first comments after reviewing and trying the new matrix factorization code

[Barry Smith]

On Aug 7, 2008, at 2:24 PM, Lisandro Dalcin wrote:

> I've started to wrap the functionality of the new code in petsc4py.
> However, I'm detecting some inconsistencies and possible
> improvements/simplifications and then I need some directions in order
> to push good fixes in petsc-dev.
>
> Then, here is my tow first questions.
>
> 1) Suppose I do
>
> MatGetFactor(A, solver_package, factor_type, &F)
>
> Then, if the call is successful, the following should happen
>
> a) F->factor should be equal to 'factor_type'

    Yes.
>
> b) F->preallocated sould be PETSC_TRUE

    I guess this is ok.
>
> c) F->assembled sould be PETSC_TRUE

    Hmm, not sure, should we wait until after the numerical  
factorization to mark this as true?
>
>
> Is this OK?
>
>
> 2) What should be the output of MatView(F, viewer) ?

   ASCII info should print matrix size, number nonzeros and solver type.

    Otherwise I am fine with not printing anything and generating an  
unsupported viewer type or format.


>
>
>
>
> -- 
> Lisandro Dalcín
> ---------------
> Centro Internacional de Métodos Computacionales en Ingeniería (CIMEC)
> Instituto de Desarrollo Tecnológico para la Industria Química (INTEC)
> Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET)
> PTLC - Güemes 3450, (3000) Santa Fe, Argentina
> Tel/Fax: +54-(0)342-451.1594
>