Yambo Community Forum

Dear all,

I hope you are doing well. Recently, I have been trying to converge Hartree-Fock parameters; namely, EXX components and VXC components. The direct band gap converged with EXXRLvcs of 50 Ry with a difference of around 0.026 eV with that of 40 Ry. However, due to memory issues I had to lower the number of Vxc components. The Quantum ESPRESSO xc contribution calculated in the scf step was "-384.75302105 Ry".
Each time I sat VXCRLvcs to any value between 20 Ry to 90 Ry I got almost the same value of "-259.5259 [Ry]", which is way far from that of DFT. Is that to be expected? And if not, is there a way to get a value closer to that of DFT?

Please find attached samples of the input, output, and report files of the calculations.


Best regards,
Mohamed
Ph.D. student, Ain Shams University, Egypt

Dear Mohamed,

Usually is a good practice to maintain the larger (density) energy cutoff for the Vxc. This part of the calculation is fast and should not need much memory. Anyway, you can lower it if it does not preclude accuracy. In your case, passing from 50 to 90 Ry you have a sizable difference of about 0.12 eV.

You are not obtaining the QE result because your pseudopotential contains Non-linear core correction. By default, in Yambo are neglected, but you can activate them using the keyword UseNLCC in your input.

In general, it is better to use pseudo not containing NLCC (as e.g. sg15), as while they can be taken into account in the Vxc part they are not considered in the Fock term, this is exactly the same situation you encounter in QE when pseudo with NLCC are used in hybrid functional calculation.

Best,

Daniele

Dear Daniele,

Thank you so much for your time and concern, I really appreciate it.

"Usually is a good practice to maintain the larger (density) energy cutoff for the Vxc."
I have learned this in the tutorial, however in my case the G vectors are too much "401131".

Also, I didn't know about the "sg15" library. I tried them and it worked very well. Thanks again.

All the best,

Mohamed