Hi all,

I try to calculate the spin orientation of an electronic state (noncollinear calculation). The FatBandstructure class allowed me to project the band structure onto spin.Up, spin.Down, spin.X, spin.Y, and spin.Z but I'm not sure about how to interpret the returned values.

For each state the sum

spin.Up + spin.Down = 1

which is the normalization condition for the spinor. Apparently, the projections onto spin.Up and spin.Down return the absolute value squared of the respective components.

If I project onto x, y or z instead, I get positive and negative values and I therefore assume that these are the expectation values. So I would expect

sqrt( spin.X**2 + spin.Y**2 + spin.Z**2 ) = const.

because only the orientation of the spin changes for different states but not the absolute value. However, this is not the case and the length of the vector changes for different bands and k points.

Could someone please explain what the return value of the spin projections is?

Thanks.