[Pw_forum] Selection of K point for band structure calculation

[Lorenzo Paulatto]

On 23/10/17 18:55, Madhurya Chandel wrote:
> Thank you for your kind reply.
>
> The number of points is free, but you have to take it large enough 
> that the bands look smooth and not jagged.
>
> How to decide that? For inverse spinel (CuFe2O4) with FCC structure 
> (56 atoms), what is the minimium number of K point (approx) required?

It is just a tradeoff between when the picture looks good enough, and 
when you have too many point to compute... I usually take between 20 and 
100 for each line segment
>
> Also, the continuity analysis done by bands.x
> what is the continuity analysis?

when two bands cross, the code has to decide if it is an actual 
crossing, or if the two bands just narrowly miss each other and then 
diverge, this can be done by symmetry or by overlap of the periodic part 
of the wavefunction u_k(r) at consecutive points.

-- 
Lorenzo Paulatto - Paris