Ionic Contribution to Dielectric Tensor and LREAL

[flowjm]

Hello all,

I'm attempting to calculate the ionic contribution to the dielectric tensor for the tetragonal phase of ZrO2 using VASP vasp.5.2.12. However, I get very different values for two of the components depending upon whether the LREAL parameter is set to AUTO or FALSE:

With Auto:

MACROSCOPIC STATIC DIELECTRIC TENSOR IONIC CONTRIBUTION
---------------------------------------------
136.788 -0.000 0.002
-0.000 136.815 0.002
0.002 0.002 13.398

With False:

MACROSCOPIC STATIC DIELECTRIC TENSOR IONIC CONTRIBUTION
---------------------------------------------
38.867 -0.000 -0.003
-0.000 38.871 -0.003
-0.003 -0.003 14.954

Everything else is identical: POTCAR, INCAR, KPOINTS, POSCAR. Has anyone experienced a similar problem, and/or can shed any light on what might be going on?

Many thanks in advance.

[flowjm]

Here's the contents of the INCAR file:

ALGO = Fast
AMIX = 0.2
BMIX = 0.0001
AMIX_MAG = 0.8
BMIX_MAG = 0.0001

IBRION = 7
!NSW = 20

NELM = 30
ISYM=2
EDIFF=1E-05
EDIFFG = -0.01
LREAL=.FALSE.
PREC=High
ENCUT=500

LEPSILON=.TRUE.

LORBIT = 10

ISMEAR = 0
SIGMA= 0.05