Question about band structure calculation

[chenweiguang]

In the VASP Manual(March 1, 2007 version, pp124), we know the band structure basically need following 2 steps:
1), generates a charge density through a static scf calculation
2), set parameters ICHARG=11, use the above charge density, perform a non-scf calculation for the supplied k-points

But i had a puzzle about above the second step. In my opinion, the supplied charge density is a find solution of K-S equation, and which keep constant in the non-scf calculation. However, I found the cycle calculation was still exist in the OUTCAR and OSZICAR, i don't know why.
In the other hand, if the supplied k-points is more sufficient than the k-points set in the first step( for some large system), whether it can be concluded that the latter result is better.

I hope it's clear for you
Sincerely
Thanks
<span class='smallblacktext'>[ Edited ]</span>

[admin]

1) please mind that VASP starts with a random wavefunction. Therefore, even if the charge (and potential) are kept fixed, a few electronic steps are necessary to converge the new wavefunctions to the self-consistent solution of that fixed Hamiltonian.
2) remind that band structures are calculated along special high-symmetry lines in the BZ, whereas the BZ has to be fully and evenly spread and to obtain a self-consisent solution for the structure. It is not a question of being better, you simlpy need a completely different k-mesh to plot a bandstructure.

[chenweiguang]

[quote="admin"]2) remind that band structures are calculated along special high-symmetry lines in the BZ, whereas the BZ has to be fully and evenly spread and to obtain a self-consisent solution for the structure. It is not a question of being better, you simlpy need a completely different k-mesh to plot a bandstructure.
[/quote]

For example, to the large system, usually a Gamma K-points is used, but the band calculation need more K-points.