ABINIT Discussion Forums

Hai

I am trying to obtain the accurate density of states (dos) of intrinsic graphene in which the density of states is zero at Fermi level set at 0 eV.
I got the dos plot of graphene, but it contains artificial wiggles i.e oscillations.
I found from the abinit tutorial that increasing the number of k points the oscillations can be reduced.
How can I increase the k points? For that as the first step, I need to calculate nkpt.
As the number of k points in the Brillouin zone is automatically generated by ngkpt, kptopt and shiftk, how to calculate the nkpt from the following input variables used:
kptopt 1
ngkpt 45 45 1
nshiftk 1
shiftk 0.0 0.0 0.0

Can anyone guide me in this regard.
Does anyone have the input file for dos computation of intrinsic graphene

I will try to help on the nkpt issue....
kptopt = 1 auto generates k point grids based on the variables ngkpt, shiftk, and nshiftk in the irreduceable BZ.

To test what the value of nkpt is, just look at your log file or output file and look in the echo of input variables section
probably nkpt = 28 but I am not exactly sure.
To see what the value of the k points are, also look at kpt = (which would be a list)
With your values, kpt probably looks something like
kpt = 0.015 0.0 0.0
0.03 0.0 0.0
... ... ...

While using kptopt = 1, to adjust the number of k points, adjust the value of ngkpt
It looks like you have a one-dimensional wire in the x direction and adjusting the value of the 1st number will add or subtract more k points in the B1 reciprocal vector

To better understand kptopt, ngkpt, shiftk, and nshiftk, I would look at the meanings in the abinit variable documentation and adjust values to see their effects

Seems like something has changed in this question and just to clear up my response from before.
ngkpt = 45 45 1 # a 2 dimensional structure
nkpt = 484 # 22X22X1
shiftk = 0 0 0 # makes sure that the kpt grid starts on the gamma point

I thought that you are supposed to obtain the correct k point grid through a kpt convergence in the self consistent case and then use that value the rest of the way?