[Pw_forum] ph.x: diagonalization (ZHEGV*) failed at some q-points

[Paolo Giannozzi]

On Dec 3, 2012, at 19:34 , Thomas Gruber wrote:

>       from cdiaghg : error #       868
>       diagonalization (ZHEGV*) failed
>
> Can someone tell me why

almost invariably, non-positive overlap matrix.

> and what I should do to get my representations?

It may turn out not to be an easy task. Try the following: edit PH/ 
set_defaults_pw.f90,
change the line
   isolve = 0
to
   isolve = 1
You may also need to add there a line
   max_cg_iter=20
and add variable max_cg_iter to line
   USE control_flags, ONLY: ...
Recompile. No warranty: if the Davidson algorithm says that the  
overlap matrix is
non-positive, it is usually true and it usually comes from less than  
optimal
pseudopotentials. Sometimes a different diagonalization algorithm can  
go through

  P.
---
Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222