[QE-users] About the k-points weights used for nscf calculation of pwscf.

[Hongyi Zhao]

On Tue, Jun 30, 2020 at 7:12 PM Nicola Marzari <nicola.marzari at epfl.ch> wrote:
>
> On 30/06/2020 13:07, Hongyi Zhao wrote:
> > As you can see, it uses 0.0156250 as the weight of all k-points. I
> > can't figure out how to obtain this value.
>
> How much did you try? You have 64 k-points, the weights are all equal,
> what could possibly be this value? Anyhow, here is the magic:
>
> 1/64=0.015625

OMG. I really did not try this. And if so, do you mean this is an
initial conjecture so that it can help wannier90 to do the job step by
step?


>
> > It said that for a non-scf calculation, weights do not affect the
> > results. If so, why they still set weights for the k-points in the
> > example I mentioned above?
>
> Because one could go into a mode, in the future, where the wannier code
> uses only the points in the irreducible wedge, or at the very least it
> uses time reversal.

Do you mean this is just a future possibly compatible usage but still
not implemented yet?

-- 
Hongyi Zhao <hongyi.zhao at gmail.com>