tolerance for diagonal entries + Slepc

[Umut Tabak]

Umut Tabak wrote:
> Dear all,
>
> I am reading some matrices into petsc, but some of the diagonal 
> entries are really small, on the order of e-8. Petsc complains about 
> missing diagonal entries, I guess it compares them with some 
> predefined tolerance value, is there a way to change/set these 
> tolerances?
>
> Error is "Matrix is missing diagonal entry 1847!" after read from a 
> matrix market formatted text file.
>
> Best regards,
> Umut
>
>
sorry that I forgot to explain my problem completely, I get this error 
when I try to use Slepc's EPSSolve function, so that the question is 
also related to Slepc I guess. I mean the tolerancing issues for the 
matrices.