WARNING: Sub-Space-Matrix is not hermitian in DAV

[dcosta]

Hi,
I'm trying to calculate Fe-Cr interface formation energy. I'm using supercells composed by a superposition of Fe bcc cells on Cr bcc cells in the direction orthogonal to the interface, so I deal with a column of pure Cr bcc single cells superposed to a column of pure Fe bcc single cell. For a relatively large number of ions (~52/60 , half Fe and half Cr) I cannot achive the vasp run, the problem which usually occurs is:

WARNING: Sub-Space-Matrix is not hermitian in DAV

I've not the same problem if I look at a superposition of pure Fe (or Cr) bcc single cells, so i guess it should have something to do with ions relaxation algorithm.
I'm using vasp.4.6.28, my INCAR an POSCAR files for a 52 ions Fe-Cr supercell look like these:

System = Fe-Cr
LWAVE = .FALSE.
LREAL = .FALSE.
IALGO =38 !I tried with IALGO = 48 as well

ISPIN = 2
MAGMOM =27*2.8 13*-2.8 12*2.8

PREC = med
IBRION = 3
NSW = 45
ENMAX = 300
ISMEAR = 1
SIGMA = 0.3
______
FeCr
2.831
1 0 0
0 1 0
0 0 26
27 25
d
.00000000 .00000000 .00000000
.49999999 .50000000 .01923077
.00000000 .00000000 .03846154
.50000000 .50000000 .05769231
.00000000 .00000000 .07692308
.50000000 .50000000 .09615386
.00000000 .00000000 .11538462
.50000000 .50000000 .13461539
.00000000 .00000000 .15384616
.50000000 .49999999 .17307693
.00000000 .00000000 .19230770
.50000000 .50000000 .21153846
.00000000 .00000000 .23076923
.50000000 .50000000 .25000000
.00000000 .00000000 .26923078
.49999999 .50000000 .28846154
.00000000 .00000000 .30769232
.50000000 .50000000 .32692307
.00000000 .00000000 .34615386
.50000000 .50000000 .36538461
.00000000 .00000000 .38461539
.50000000 .50000000 .40384614
.00000000 .00000000 .42307693
.50000000 .50000000 .44230768
.00000000 .00000000 .46153846
.50000000 .50000000 .48076922
.50000000 .50000000 .98076957
.00000000 .00000000 .50000000
.00000000 .00000000 .53846157
.00000000 .00000000 .57692313
.00000000 .00000000 .61538470
.00000000 .00000000 .65384626
.00000000 .00000000 .69230783
.00000000 .00000000 .73076940
.00000000 .00000000 .76923096
.00000000 .00000000 .80769253
.00000000 .00000000 .84615409
.00000000 .00000000 .88461566
.00000000 .00000000 .92307723
.00000000 .00000000 .96153879
.50000000 .50000000 .51923078
.50000000 .50000000 .55769235
.50000000 .50000000 .59615391
.50000000 .50000000 .63461548
.50000000 .50000000 .67307705
.50000000 .50000000 .71153861
.50000000 .50000000 .75000018
.50000000 .50000000 .78846174
.50000000 .50000000 .82692331
.50000000 .50000000 .86538488
.50000000 .50000000 .90384644
.50000000 .50000000 .94230801
_____

Thank you

[admin]

Usually this error indicates an unreasonable charge density
please check if this error shows up at the first or some later ionic step.
1) first step: please check your input geometry (eg the interatomic distances as written in OUTCAR)
2) at some later step: please check the relaxation history:
have all ionic steps been converged electronically?
does the total energy decrease smoothly?
it may help to decrease POTIM and the mixing parameters (AMXIX,BMIX, AMIX_MAG and BMIX_MAG)