Yambo Community Forum

Dear developers,

I am using Yambo to calculate GW eigenvalues of a crystalline porous material The unit cell of the system is quite large with lattice parameter of 18.5 Ang. The cell contains a lot of vacuum (pore) in the middle. I have used the Coulomb cut-off for better integration since I am using only Gamma point for DFT calculation. As suggested in FAQs, I used Covariant DipApproach since I am trying to correct a GauPBE DFT ground state. In the log file and in report file, Yambo issues a warning:

"Covariant Dipoles not in 3d. Assuming non periodic system."

What does the warning mean and how can it be avoided?

Note that the system that I am working on is a periodic one.

Thanks and regards,
Aseem

Dear Aseem,

Two issues:

1) in covariant dipoles a derivative in k space is needed so I'm afraid they are not allowed using gamma point only, but maybe someone more expert than me on the use of covariant dipole can correct me.
2) if your system is periodic why are you using a Coulomb-cutoff?

Best,
Daniele

Dear Aseem

you cannot use covariant dipoles because you do not have k-points.
I think normal dipoles should be fine in your system.

for the rest, I will write to you by email

best
Claudio

If you want to calculate GW corrections on top of Gau-PBE, regular dipoles will neglect the contribution of the commutator with the non-local part of the exchange-correlation potential. I cannot say how this part can be relevant.

Note that covariant-dipoles in non-periodic direction are switched to a real-space approach, but this is not meant for periodic systems.

Another option is to consider shifted grid for dipoles, you will need to perform other 3 nscf calculations slightly shifted from gamma.

Best,
Daniele

Thanks a lot for quick replies.
I shall use the default G-space DipApproach.

@ Dr. Varsano, the reasons for using Coulomb cut-off are that use of only gamma point at DFT level and few other specificities of the system.
Thanks.

Regards,
Aseem

Thanks Dr. Varsano.

The shifted grid could be very expensive for the system that I am working on because: (1) it is quite large system (468 electrons), and (2) nscf with hybrid is not possible so I will need to do 3 more SCF calculations with k-mesh more expensive than Gamma.

Regards,