changing the axis of the magnetic moment

[alpinnovianus]

In a collinear magnetic moment calculation for tetragonal cell with in-plane axes (a; b=a) and out-of plane axis (c>a), does the resulting magnetic moment always parallel to the c-axis (001)?

I was reading the SAXIS definition in vasp wiki, and I wonder if this is true.

If yes, suppose I want to model a collinear magnetic state where the moments are oriented along a-axis (100) instead of along the c-axis (001).

I can think of two possible ways:

• "rotate" the computational cell by swapping the a and c lattice parameters and coordinates (so the longer side is now horizontal instead of vertical). It is possible to use the same MAGMOM, collinear calculation with vasp_std this way, am I correct?

• use vasp_ncl, saxis = (100), and supply three numbers for MAGMOM to run a noncollinear calculation. This adds value in a sense that the resulting magnetic moment vector can be directed along any axis, but I expect this noncollinear calculation to be take longer and more complicated. It may not work with some functionals or some cases where vasp_std can work well. (I don't know this for sure)

Since I only want to change the axis of collinearity from c-axis to a-axis, would it be fine to just use the first method that rotates the cell?

[henrique_miranda]

@martin.schlipf brought to my attention that a similar question has been asked recently:
forum/viewtopic.php?f=4&t=18298

In a collinear magnetic moment calculation for tetragonal cell with in-plane axes (a; b=a) and out-of-plane axis (c>a), does the resulting magnetic moment always parallel to the c-axis (001)?

In a collinear calculation, there is no SAXIS because there is no coupling between the spin degrees of freedom and the lattice (spin-orbit coupling). Basically, this means that there is no difference in energy independently of the magnetization direction you choose. Magnetization is a scalar in the case of a non-collinear calculation.

If you want to look at preferential magnetization directions you need to include spin-orbit coupling and perform a non-collinear calculation (vasp_ncl).
That being said you can indeed either rotate the lattice and atomic positions or change the SAXIS in combination with the magnetization directions.