Steps in case of ICHAR=11.

[bikash.patra]

Dear experts,

I am confused regarding the steps when we use ICHARG=11.
This is used to plot the band structure.
This means we are starting the calculation from a pre-converged
CHGCAR file and keeping the density fixed throughout the calculation.
If we know the density, we can calculate the potential (at least upto
GGA level), and after solving the KS equation in one step, we can get
the eigenvalue at any k-point we want. But, when running with ICHARG=11,
there are some scf like steps in the output. What are these steps?

With regards,
Bikash Patra
TIFR, Mumbai

[martin.schlipf]

This is a common misunderstanding about the way that VASP works. If you would use an exact diagonalization (ALGO = Exact), you could indeed stop after a single diagonalization and obtain converged eigenvalues. However, VASP obtains the orbitals with an iterative diagonalizer, i.e., there are two self-consistencies, one with respect to the orbitals and one with respect to the density. You need multiple steps to get good enough orbitals so that your subspace diagonalization yields the correct eigenvalues. By the way, this is also the reason that for delaying the density mixing (NELMDL) in the beginning of every calculation.