Clarification on Non-Self-Consistent Calculations for Magnetocrystalline Anisotropy Energy in VASP

[farah_shahzadi]

I am following the procedure described in the VASPwiki to calculate Magnetocrystalline Anisotropy Energy (MAE). The process involves performing an accurate collinear calculation first and then running non-self-consistent calculations with spin-orbit coupling. My question is regarding the second step: should I use only the CHGCAR file, or should I also include the WAVECAR file? When I include both WAVECAR and CHGCAR, I encounter an error, but the calculation runs successfully when using only CHGCAR. Could someone clarify this for me? Thank you.

[marie-therese.huebsch]

Hi,

as you have basically discovered by testing, you need to restart from CHGCAR and not use WAVECAR.

The nonSCF approach to computing the MAE involves
1. a collinear calculation with ISPIN=2 which yields a charge density and magnetization written to CHGCAR (vasp_std)
2. restarting the calculation with LSORBIT=T (vasp_ncl) and specific direction of the magnetic moment specified by MAGMOM and SAXIS with fixed charge density (and magnetization).

The wavefunction is computed based on the provided charge density and magnetization.

Best regards,
Marie-Therese

[farah_shahzadi]

Thank you for your guidance. I want to clarify that it does not matter if I do not use the WAVECAR file. Using just the CHGCAR file from the first process in the second process will yield correct results, right?

Regards
Farah Shehzadi

[marie-therese.huebsch]

Yes. Be careful to set LMAXMIX and do a proper convergence until reaching the energy criterion set by EDIFF.

Marie-Therese