I am trying to do constrained magnetic moment calculation, and as far as I've figured so far, in order to do that I have to do noncollinear spin polarized calculation. Bellow is the INCAR file for the collinear calculation, kind of which I perform routinely and it always converges in 20 iterations or so.
Code: Select all
INCAR
ENCUT=300.000000
BMIX=0.000100
AMIX_MAG=0.800000
BMIX_MAG=0.000100
AMIX=0.200000
SIGMA=0.100000
EDIFF=1.00e-06
PREC=Normal
LORBIT=11
NPAR=4
NSIM=4
LPLANE=.TRUE.
LWAVE=.TRUE.
LCHARG=.TRUE.
LREAL=Auto
NELM=200
ISPIN=2
MAGMOM=4*0 1 1 -1 -1 1 1 -1 -1 8*0
Now, for non-collinear calculations I have just converted magnetic moments to 3D vectors and added LNONCOLLINEAR = .TRUE. so that instead of last two lines in the above INCAR file I have
Code: Select all
LNONCOLLINEAR=.TRUE.
MAGMOM=12*0 1 0 0 1 0 0 -1 0 0 -1 0 0 1 0 0 1 0 0 -1 0 0 -1 0 0 24*0
This calculation converges also in around 20 iterations, but magnetic moments are 0. So I have tried following
Code: Select all
LNONCOLLINEAR=.TRUE.
MAGMOM=12*0.01 1 0 0 1 0 0 -1 0 0 -1 0 0 1 0 0 1 0 0 -1 0 0 -1 0 0 24*0.01
Now, this calculation does not converge. After 7 or 8 iterations dE goes down to 1e-3 or 1e-4 and does not decrease any more. When NELM iterations are executed output shows magnetic moments equal to what I had in the collinear case, but convergence is not full.
Does anyone have any advice?
<span class='smallblacktext'>[ Edited ]</span>
