How to fix local magnetic moment of each ions in VASP?

Question

In spin-polarized calculation,total magnetic moment can be fixed to a specific value using NUPDOWN tag. In non-colinear spin calculation, tag I_CONSTRAINED_M = 1 or 2 and can be used to contained direction and magnitude of magnetic moment. Neither of above two tag can fix local magnetic moment to fixed value. Is there any way to fix local magnetic moment to specific value? or Kindly suggest some changes in INCAR given below

POSCAR
3.50
1 0 0
0 1 0
0 0 1
  Fe
  4
Direct
0 0 0
0.5 0.5 0
0.5 0 0.5
0 0.5 0.5

SYSTEM = FCC
ISTART=0         # start new job
ICHARG=2         # 2 for atom
INIWAV=1         # initial wave function
ENCUT= 400.00 eV # Energy cut-off for calculation
LREAL= Auto   # real space projection .FALSE.
LCHARG= .FALSE  # no charge file output
ALGO=F
PREC=A
NCORE=8
KPAR=4
MAGMOM=0 0 1 0 0 -1 0 0 1 0 0 -1
ISYM=-1
LWAVE= .FALSE    # no wavefile output
ISMEAR = 1;     # tetrahedron method with Bloech 1 correction
SIGMA = 0.1      # smearing value
ISPIN=2         # Spin Polarisation
NELM=200         # Maximum number of iteration
NELMIN=3         # Minimum number of iteration
NSW=300          # number of step for ionic update
EDIFF=1E-4       # stopping criteria for electronic update
EDIFFG=-0.02
ISIF=3
IBRION=2
LORBMOM = .TRUE.
#NUPDOWN=0
I_CONSTRAINED_M=2
LNONCOLLINEAR = .TRUE.
M_CONSTR= 0 0 1 0 0 -1 0 0 1 0 0 -1
#ADDGRID=T
LORBIT=11

OUTPUT OSZICAR

   2 F= 0.65080119E+03 E0= 0.65080992E+03  d E =0.678114E+03  mag=     0.8391    -0.2230    -1.7841
   3 F= -.31196102E+02 E0= -.31191802E+02  d E =-.388330E+01  mag=    -5.0040    10.2547     2.8772
   4 F= -.31256210E+02 E0= -.31248490E+02  d E =-.394341E+01  mag=    -5.0955    10.5249     2.6127
   5 F= -.31250190E+02 E0= -.31243811E+02  d E =-.393739E+01  mag=    -5.0849    10.5771     2.5370
   6 F= -.31248427E+02 E0= -.31241528E+02  d E =-.393563E+01  mag=    -5.0749    10.5611     2.5074
   7 F= -.31256719E+02 E0= -.31250763E+02  d E =-.394392E+01  mag=    -5.1238    10.5592     2.4966
   8 F= -.31241928E+02 E0= -.31235361E+02  d E =-.392913E+01  mag=    -5.1036    10.5743     2.4683
   9 F= -.31250870E+02 E0= -.31243907E+02  d E =-.393807E+01  mag=    -5.1712    10.5782     2.3745
  10 F= -.31230308E+02 E0= -.31233397E+02  d E =-.391751E+01  mag=    -4.8791    10.5239     2.3687
  11 F= -.31234520E+02 E0= -.31229874E+02  d E =-.392172E+01  mag=    -4.8841    10.4645     2.4477
  12 F= -.30748687E+02 E0= -.30743533E+02  d E =-.343589E+01  mag=    -4.8649    10.4377     2.2561

```

Answer 1

As far as I am aware, this is simply not implemented in VASP. This is a tricky thing for codes to do correctly, so you will need to just converge to the correct state. You may overestimate some magnetic moments so they relax to their correct state. You can also try a range of NUPDOWN.

Its possible that you should be using a different approach though. Fixing all magnetic moments is something you really shouldn't be doing outside of debugging calculations.