[petsc-dev] Runtime selection of MatType for DMGetMatrix

[Jed Brown]

On Thu, Nov 25, 2010 at 21:07, Jed Brown <jed at 59a2.org> wrote:

> On Thu, Nov 25, 2010 at 21:04, Barry Smith <bsmith at mcs.anl.gov> wrote:
>
>> Sound like the right approach?
>
>
> Yes, this sounds good.
>

I just pushed some patches that broke up DMGetMatrix_Composite, be sure to
pull those if you are going to work on this.  I'm headed home and plan to
get a simple 2-physics demo working with FieldSplit and use of MatNest or
AIJ determined by -dm_mat_type.  I can probably move the options around
tomorrow if it's not already done by then.

Jed
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20101125/fc6187f1/attachment.html>