a few more questions on magnetic anisotropy

[decai]

Dear Administrator and Users,

I have a few more questions on calculating the magnetic anisotropy energy.
(1) According to the explanation under LNONCOLLINEAR tag,
MAGMOM line supplies initial magnetic moments only if ICHARG is set to 2, or if the CHGCAR file contains only charge but no magnetisation density.
Then, when I do anisotropy calculations, and I set ICHARG=11, to read in initial magnetic moments by MAGMOM line, do I have to do a non-magnetic (non spin-polarized) calculation in the first step?
(2)for the LSORBIT tag,
Is the magnetic moment in the OUTCAR always the external represention maxis?
(3)If I do conllinear spin-polarized calculation (in the first step), and then do a non-collinear calculation in the second step(ICHARG=11).
To change the spin orietation, the only choice is
MAGMOM = 0 0 total_magnetic_moment
SAXIS = x y z !
by changing the SAXIS.
Am I right?

[admin]

1) yes, you have to calculate a converged, non-magentic charge density first, then calculate the total energies corresponding to the requested spin orientations to estimate the energy differences of different spin orientations.
2) the read values of MAGMOM are just starting guesses, you are right.
3) LSORBIT: if the axis is not given explicitely, the principal axis is the z-axis of the cell, by default.
4) choice the spin orientation: there are 2 ways to do this, as indicated in the handbook:
--) keep the principal axis (SAXIS) along z (which is the default value) and choose the magnetization axis with respect to SAXIS:
e.g. SAXIS=0 0 1; MAGMOM= 1 0 0 (along x) or 0 1 0 (along y)
--) if MAGMOM is set to 0 0 1, this implies that the magnetic orientaion is parallel to the axis you choose via SAXIS
(1 0 0 ---> x; 0 10 --> y)

[decai]

Thanks for the reply.
I am confused with your answer (1) and (4).
According to (1), I should do a non-magnetic (i.e. non spin-polarized calculation).
According to (4) to my question (3), it seems that I can do a spin-polarized calculation in the first step.

My question (2) is the magnetic moment output in OUTCAR always the actual magnetic moment in the cell (i.e. along the same axis)?

I don't quite understand the internal and external representation under LSORBIT tag.
Can the external representation mx(axis),my(axis),mz(axis) be different from SAXIS?