Groundstate Structure determination

[bruent]

I am thinking of determining the ground state structure of PZTN while N% is 20%, whose structure hasn't been reported by experiement. Thus I plan to start from different structures, and relax the lattice parameter and ions. However, I cannot find a proper ISIF to realize it, say, fix the cell shape and relax ions and volumes. Anyone has any idea? Thanks

[admin]

please have in mind that VASP keeps the symmetry type of the lattice of the input POSCAR file (i.e., a cubic lattice remains cubic, a tetragonal lattice remains tetragonal,...)
for example: 'change the cell shape and the volume ' (IBRION = 3) of a tetragonal lattice will relax c/a, and V, but will never lead to an orthorhombic lattice or any other lattice type of lower symmetry (this, in some sense, fixes the cell shape).

Hence you can choose 2 ways to find the ground state geometries of your input structures:

1) do it automatically, but allow for a change in the shape
ISIF=3 #changes volume, cell shape, ionic positions
PREC= High #necessary to obtain stress tensors of the required accuracy
NSW= #number of steps
IBRION = 2 (or 1)

2) do it 'by hand' by fitting the energy parabola of a set of calculations of
cells of equal shapes, but different volumes
ISIF = 0, 1 or 2
PREC = High
NSW = ...
IBRION = 2 (or 1)

[Tyx]

" please have in mind that VASP keeps the symmetry type of the lattice of the input POSCAR file (i.e., a cubic lattice remains cubic, a tetragonal lattice remains tetragonal,...)
for example: 'change the cell shape and the volume ' (IBRION = 3) of a tetragonal lattice will relax c/a, and V, but will never lead to an orthorhombic lattice or any other lattice type of lower symmetry (this, in some sense, fixes the cell shape). "

Unless ISYM=0, right? Also, what are the dangers in setting ISYM=0, apart from VASP using more k-points than the irreducible set of your structure, and thus taking longer?
(when only interested in the total energy).

Thanks


<span class='smallblacktext'>[ Edited Tue Apr 18 2006, 06:46PM ]</span>

[admin]

yes, you are right, sorry for beeing not precise in that point. There are no dangers in setting ISYM=0, the only effect may be that the calculation takes longer because the complete BZ is taken instead of the irreducible part.