how to sample the k-points?

[vboyz]

In the ADVANCED TUTORIAL of "Graphene nanoribbon", it use a 1*1*100 k-points to calculate a armchair GNR. In my opinion, 1*1*100 is suitable for the 1D system? So why use such a k-point sampling in a 2D system? And how should we sample the K-points?
BTW, do you compare the electronic calculation to other DFT software such as Win2K?

[Nordland]

If you ask me, a 1,1,100 for a 1D system is the perfect choice, it is as good as it gets. Normally no reason for having more k-points in the z-axises than 100.

For 2D systems and 3D it depens alot on geometry, but one thing is sure, you will need kpoints in the other directions when you increase the dimensionality of the system.

So in a graphene sheet, I would use 1,20,100 and in a typical FeMgOFe junction I would use 6,6,25.

Usual prior to starting out on a new project, I always compare the bandstructure with one calculated by Abinit, and normally the agreement is pretty good and then I don't more comparisons for this study.