DFT+U Hubbard Correction on an Unconventional Atom not already configured in Quantum ESPRESSO

In this Quantum ESPRESSO tutorial, I show you how to run a DFT+U calculation for an element that is not already configured.
Essentially, one has to supply the n,l and occupancy of the orbitals for which the correction is desired.
The necessary files to be edited are PW/src/tabd.f90, Modules/set_hubbard_l.f90, Modules/set_hubbard_n.f90 (only in newer versions)

[@anibalbezerra]

very nice!! tks by elucidating such a specific step within QE.

[@anishdas39]

great job. It took me one month to implement U for non implemented elements. I wish this video was available that time :)

[@PhysWhiz]

Thank you so much for your kind words :)

[@iyorzorbenjamin364]

Thank you very much for this video, it has helped me a lot.

[@PhysWhiz]

You're welcome, but I should let you know that the Hubaard correction related stuff has changed in the newer versions of Quantum ESPRESSO and the method shown in this video may not work anymore. I haven't tried the newer versions yet.

[@haaardcoooreee]

Sir, I'm gonna write your channel name on my thesis acknowledgments. THANK YOU SO MUCH.

[@PhysWhiz]

I'm glad you found it useful.

[@amirhosseintalebi6770]

very nice. Let us know about more upcoming videos.

[@biprojitsana9873]

It is a very useful viseo. Thanks a lot. I want to do DFT+U+SOC. For that I need fully relativistic pseudopotential for Yb element. DO you gave any idea how to get it? I have checked QE and SSSP library. It is not available.

[@yuhitgupta3388]

Thank you so much for these steps I was trying it from last three months. Will you please suggest how to run for GGA +U calculations, as it was for LDA + U

[@PhysWhiz]

This was GGA+U indeed. Although the keyword is called lda_plus_u, if the PP you're using is GGA, ex:PBE, then Quantum Espresso runs GGA calculations. You could also specify the XC functional using input_dft keyword. Although, even if you don't, it ises the XC according to your PP.

[@yuhitgupta3388]

@@PhysWhiz Thank you so much for your valuable reply.

[@yuhitgupta3388]

@@PhysWhiz please help in how to choose hubbard_U(1), U(2) and U(3) for different atoms like Mn, Y and Al

[@ManasSharma07]

@@yuhitgupta3388 The hubbard parameters can be figured out either semi-empirically, i.e. by matching some experimental property like Band-gap, bandstructure, bandwidths, etc. But ALWAYS REMEMBER, that the Hubbard parameters for the same element can be different when the element is in different molecules. So, you need to figure out the values for the atoms you mentioned, for the molecules (crystal systems) that you want to study. Also, they depend from software to software, based on the implementation. So, I would recommend that you search for the Hubbard parameters in literature but check if they used them for the same system and using Quantum ESPRESSO. More tips: hjkgrp.mit.edu/content/right-hubbard-u-any-element hjkgrp.mit.edu/content/quick-tips-more-about-u One can also, determine the Hubbard parameters, by first principles (ab-initio). Refer to literature for these. In the 2018 update, QE added a feature to calculate the Hubbard parameters ab-initio. Check that out. hjkgrp.mit.edu/content/calculating-hubbard-u hjkgrp.mit.edu/content/calculating-hubbard-u-periodic-systems

[@yuhitgupta3388]

@@ManasSharma07 Thank you

[@kunsahaho607]

Zn.peb-van.UPF is LDA or GGA? can we do GGA+U or LDA +U? which one is correct and better approximate?

[@ammar1099]

Good video ! . could you tell me how to add Bismuth into these files set_hubbard_l.f90 set_hubbard_n.f90 and tabd.f90 ?.

[@stephanalbrecht8479]

I did all the steps you showed in this video including reconfiguration and rebuilding the pw-package, but for Silicon atoms, and if I run the calculation I still get the same error: 'Error in routine set_hubbard_l (1): pseudopotential not yet inserted' I watched the video several times now and checked the files twice... I inserted 'Si' in all files

[@PhysWhiz]

Which version of QE are you using?

[@stephanalbrecht8479]

@@PhysWhiz qe 7.0 should be the newest, I just installed it couple of weeks ago

[@stephanalbrecht8479]

@@PhysWhiz I am trying to correct the underestimated band-gap for SiC and thinking now of just using a hubbard-U value for Carbon only. Do you think this will solve the problem, as C was allready included before?

[@PhysWhiz]

Is there any literature on this? You can try what others are doing

[@PhysWhiz]

Also, you can try using the same version of QE as I am using in the video

[@andyvo314]

i have a trouble with Sn (Thin) can anyone help me where i have to put Sn on tabd.f90, set_hubbard_l.f90, set_hubbard_n.f90. tks kiu

[@aiswaryaprasad4504]

How to calculate the Hubbard vale ?

[@alljobzinpakistan5070]

Hi, I need help to solve the error from QE by using u+v method.. I pay some money if you solve it

[@yuhitgupta3388]

One question can we compare the results of Electronic band structure of spin polarized calculations with band structure with spin orbit coupling...for magnetic material.. please reply if anybody know

[@PhysWhiz]

Spin orbit coupling should give slightly different but better results. But I am not too sure what you want to do.

[@yuhitgupta3388]

Thank you sir...

[@sancharibhattacharya4984]

"Error in routine offset_atom_wfc (2): wrong offset" This is coming when I am trying to incorporate La in the code

[@PhysWhiz]

www.researchgate.net/post/How_can_I_solve_wrong_offset_problem_in_quantum_espresso_relax_calculation

[@debojyotinath4602]

How to do this in Burai

[@PhysWhiz]

Unfortunately you cannot do this in BURAI, as it comes with pre compiled binaries. But here, we have to change the source code, therefore we have to compile the executables/binaries again. I only know how to compile the binaries on linux and macos. The video shows how to compile them.

[@PhysWhiz]

If you're a windows 10 user, you can also run QE using Windows subsystem for linux (WSL). kzbin.info/www/bejne/eafVoqCnbqZmqqM This will behave like a proper linux machine and you will be able to use the linux bash commands and also install fortran and QE just like you would on a linux computer.