Inconsistency in the WAVECAR plane wave coefficients

[emilhfr]

I am presently doing calculations on a ferrous material which i suspect to exhibit non-collinear magnetic ordering.

My approach is standard fare; i calculated a non-magnetic ground state (ISPIN = 1), and used this WAVECAR and CHGCAR as the initial state for the non-collinear arrangements. For the most part, this works fine.

However, while trying the various possible magnetic orderings, i encountered a strange issue: For a certain arrangement of initial MAGMOM values, i get the error message "ERROR: while reading WAVECAR, plane wave coefficients changed", and the calculation fails.
There is no difference between this calculation and the others which work normally, except for the MAGMOM. Is this a bug, or something i don't know about?

Here is the INCAR i use. The hashtag-ed MAGMOM is an example of a MAGMOM that works fine. The not-hashtag-ed MAGMOM breaks the calculation. Other MAGMOMS with the same direction but different magnitude of the MAGMOM vector also breaks the calculation.

SYSTEM = FerrousArrangement

LREAL= Auto

#ISTART = 0

ALGO = Normal

ISMEAR = 0
SIGMA = 0.02
LORBIT = 11

ICHARG = 1

ENCUT = 500.0
EDIFF = 1E-6
NELM = 100

EDIFFG = -1E-02
NSW = 0
IBRION = -1
ISIF = 2
ISPIN = 2

Mixer
AMIX = 0.2
BMIX = 0.00001
AMIX_MAG = 0.8
BMIX_MAG = 0.00001

GGA+U
LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = -1 -1 2 -1
LDAUU = 0.00 0.00 3.70 0.00
LDAUJ = 0.00 0.00 0.00 0.00
LDAUPRINT = 2
LMAXMIX=4


LNONCOLLINEAR = .TRUE.
MAGMOM = 72*0 36*0 0 6 6 0 -6 6 0 6 6 0 -6 6 0 6 6 0 -6 6 0 6 6 0 -6 6 84*0
#MAGMOM = 72*0 36*0 0 6 6 0 6 -6 0 6 6 0 6 -6 0 -6 -6 0 -6 6 0 -6 -6 0 -6 6 84*0


NCORE = 4

[ferenc_karsai]

Can you please upload all necessary input and output files? POSCAR, POTCAR, KPOINTS, OUTCAR and INCAR. Please also upload this set of files used for the previous standard calculation.
We need to try to reproduce every step that you made.

[emilhfr]

Uploading the complete files would be problematic. It is a phase that was recently discovered and it has not been reported yet, so i can't really put the exact composition out prior to publication.

[ferenc_karsai]

In that case please send the files to support@vasp.at.