Hubbard U derivation via DFPT

[william_watson]

Dear all,

I have been using VASP to derive Hubbard U_eff values using the constrained DFT linear response method which is well documented. However, I am interested to know whether it is possible to derive U values using density functional perturbation theory (DFPT)? A comparison of the two methods is provided here: https://indico.cern.ch/event/611331/con ... ov_625.pdf. This question has also been asked previously on the forum: forum/viewtopic.php?f=4&t=17598&p=18681 ... fpt#p18681.
This procedure is automated in Quantum Espresso via the HP code but I wondered whether something similar was available in VASP given this is the software that my group uses.

Many thanks,

Will

[ferenc_karsai]

Please check out this example on the VASP wiki:

wiki/index.php/Calculate_U_for_LSDA%2BU

Mind that it uses "LDAUTYPE = 3" which is still not explained on the LDAUTYPE site, but hopefully the tutorial will give you an explanation.

[william_watson]

The method you reference is DFT linear response rather than DFPT. Does your response mean that only DFT linear response is possible in VASP?