[Pw_forum] Warning in the 'nscf' calculation

[Paolo Giannozzi]

On Nov 30, 2007, at 24:20 , alan chen wrote:

>      WARNING:     3 eigenvalues not converged [...]
> The band structure is finally calculated at the end of output file
> but shall I believe in these E_{nk} values?

the code counts unconverged eigenvalues and makes
another tour of iterative diagonalization, until the desired
number of eigenvalues (all of them for non-scf calculations)
have converged: see function test_exit_cond. So yes, you
should believe these values, or more exactly, you shouldn't
be any more suspicious that for all other numbers.

> How can I get rid of these warning messages?

try the "see no evil" approach: comment the line yielding the
warning message. The alternative is the "computer geek"
approach: find a better algorithm.

Paolo
---
Paolo Giannozzi, Dept of Physics, University of Udine
via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222