ABINIT Discussion Forums

Dear all users
As every one knows the band structure of graphene has a touching in the cone point (Dirac point) that this touching should be preserved by GW correction.
I applied the GW approximation on the graphene sheet. when i apply the GW on the normal hexagonal cell of graphene with two atom in the unit cell, i received a converge behaviour for GW gap versus k-mesh at the Dirac point. It should be mentioned that in this case i have a 24 symmetry operators for the system. But when i construct the supercell of graphene (1*2 cell) with four atoms in the cell the symmetry operators reduced to 4. In this case the primitive unit cell is a rectangular not hexagonal and i received strange manner for GW band gap at the Dirac point and i never received touching (or near touching) at the Dirac point by GW approximation. In each case LDA yields a good touching at the Dirac point.

"I think that this problem is only due to the symmetry and i think that ABINIT code is not symmetrized well and there is a bug in the code."

I attached to this e-mail the plots of band gap variation versus k-mesh at the Dirac-point for the normal cell and super-cell of graphene-sheet.
and i also attach one of my input file for GW correction on the super-cell of graphene sheet.
Any suggestions will be appreciated.
Peiman

It seems to be related to the convergence vs k-points.
The converge of the exact exchange vs k-points is noticeably slower than the convergence of LDA.
A way to improve the convergence is to use a truncated Coulomb interaction as proposed by Spencer and Alavi in their PRB.
Could you make a try using

?