Oscillations when using LCALCEPS tag

[rohan maniar]

Dear VASP Community,

I am a new user of VASP. I recently attempted to compute the static dielectric constant of NiO at its experimental geometry. As per the manual, there appear to be two ways to do this- (1) using DFPT (LEPSILON) and (2) response to applied fields( LCALCEPS). I can run the calculations using the LEPSILON tag with no issues. However, I see large oscillations in the DMP block of my output file when running the LCALCEPS calculation.

....................................................................................................................................................................................
Section of output
.....................................................................................................................................................................................

DMP: 147 -0.233881481206E+02 0.10210E-04 -0.63524E+00 12000 0.236E+01-0.150E+01
p_tot=( 0.158E-02 0.586E-02 0.177E-02 )
dp_tot=( 0.158E-02 0.586E-02 0.177E-02 ) diag[e(oo)]=( --- 30.24413 --- )
gam= 0.519 g(H,U,f)= 0.237E+01 0.000E+00 0.000E+00 ort(H,U,f) =-0.150E+01 0.000E+00 0.000E+00
DMP: 148 -0.233881176330E+02 0.30488E-04 -0.63530E+00 12000 0.237E+01-0.150E+01
p_tot=( 0.502E-04 0.780E-03 -0.138E-03 )
dp_tot=( 0.502E-04 0.780E-03 -0.138E-03 ) diag[e(oo)]=( --- 4.89153 --- )
gam= 0.519 g(H,U,f)= 0.237E+01 0.000E+00 0.000E+00 ort(H,U,f) =-0.150E+01 0.000E+00 0.000E+00
DMP: 149 -0.233881077337E+02 0.98993E-05 -0.63540E+00 12000 0.237E+01-0.150E+01
p_tot=( 0.157E-02 0.585E-02 0.175E-02 )
dp_tot=( 0.157E-02 0.585E-02 0.175E-02 ) diag[e(oo)]=( --- 30.18597 --- )
gam= 0.519 g(H,U,f)= 0.237E+01 0.000E+00 0.000E+00 ort(H,U,f) =-0.150E+01 0.000E+00 0.000E+00
DMP: 150 -0.233880775922E+02 0.30141E-04 -0.63547E+00 12000 0.237E+01-0.150E+01
final diagonalization
p_tot=( 0.356E-04 0.764E-03 -0.153E-03 )
dp_tot=( 0.356E-04 0.764E-03 -0.153E-03 ) diag[e(oo)]=( --- 4.81351 --- )

............................................................................................................................................................................................

I have tried to alter the EDIFF, EFIELD_PEAD, and ISMEAR tags in the INCAR to no avail.

I have also attached my input and output files. Any clues on how I can fix this problem would be much appreciated!

Thanks,
Rohan

[ferenc_karsai]

I ran your calculations, but I can't reproduce your results. For me starting from your input (with and without previous WAVECAR, CHG etc.) the calculations look ok. I also saw you have many DMP steps (over 100, which is also very suspicious). Probably your VASP installation is not ok.
I've run VASP 6.4.3 on 32 cores using the gnu compiler 11.2 with mkl 2022.0.1 and open mpi 4.1.2. If you can please try this configuration or at least a similar one.

Here is a snippet of the final diagonilization steps in each direction:
DMP: 6 -0.235393910223E+02 -0.33296E-06 -0.90157E-06 4800 0.321E-05-0.184E-05
final diagonalization
p_tot=( -0.667E+02 -0.667E+02 -0.667E+02 )
dp_tot=( 0.155E-02 0.199E-04 0.198E-04 ) diag[e(oo)]=( 8.73131 --- --- )
...
DMP: 3 -0.235393402920E+02 -0.29661E-06 -0.42510E-06 4800 0.118E-05-0.234E-06
final diagonalization
p_tot=( -0.667E+02 -0.667E+02 -0.667E+02 )
dp_tot=( 0.116E-04 0.109E-02 0.114E-04 ) diag[e(oo)]=( --- 6.41816 --- )
...
DMP: 2 -0.235394117580E+02 -0.22892E-07 -0.41790E-06 4800 0.822E-06 0.429E-06
final diagonalization
p_tot=( -0.667E+02 -0.667E+02 -0.667E+02 )
dp_tot=( 0.771E-05 0.762E-05 0.762E-03 ) diag[e(oo)]=( --- --- 4.80437 )