Page 1 of 1

In the case of ISIF=3, different free energy

Posted: Fri Dec 02, 2011 7:11 am
by gdlee
I used ISIF=3 to relax stress and unit-cell. After the finish of the calculation, I calculated again with the final unit-cell and atomic positions in ISIF=0. Then, the final free energies are different each other. I don't know reason why. Which free energy is right ? I guess the difference comes from symmetry. But in both calculation, I used ISYM = 0. Could any master let me know the origin of the difference?
These are INPUTs in my calculation.
INPUT in first calculation
Startparameter for this Run:
INIWAV = 1
ISTART = 0
LREAL = .TRUE.
ENCUT = 286.567
Electronic Relaxation 1
EDIFF = 1E-4
ICHARG = 2
NELM = 200
Ionic Relaxation
NSW = 550
IBRION = 2
ISIF = 3
POTIM = 0.30
ISYM = 0 0-nonsym 1-usesym 2-fastsym
DOS related values:
ISMEAR = 0;
SIGMA = 0.07
.....................................
INPUT in second calculation
Startparameter for this Run:
INIWAV = 1
ISTART = 0
LREAL = .TRUE.
ENCUT = 286.567
Electronic Relaxation 1
EDIFF = 1E-4
ICHARG = 2
NELM = 200
Ionic Relaxation
NSW = 550
IBRION = 2
ISIF = 0
POTIM = 0.30
DOS related values:
ISMEAR = 0;
SIGMA = 0.07
~

In the case of ISIF=3, different free energy

Posted: Mon Dec 05, 2011 8:11 am
by tlchan
You can refer to the VASP manual for a detailed explanation.

In short, during optimization of the lattice vectors, the reciprocal lattice vectors of the plane wave basis will change. As a result, the energy cutoff is changed effectively. However, VASP does not redefine the plane wave basis during the optimization. You get different total energies for the two calculations because they correspond to slightly different energy cutoff.