ndb.kindx

[xjxiao]

Dear Daniele,
I reran the calculations, performing structural optimization while preserving symmetry, and confirmed that the plotted band structure is correct. After initializing Yambo, the generated k-grid is uniform.
However, the temperature is no longer zero.

Code: Select all

  [X] Electronic Temperature                        :  0.861734E-5  0.100000    [eV K]
  [X] Bosonic    Temperature                        :  0.861734E-5  0.100000    [eV K]

Why does this happen?
Thank you very much!
Xiao

[Daniele Varsano]

Dear Xiao,

it happens that your system is seen as metallic, you have some bands with partial occupations. e.g. k point=4.
Check if this is consistent with shat you found in the QE calculation. Is it possible you used a large smearing factor in your QE ground state?
If you think that it should be a semiconductor, try to run the ground state using occupation fixed.

Best,

Daniele

[xjxiao]

Dear Daniele,
Currently, I have not set occupations in my calculation. The system exhibits a positive direct bandgap but a negative indirect bandgap in PBE-DFT calculations within QE, while it becomes a positive-gap semiconductor under HSE.
For the PBE-DFT calculation in QE, should I treat it as a semiconductor and set occupations = 'fixed'?
Thanks!
Xiao

[Daniele Varsano]

Dear Xiao,

for what you say, the system is semimetal, so it is correct you have fixed occupation and a finite temperature.

For the PBE-DFT calculation in QE, should I treat it as a semiconductor and set occupations = 'fixed'?

Not having the system a finite gap, I do not think that the calculation will converge. And even if so, Iìm not sure you'll get a meaningful solution.

Best,

Daniele

[xjxiao]

Dear Daniele,
After setting occupations = 'fixed', the calculated temperature remained the same as before.

r_setup_01.txt

Strangely, the output shows Finite Temperature mode = no. Does this mean non-zero temperature has no effect?

Code: Select all

  [X] Electronic Temperature                        :  0.861734E-5  0.100000    [eV K]
  [X] Bosonic    Temperature                        :  0.861734E-5  0.100000    [eV K]
  [X] Finite Temperature mode                       : no

After performing a G0W0_ppa calculation, a bandgap of over 0.2 eV opens up. Preliminary parameter tests suggest the bandgap appears to converge. Why do you say the results might be meaningless? Can Yambo perform GW calculations for semimetals? What impact does non-zero temperature have on GW+BSE calculations?

Thanks!
Xiao

[Daniele Varsano]

Dear Xiao,

Strangely, the output shows Finite Temperature mode = no. Does this mean non-zero temperature has no effect?

That's not strange, your system is semimetal, so a smearing is needed in the occupation function (finite temperature).

Why do you say the results might be meaningless?

Because occupation fixed should not be used in semimetal case. Anyway, Yambo recalculate occupation factors so you ended up in the same situation as before which seems to be correct.

Can Yambo perform GW calculations for semimetals?

Yes it can.

What impact does non-zero temperature have on GW+BSE calculations?

You have partial occupations across the Fermi energy, and that's correct in the case of a semimetal.

Best,

Daniele

[xjxiao]

Dear Daniele,
Thank you for your response, but I’m sorry—there are still some parts I don’t fully understand. Could you please clarify further?

That's not strange, your system is semimetal, so a smearing is needed in the occupation function (finite temperature).

Do you mean finite temperature is needed? If so, why does the output show Finite Temperature mode = no instead of yes?

Because occupation fixed should not be used in semimetal case. Anyway, Yambo recalculate occupation factors so you ended up in the same situation as before which seems to be correct.
You have partial occupations across the Fermi energy, and that's correct in the case of a semimetal.

For semimetals, a non-zero temperature is physically reasonable and shouldn’t disrupt GW/BSE calculations. I should not use occupations='fixed' in QE for such cases, but in my tests, Yambo still produced correct results even with this setting. Is my understanding accurate?

Best,
Xiao

[Daniele Varsano]

Dear Xiao,

1) This is because there is a threshold to turn it on.
You have in the report the eigenvalues and occupation number, so you can realize that all is fine.

2) Yes, it seems all is correct. As an advice, do not use occupation fixed when dealing with metals/semimetals.

Best,

Daniele