K point sampling

[jerry]

Dear,
    I want to know how to sample K points in a two or three probe (with gate electrode) system. Is the K point sampling is only for the electrode? The central region is handled using non-equilibrium Green function? Can you give me some examples for three dimensional systems rather than one dimensional (atomic chain)?
   If the K point in the new calculator is set as 2*2*10(along a, b, c respectively). How to set the K point in the transmissioneigenvalues? Is the K point in the transmissioneigenvalues fractional or integral?
 Thanks a lot!

[zh]

You can refer to the tutorial of Fe/MgO/Fe. In that tutorial, the k-dependent transmission spectra and the convergence of transmission coefficients with respect to the k-point sampling are demonstrated.

[Anders Blom]

kC (or kz) is only for the electrode, kA and kB are for the whole calculation, these are good quantum numbers, thus an electron entering from the left electrode with a given (kA,kB) will exit in the right electrode with the same k-vector.

As discussed in many places in the Forum, you need to have enough k-points in A/B to sample the electron density well enough. How many that is, depends on the system and its size. It should be studied systematically.

The k-point to be given for analysis quantities like TransmissionEigenvalues is to be specified as a specific point (kA,kB) in fractional reciprocal coordinates, not as a sampling. It's one point, in the interval 0.5<=kA,kB<=0.5, as appropriate.