03 March 2018 0 826 Report

Hi all,

I have a question about the KPOINT setting for non orthorhombic supercell.

I use Monkhorst-Pack sampling scheme. As far as I known, the number of k points is decided by the norm of reciprocal vector of supercell. I will show vector matrix in POSCAR of Mg supercell (32 atoms) first to explain my question:

New structure

1.0

12.8000001907 0.0000000000 0.0000000000

-6.4000000954 11.0851253336 0.0000000000

0.0000000000 0.0000000000 5.1999998093

For orthorhombic supercell, it is very easy to define the norm of three real vector, because the non-diagnosed element in matrix are zero. However, for a hexagonal supercell, which element in matrix should I refer to when setting the number of Kpoint?

Usually I don't use this matrix because the supercell has three vectors defining it already, and I set number of k points by checking these three vectors. In this case, I will treat norm of three vectors (12.8,12.8,5.2). However, should I use (12.8, 11.085, 5.2) instead of (12.8,12.8, 5.2) when considering number of k points?

I ask this because I find that, compared with orthorhombic supercell, the non-orthorhombic (e.g. hexagonal, rhombic) will be much easier to meet such problem:

please rerun with smaller EDIFF, or copy CONTCAR

to POSCAR and continue

Is it a inherent issue for VASP or I set the inappropriate number of k points?

Any answer will help. Thank you in advanced!

Similar questions and discussions