Hybrid functional with libxc
Brought to you by:
jkdewhurst,
sangeetasharma
Menu
▾
▴
Using elc code (3.1.12), I am trying to compute Si bandstructure using HSE06 hybrid functional. To do this, I tried to change in the example <elk_root>/examples/hybrid/Si-hybrid-libxc.
I do as proposed, first run with PBE, then chane the task to 5 and change+uncomment line after 'xctype' to '100 0 428' which should correspond to HSE06 in libxc. Also, I added task 20 to see the bandstructure.
The result is extremely strange: the result lookls like if the Fermi level goes far to conduction band and the band gap is not what expected for HSE06 at all, more or less the same as with PBE (WIEN2k gives 0.62 with relatively small k mesh and the same positions of atoms).
Am I doing something wrong?
Also, I was wandering whether it is a way to control parameters of hybrid functionals or such mechanism is not assumed in (either) libxc/elk?
Also, do hybrid functionals work with spin-orbit in current version of elk?
Dear Mikhail,
the answer is simple: elk does not support screened hybrids. You can however use PBE0, but I don't think it is reliable.
Regards
Markus
Dear Markus, thank you for the explanation. It was not completely clear from the documentation.
Do you plan to include them in future releases?
Dear Mikhail and Markus,
Elk supports some types of hybrids but not those that require short-range and long-range exchange.
The reason is that it is difficult to make a SR/LR Poisson solver for LAPW (difficult but not impossible), so we haven't got around to it yet.
We'll almost certainly put it in at some point, but it's not priority at the moment. We're currently working on getting the GW approximation working.
Regards,
Kay.
why would PBE0 not be reliable?
regards
manuel