ABINIT Discussion Forums

Dear Abinit users

I have a doubt regarding the value of the converged ngkpt from the Abinit simulation of ngkpt convergence for a 4 by4 graphene supercell.
I plotted the Etot versus ngkpt values in Excel, but I am not sure of which value I need to select as the converged ngkpt.
The Etot versus ngkpt values are as follows:

ngkpt Etot in Hartree
3 3 1 -182.55416858
4 4 1 -182.55914084
5 5 1 -182.55868711
6 6 1 -182.55800865
7 7 1 -182.55866877
8 8 1 -182.5586083
9 9 1 -182.5584028
10 10 1 -182.55860418
11 11 1 -182.55858787
12 12 1 -182.55850248
13 13 1 -182.55858637
14 14 1 -182.55858051

I found oscilaltions in the total energy as the k-point mesh is increased. Sometimes the Etot go up or down as the mesh is changed. Is the converged value 5 5 1 or greater than that. How could I select the converged ngkpt value from these values.

Can anyone please guide me in this regard.

Any help on this is highly appreciated.

Thanks in advance
Seba

Hi!

All obtained values should be compared with the most converged one. In your case with the one for mesh 14x14x1. If you take the usual Etot tolerance of 5x10^(-4) Ha, then your converged value would be the one for 7x7x1.
The only question remains is whether 14x14x1 is a completely converged mesh.

Best regards,

Igor