Moments Disappear in Some Constrained Calculations

[HSims]

I am running noncollinear constrained moment calculations with VASP 4.6.28, and in order to convince myself that everything is as it should be, I decided to run three tests: one in which the moment lie parallel along the x direction, one along the y direction, and one along the z (the direction of the supercell is along the x [i.e. there are two unit cells in the x direction]). In the x direction, the calculation runs as it should, yielding good energies and moments. However, in the y and z directions, the calculations yield no moment in any direction, and VASP finds a much higher energy for these calculations.

Does anyone know what might cause this? I had initially suspected that it might be some issue in the CHGCAR (I typically run a collinear calculation to get a CHGCAR before starting a noncollinear one), but even without reading a CHGCAR the moments vanish. I have not been using the WAVECAR at all.

[admin]

I fyou have done everything correctly, this means that the easy direction of magnetization is along x.