Magnetic moment in an antiferromagnet

[prasen]

I am trying o figure out how to get the magnetic moment on each atom in an antiferromagnet. "Magnetization" is written out at two places: One after each step in the SCF cycle:
----------------------------------------- Iteration 1( 1) ---------------------------------------


POTLOK: VPU time 28.60: CPU time 28.61
SETDIJ: VPU time 0.30: CPU time 0.30
EDDAV : VPU time 76.99: CPU time 77.02
DOS : VPU time 0.00: CPU time 0.00
------------------------------------------
LOOP: VPU time 105.89: CPU time 105.93

eigenvalue-minimisations : 144
total energy-change (2. order) : 0.5255710E+02 (-0.2570726E+03)
number of electron 26.0000000 magnetization 0.0000000
augmentation part 26.0000000 magnetization 0.0000000

... etc
But this is zero in case of AFM.

The other place it is written is after each ionic step:

magnetization (x)

# of ion s p d tot
----------------------------------------
1 0.120 0.081 3.371 3.571
2 -0.116 -0.079 -3.370 -3.565
3 -0.012 -0.017 0.000 -0.029
4 -0.014 -0.019 0.000 -0.033
5 -0.010 -0.018 0.000 -0.028
6 -0.024 -0.015 0.000 -0.040
7 -0.022 -0.015 0.000 -0.037
8 -0.022 -0.015 0.000 -0.037
9 -0.013 -0.017 0.000 -0.029
10 -0.020 -0.015 0.000 -0.035
11 0.016 0.019 0.000 0.035
12 0.022 0.016 0.000 0.038
13 0.019 0.015 0.000 0.033
14 0.025 0.015 0.000 0.039
15 0.017 0.018 0.000 0.035
16 0.015 0.016 0.000 0.031
17 0.011 0.018 0.000 0.028
18 0.023 0.015 0.000 0.038
------------------------------------------------
tot 0.014 0.000 0.001 0.015

In which the angular momentum decomposed magnetic moment on all the atoms are printed out. But are these numbers correct? . Because, in case of a ferromagnet, these atomic moments do not add up to the total magnetization given at the end of each SCF cycle (and which seems more reasonable for known systems).

Where does one find the correct magnetic moments on atoms in AFM systems?

[forsdan]

This issue have been discussed many many times. The magnetization values are based on the charge inside the RWIGS radius. You can change the RWIGS values to account for more charge but you can never fill the space with spheres without double counting. However even if you choose the radii in such a way that they don't overlap, the division into which charge density belongs to which atoms is not unique. Therefore the magnetization for each atom is not unique. Search the forum for further discussions.

Best regards,
/Dan Fors
<span class='smallblacktext'>[ Edited Tue Jul 15 2008, 09:37PM ]</span>