[Pw_forum] How to calculate unoccupied states using cp.x

[Paolo Giannozzi]

On Sep 13, 2007, at 24:41 , Hongjun Xiang wrote:

> It seems that when nbnd is larger than the number of the occupied  
> states,
> CG gives even wrongs occupied eigenvalues.

before stating that eigenvalues are wrong, you should carefully verify
that they are well converged. Eigenvalues are more sensitive than
total energies to convergence, since the latter is a variational
quantity (i.e. has a minimum for the ground state charge density),
the former aren't.

There is an option (I don't remember which one) that allows calculating
empty Kohn-Sham eigenvalues in cp.x . It is described in the input
documentation, file Doc/INPUT_CP .

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy