Hi folks,

Currently running some convergence tests for magneto-crystalline anisotropy energy (MAE) of Fe (e.g. E_111 vs E_110 vs E_001).

As you know calculating MAE requires very strict convergence criteria as the relative energies can be on the order of 1 µeV, and can require k-points on the order of 10,000 or more.

However, the calculation is also dependent on the smearing width, which has a non-negligible influence on the # of k-points required for convergence, and smaller smearing requires higher k-points.

I am currently using VASP with the tetrahedron method with Blöchl corrections (ISMEAR = -5), with 0.01 eV smearing, and I find that 10k k-points is not enough to reach convergance (it has reached only ~ E_110 > E_001 of Fe is correctly predicted, but I don't know how accurate the magnitude would change with smearing, or if perhaps I should try even smaller smearing widths.

Thanks in advance,

-B