Hi, I am working on variations of LiNiO2 supercells. My supercell is around 200 atoms. The k points are 1 1 1 and 520eV is the cutoff we use. The system is too big that 1 1 1 or 2 2 1 k mesh should be used. GGAPBE is used with Hubbard U corrections. The calculation is spin polarized and PBE.usp pseudopotentials are used for all atoms.

The issue is we can only geometry optimize in coarse level for now, however to publish a well article the force per atom should be around 0.01. Even all other criteria are identified as coarse, only making force per atom 0.01 makes optimization hard (typically not optimized for 100 cycles and force per atom never gets a YES.)

Second issue we have is the volume change. We design the supercell from experimental values. There is a 10 % increase which is too much. How can we proceed with this?