Hi,
For the mesh cutoff change the meshcutoff variable from the default of
100 Ry to
500 in steps of 100 Ry, if the energy converges (also take a look at
the forces) you
are done with the grid. If it is not converged you should continue
increasing the
mesh cutoff. If you reach a very huge value (800-1000 Ry) then you
should take a look at your
orbitals, neutral atom potential and non linear core corrections (if
exist in your
pseudo). Their FFT should give you a hint of the grid, as the energy/
forces don't
converge because of their wrap around errors (FFT aliasing). Remember
that
the square of the orbitals is what is projected into the grid.
For the k-points I recommend that you use the kgrid_cutoff described
in Moreno
and Soler PRB 45, 13891 (1992). It will calculate the best k-grid for
the symmetry of your
system. Start from some small value (around 2-3 Bohr) and continue
increasing it
until the energy is converged (or the calculation takes so much time
that is not feasible).
Best regards,
Eduardo
On 07/05/2008, at 8:52, Nidhi Sharma wrote:
Hi to all,
Will anybody please guide me how to perform the convergence test for
mesh-cut-off and k-point (Mohnkorst-Pack scheme)? For this what
changes should make in the .fdf file ?
Thanks,
Nidhi
Bring your gang together. Do your thing. Find your favourite Yahoo!
Group.
