ABINIT Discussion Forums

Dear ABINITers,

I would like to automize the convergence study with respect to k points. I would also like ABINIT to choose the best grid for a given precision, controlled by ONE parameter like kptrlen. The problem is that the k-point grid must contain gamma, in order to be able to calculate the GW correction for gamma. Therefore I cannot simply use kptrlen. Am I wrong?

Question A: Is there a way to auto-generate k-point grids that contain gamma, similar to the use of kptrlen?

Question B: Should I start with a different (more efficient) grid for the ground state run and then switch to the one containing gamma? How can this be done?

Question C: How can I set istwfk for all k-points, when I do not know the number of k-points?

Best,

Martin Haeufel
TU Munich, WSI (T33)
http://www.wsi.tum.de

Dear Martin,

Martin wrote:Question A: Is there a way to auto-generate k-point grids that contain gamma, similar to the use of kptrlen?

No. But you might modify the routine testkgrid.F90 to suit your needs ...

Martin wrote:Question B: Should I start with a different (more efficient) grid for the ground state run and then switch to the one containing gamma? How can this be done?

You might indeed combine two different grids for separate SCF and GW calculations.
Well, you have to do it by using different datasets.

Martin wrote:Question C: How can I set istwfk for all k-points, when I do not know the number of k-points?

This is easy :
istwfk *1
See the description in http://www.abinit.org/documentation/hel ... tml#intro3

Best,
X