I need some suggestion for Hybrid functional calculation. Any model input would be a help and also some guideline on parallelization.
My primary goal is to calculate the electronic band structure.
Since, the LDA level of calculation underestimates the lattice parameters. My goal is to get a better bond length as well as the forces in the system. Therefore, I am also curious whether I should perform a hybrid calculation for the relaxation calculation (that would be incredible expensive I presume), or, I should only do the relaxation calculation with PBE functional and the perform HSE calculation as an SCF calculation to get a better bond length as well as the forces.
I am looking forward to suggestions from the ABINIT community.
Are you studying molecules or solids? If molecules I’d suggest PBE0, if solids, HSE06. Second, make sure to read the Hybrids topic on the web site. It’s a little sparse, but gets you started. Third, use input files tests/v7/Input/t69.abi and tests/v7/Input/t71.abi as starter input files. Hybrids are challenging in solids, there are a lot of input parameters to consider and tune. So start on a VERY simple system, to practice, for example Si or C (only two atoms in the unit cell). Forces and stresses are supported, so you can compare different relaxation strategies and eventually apply to your system.
Oh and for band structures, since you always have to use a regular grid of kpoints, you will want to use abipy tools to interpolate the band structure (rather than computing NSCF energies on given k points as usually done, see my question and its answer just below).