[Pw_forum] Erro in cdiagh or cdiaghg

[Paolo Giannozzi]

On Friday 08 April 2005 10:14, Sergey Lisenkov wrote:

> I got an error 'in cdiagh or cdiaghg'. Following the manual, I use 
> 'cg'  algoritm, but the error was the same. From 'diis' algoritm I got:
> [...]
>       from diis_step : error #         1 
>       negative norm in S

if all diagonalization algorithms have a problem, you either have
some weird input data, or bad pseudopotentials

Paolo