[Wien] magnetizatin direction in case.inso vs case.inst

[pluto]

Dear Prof. Blaha, dear All,

I would like to make a collinear magnetic+SOC calculation that simulates 
the sample magnetized in opposite directions along some easy axis. I am 
looking both at ferromagnetic and ferrimagnetic samples, where only some 
atoms are magnetic. Could you please tell me the correct procedure.

One can set the direction of magnetization in case.inso and in 
case.inst.

One can see sizes of magnetic moments through:

grep :MM case.scf

How are all the three things (inso, inst, and scf) related?

I would assume that case.inso only sets the quantization axis (but does 
not physically change the directions of magnetic moments). Therefore, 
for two identical calculations but with inso at [001] and [00-1], the 
result should be exactly the same in energy, but the spin up/dn weights 
of orbital characters should be swapped. Spin-integrated (adding up+dn) 
weights of the Ylm orbitals should not change in such case.

But then what is the precise meaning of the u/d/n when producing the 
case.inst using instgen_lapw? Are those, for the SOC calculation, 
ultimately referring to the quantization axis set in case.inso? I mean, 
would setting "up" in case.inst mean the moment oriented in real space 
along the direction set by instgen_lapw in case.inso?

Also, what do the magnetic moments in case.scf mean? They have signs, so 
they must refer to some quantization axis.

Best,
Lukasz