Dear all, I am pretty new to Quantum Espresso and Wannier90. After finding out the bandstructure of my system using QE, I am trying to find out the wannier interpolated band structure, but these two band structures show a significant discrepancy. I have tried changing the projection and disentanglement window, but I am not getting a better result. The input files and bandstructure plots can be found here https://drive.google.com/drive/folders/1kxI7TZ4UD4x3TlCX8vT-J2WVLzm0nBTX?usp=sharing . It might be a minor solution for the experts. Hoping to get a positive response. Thank You.
Hi Bhargab Kakati
I have seen your files. I can give you a few suggestions, which you may try and see whether it works for you or not.
1. You have taken num_wan = 72 and num_bands = 75. In general, it is preferred to keep num_bands as double of num_wann, so try to take num_bands = 144, it may drastically change your band structure.
2. Take note of the projections of DFT bands, which bands are contributing near the Fermi level, and then take the projections in Wannier accordingly.
3. Try without excluding bands, and see how it goes.
4. Try to keep the same dis_win_min and dis_froz_min, and increase dis_win_max a bit, say 30.
Hope this helps!
Best regards,
Payal
Thank you Payal Wadhwa for your response and suggestion. I will try running different calculations as per your suggestion and get back to you.
Hello Payal Wadhwa, I have tried running calculations as per your suggestion. The calculation with num_bands as double of num_wann did not converge even in 8000 iterations when the dis_win_max was kept as before. But it converged when the dis_win_max was increased to 30. But the band structure remained almost the same and the discrepancy with the DFT band structure remained.
In the case where I did not exclude any bands, it did not converge. I have attached the input and output files here https://drive.google.com/drive/folders/1AcgzEfYKRLe6UMOtKkQwFZq8ckOXpEAU?usp=sharing
thank you.
Hello Bhargab Kakati
I have seen your ".win" file, and I have found that you completely removed the contribution of Ga, It may also alter your results. Try to add it and see how it goes.
and sometimes, it works well even without converging the disentanglement_criterion, but don't increase the iterations too much, instead try with a small k-mesh first.
The only thing we need to ensure is Hamiltonian shouldn't have imaginary frequencies.
Hope this helps!
Best regards,
Payal
Hi Payal Wadhwa , thank you for your response. I excluded Ga because it barely contributes to the density of states. But even after including it the bandstructure does not significantly change.
I have one question. Can I proceed with postw90.x calculations even though the DFT and wannier90 interpolated bandstructures do not match? Will it significantly affect various properties like anomalous hall conductivity, spin hall conductivity etc?
thank you.
yes Bhargab Kakati
If the band structure doesn't match properly, then it may have a significant effect on properties.
and try to keep dis_win_min and dis_froz_min the same.
I use VASP, where i never used such a huge value of dis_win_min and dis_froz_min. I generally take it as negative, but in your case, Fermi energy is too high, so i believe it should be correct whatever you have chosen.
But, try with their same value.
I agree that there are a lot of hits and trials, but this is how this code works.
Best regards,
Payal
Payal Wadhwa , I will keep that in mind in further calculations. Thanks for your help.
- Badges
- Science topic