energy difference of the elastic constant calculation

[vasp2010]

hi everyone
I am a vasp novice ! when I calculate the elastic constant of Ce2Zr2O7 pyrochlore on the basic of stress - energy relation。I find some slight deformation and render great difference of energy。However, I do not kown the reason.

slight deformation total energy (ev)

0.02 -199.542615
0.015 -199.722497
0.010 -199.882383
0.005 -199.933486
0 -199.959135
-0.005 -199.933779
-0.010 -199.856072
-0.015 -198.644631
-0.02 -198.987394
I find the slight deformation -0.015 and the energy is 198.644631ev . I initial guess that the energy should be close to 199.722ev. Deformation -0.02 should nearly be -199.54ev . I make some energy calculation where those deformations are -0.016 -0.018 .the results are :
Deformation total energy
-0.016 -199.695020
-0.018 -199.625938
Surprisingly, The difference between -0.015 and -0.016 is negligible, but the discrepancy of energy is tremendous (-198.644631 and -199.625938).
If someone could account for the discrepancy, I will highly appreciate your assistance.
About my calculation information:
pyrochlore Ce2Zr2O7 : space group：Fd3m face-centered cubic lattice
I calculate this in the primitive cell (22 atoms).

POSCAR (without deformation ):

system=cpp
10.86589395135278029436
0.0000000000000000 0.5000000000000000 0.5000000000000000
0.5000000000000000 0.0000000000000000 0.5000000000000000
0.5000000000000000 0.5000000000000000 0.0000000000000000
4 4 14
Direct
0.5000000000000000 0.5000000000000000 0.5000000000000000
0.5000000000000000 0.5000000000000000 0.0000000000000000
0.5000000000000000 0.0000000000000000 0.5000000000000000
0.0000000000000000 0.5000000000000000 0.5000000000000000
0.0000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 0.5000000000000000
0.0000000000000000 0.5000000000000000 0.0000000000000000
0.5000000000000000 0.0000000000000000 0.0000000000000000
0.9161878673778846 0.3338121326221152 0.3338121326221152
0.3338121326221152 0.9161878673778846 0.9161878673778846
0.3338121326221153 0.9161878673778846 0.3338121326221153
0.9161878673778846 0.3338121326221153 0.9161878673778846
0.3338121326221153 0.3338121326221153 0.9161878673778846
0.9161878673778846 0.9161878673778846 0.3338121326221152
0.0838121326221153 0.6661878673778846 0.0838121326221153
0.6661878673778846 0.0838121326221153 0.6661878673778846
0.6661878673778846 0.0838121326221153 0.0838121326221153
0.0838121326221153 0.6661878673778846 0.6661878673778846
0.6661878673778846 0.6661878673778846 0.0838121326221153
0.0838121326221153 0.0838121326221153 0.6661878673778846
0.3750000000000000 0.3750000000000000 0.3750000000000000
0.6250000000000000 0.6250000000000000 0.6250000000000000

POSCAR (-0.015 of deformation)

system=cpp
10.865341916374798
-0.0150000000000000 0.4926130000000000 0.4926130000000000
0.4926130000000000 -0.0150000000000000 0.4926130000000000
0.4926130000000000 0.4926130000000000 -0.0150000000000000
4 4 14
Direct
the same as above 22 atoms
………………………….
POSCAR (-0.018 of deformation)

system=cpp
10.865341916374798
0.0180000000000000 0.5091620000000000 0.5091620000000000
0.5091620000000000 0.0180000000000000 0.5091620000000000
0.5091620000000000 0.5091620000000000 0.0180000000000000
4 4 14
Direct
the same as above 22 atoms
………………………….
INCAR:
SYSTEM=cpp
ISTART=0
ICHARG=2
ISIF=2
NSW=60;IBRION=2
NELM=80
ISMEAR=0
SIGMA = 0.05
NBANDS=250
LREAL=.TRUE.
ISPIN=2
ENCUT = 900
PREC=Accurate
EDIEF=1E-5
EDIEF=-1E-2
POTIM=0.2

NELMDL=3
LMAXMIX=6

LWAVE=.FALSE.

LDAU = .TRUE.
LDAUTYPE = 2
LDAUL = 3 -1 -1
LDAUU = 5.0 0.00 0.00
LDAUJ = 0.0 0.00 0.00
LDAUPRINT = 2

IALGO = 48
NSIM =4
LPLANE = .TRUE.
LSCALU=.FALSE.

KPOINTS:
Automatic mesh
0
Monkhorst - pack
6 6 6
0.0 0.0 0.0

[alex]

Hello vasp2010,

welcome to the funny world of f-states! Your system probably ended up with different states/occupations/localisation in the two cases. This does not necessearily mean different magnetisation, but could also be different degree of localisation (check e.g. alpha vs gamma Cerium (metal)).
Go to the INCAR website and check how you can manipulate spin states of the system.

Cheers,

alex