non-integer transmission possibility in zero-bias

[beark]

I calculated a 1-D homogeneous system under zero-bias and found
that some points in the transmission curves are not integer, while other
points are all nearly integer (1.999 etc) and match the band structure well.

The nearest distance of atoms in the near cells is larger than 15A. I then
enlarge the cell, the transmission possibility change from 3.39 to 3.5.
But it suppose to be 4. The unit cell I used is very large and my computer
's memory is nearly all used. Could anyone give some suggestions on how
to get the correct result?

[zh]

Do you check the k-point sampling? Is the size of k-grid large sufficient?

[Nordland]

It is quite important to know the length of the your electrode.
If it is 15 Ang long in the z-axis, then it is okay. If the electrode is too short,
it will give these kind of artificial scattering.

Alternative check the coordinates for any small errors - even the smallest error in coordinates
can and will cause scattering of the electrode.

When it comes to 1D system, I have not seen a case where doing one of the two above didn't solve the problem.

[beark]

I enlarge the length of the electrode and that solve the problem.
Thank U all!