QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: mephisto3142001 on March 13, 2012, 15:56

Title: Question about k-point Sampling and Monkhorst Pack Grid method
Post by: mephisto3142001 on March 13, 2012, 15:56

Greetings, I have three questions concerning k-points sampling.

1.As we set parameters for "k-point Sampling", we type three integers "(nA, nB, nC)". If I typed (3, 3, 50), whether it means that I want   
  the whole Brillouin zone to be sampled by 3x3x50=450 k-points? 

2.Does Monkhorst-Pack-Grid method calculate reciprocal lattice vector of each k-point in the whole Brillouin zone via the formula used in "
  Phys. Rev. B., 13, 5188, 1976" ? If I typed (3, 3, 50), can I say that I have Monkhorst-Pack-Grid method generate reciprocal lattice 
  vectors for 450 points?

3.When I calculate TransmissionSpectrum, why can k-points along the C-reciprocal direction be ignored?

Title: Re: Question about k-point Sampling and Monkhorst Pack Grid method
Post by: Anders Blom on March 13, 2012, 21:50

1. Yes. Although at least about half of these can be reduced in the calculation due to time-reversal symmetry.

2. Yes. Try

Code: python

sampling = MonkhorstPackGrid(3,3,5)
print sampling.kpoints()
sampling = MonkhorstPackGrid(3,3,5,force_timereversal=False)
print sampling.kpoints()

(5 instead of 50 so you can actually see the output properly.)

3. Devices are not periodic in C (Z) hence it makes no sense to have any k-point sampling in this direction (or rather, 1 point is always enough). The electrodes, however, are periodic, and that's why you have 50 or 100 points for the electrode calculation.