[petsc-dev] controlling vector values while doing matrix free operations

[Dmitry Karpeev]

On Wed, Dec 14, 2011 at 9:56 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

> On Wed, Dec 14, 2011 at 19:52, Dmitry Karpeev <karpeev at mcs.anl.gov> wrote:
>
>> I guess I'm missing something: why would you add pressure to the set of
>> conservative variables?  It's redundant, as it can be obtained at any time
>> via the equation of state. The only reason to add it is to impose a bound
>> constraint on it and use it to monitor for departures from the feasible set.
>
>
> This is exactly the reason. I'm saying that since it is defined by the
> equation of state, we might be able to avoid adding it explicitly.
>
> Note that in DG, it will be evaluated at quadrature points, and would thus
> break the block structure satisfied by the conservative variables. This
> likely also makes it more difficult to eliminate (while using to impose
> constraints).
>
If you don't care about the extra storage, you could put the conserved
variables and the  "constraint variables" in separate splits, make the
Jacobian a MatNest and solve with the consevative subblock most of the time
(away from the feasible set boundary).  That submatrix would retain the
nice structure and all.  I'm thinking about adding "bounded variables" to
libMesh in that way.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111214/98bbb829/attachment.html>