Best,

Daniele