[Wien] fixed-spin-moment and normal spin-polarized calculations

[carl]

Dear Wien2k Users

For Mn substitutional doped AlN supercell system, when I tried to compare 
the total energy from the fixed-spin-moment (FSM) scheme with the one from 
normal spin-polarized calculations, I observed quite weird behavior:

For the normal spin-polarization calculations, the calculation runs well 
and the produced result is as expected---which gives a 4 mu_B/cell, and the 
total energy is -11357.236702 Ry.

However, if I use the FSM method with the fixed moment = 4 mu_B/cell, the 
calculation looks (only looks!) good, that is, the energy and the charge 
density converged very well, and without any suspicious signals. (in 
*in2cup, the number of spin-up electron is 232, and in *in2cdn, the spin-dn 
electron number is 224 --- which are as expected). But the converged total 
energy is -11356.396458 Ry!  Moreover, the converged total magnetic moment 
is 5.196 mu_B, not the expected 4 mu_B!

Note in these two calculations, I used same structure, same parameters. The 
other difference is that for normal s-p using command "runsp_lapw -cc 
0.00001 -i 40"
and for the FSM using
"runfsm_lapw -m 4 -cc  0.00001 -i 40"

Also I have tried the several bcc-Fe using FSM with different fixed moments 
--- FSM works fine, I got the expected magnetic moments.

Now I got really puzzled --- it seems that the current FSM is not capable 
for Mn-doped AlN case? --- FSM should be rather straightforward .

I attached the supercell *.struct file for your inspection.

Many thanks in advance and best wishes to you all,
Carl CUI

School of Physics,
The University of Sydney, Australia 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: AlN-Mn-fsm.struct
Type: application/octet-stream
Size: 6608 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20051110/8d0a4636/AlN-Mn-fsm.obj