[Wien] MAE; S-O calculation

[Stefaan Cottenier]

> > What I am trying to do is that I want to calculate the
> > difference in energy between two magnetization
> > directions [001] and [111] of Fe bcc.
>
> As far as I know there are several calculations of the magnetocristalline
> anisotropy for Fe, Co and Ni in literature. The problem is, that these
> values are extremely small (micro Ry) and require a k-mesh of several
> 100000 k-points (and even then it might not be converged.). Freeman etal
> even "invented" the "state-tracking method" (what ever this is ?) to get
> some numerical stability.
> In addition I think to remember, that the magnitude (too big) but even the
> sign of it does not fit experiment for all/some of those 3 elements.

A relevant reference might be Trygg et al., PRL 75 (1995) 2871. They report
on full potential calculations using a few thousand k-points (less than the
hundreds of thousands used previously) for Fe, Co and Ni. Order of magnitude
of the MAE is correct in all cases, but the sign for Ni is wrong. They
discuss in detail how they checked the stability of their results.

Stefaan