[QE-users] Fair distribution of k-mesh grid points in triclinic systems

[Arvin Kakekhani]

Dear QE community,

Usually for orthorhombic crystal structures in order to have a fair sampling we weight the k-mesh density in each a, b and c direction with the inverse of the length of the cell in that direction, a direction twice long generally needs half k-mesh grid points.
Now my question is what should one do for systems that are significantly triclinic. For example this:
CELL_PARAMETERS (angstrom)
16.651839677   2.491547706  -2.073450444
  2.443005780  16.525077990   1.297196607
 -0.788930778   0.422305589   4.624004634

How should one decide on what ratio of grid points one should have in k-space for such a system? Will the fact that the system is triclinic complicates the k-mesh convergence? Also does shifting the k-grid help convergence for such systems?

Thanks,
Arvin Kakekhani
University of Pennsylvania