[Wien] why Fe spin at [A] site is not showing negative value than that in [B] site in a ferrimagnetic calculation?

[Stefaan Cottenier]

>        I am doing a ferrimagnetic calculation for [A][B]O4 type material
> in which I have flipped the spin of Fe atoms at [A] site. I have
> modified case.inst file after input file generation (just before giving
> the number of K points) and flipped the spin of Fe atoms. Then as usual
> I go for dstart -up/dn and then run a spin polarize calculation. But
> after 24 cycles (it is still running) when I checked Fe moments at [A]
> site, *the moments are positive !! (although different from that in [B]
> sites)*. I have checked again the case.inst file and see that it has
> sticked to the modified form, i.e., Fe spins are opposite in [A] site
> than it is in [B] site. *Then why it is giving positive moments instead
> of negative?*..whether I have done something wrong in the steps?
> **Any response in this regard will be very helpful for us.

This might either be physics or a user problem.

Check how your moments evolved throughout the scf-cycle, and especially 
how they were in the first iteration. Did the sign originally correspond 
to what you have put in case.inst, and did it later gradually flip to 
the undesired sign? In that case, it is physics: your material does not 
'like' a ferrimagnetic solution. In case your moment had already the 
undesired sign in the very first iteration, you probably did something 
wrong. From your description, you seem to have changed case.inst *after* 
lstart had run, and therefore your changes had no effect. In recent 
wien2k versions, you can specify the moment (and its sign) atom by atom 
during init_lapw, before lstart is run.

Stefaan