I have read the paper of Monkhorst-Pack method and the C-C method,and in my understanding the shift of k-points in Monkhorst-Pack method with the parameter Ni which chose to be even, is to avoid the gamma point or the k-points on the boundary,and the point of gamma and the points on the boundary is high symmetry point,so there is degeneracy,and the degeneracy means extreme points,and the M-P method is take the mean value,so it should be avoid .but sometimes the shift just set to 0(just like in the QE scf namelist ),and if i do this,the gamma point and the points on the boundary which the M-P method should avoid is just take into account,what the meaning of it?
The Monkhorst-Pack set of "special points" is nothing but a regular grid of points in reciprocal space, and seeing it as such is also the best way to understand its properties. In particular, it is simple to convince oneself that shifting the origin of the grid while keeping the density of points (and hence the accuracy of the sampling) fixed allows you to save some points in the calculation, when all the symmetries are carefully exploited. The gain becomes negligibly small when the number of points grows large. In summary, you *should* set the offset to zero when (for one reason or the other) you need to know the electronic states at zone center. You *could* set it to zero without any significant loss of performance when the number of points is large (and the density of points in the Brillouin zone accordingly large).
for i dont know how to use the researchgate,it takes too long to thank you.but i am a new learner ,so i still cant understand when should i know around the zero center ,such as i just want to know the phono spectrum ?and is there any other situations?
- Badges
- Science topic