Page 1 of 1

ngkpt for geometry optimization

Posted: Mon Feb 25, 2019 2:32 pm
by mdsimulationgroup

I'm very new to planewave calculations; all of my experience is gaussian basis sets. I'm trying to get a feel for the kpoint grid. If all I want to do is geometry optimization of a slab of copper with a molecule absorbed to the surface, how should I be setting my ngkpt? Is it out to have a sparse grid such as:

ngkpt 1 1 1

Or should I have more points?


Re: ngkpt for geometry optimization

Posted: Tue Feb 26, 2019 3:52 pm
by ebousquet
Dear Stacey,
When doing a slab, you have one direction that is very large with vacuum (probably the z direction), meaning that your reciprocal space is very small and then the number of k-point can be set to 1 in this direction to have good precision. In the other directions, you need to test the convergence of the k-point grid, i.e. check how your result change w.r.t. number of k-point along x and y (it will depend if you are metallic or not and what is the size of the cell in these directions).
Best wishes,