[petsc-dev] Lagging the numerical factorization for one of the fields of fieldsplit preconditioner

[Jed Brown]

On Tue, Mar 15, 2011 at 23:43, Barry Smith <bsmith at mcs.anl.gov> wrote:

> I'm just asking that all matrix classes be able to propagate this
> information NOT just MatNest().


Okay, great. My point was just that MatNest pretty much already does this
[1], with MatAIJ, you need some new API to declare part of the matrix as
unchanged (otherwise I don't know how to efficiently determine it).

[1] MatAssemblyEnd_Nest() still calls MatAssemblyEnd() on submatrices. It
could keep track of whether submatrices have been "gotten" (if not, they
can't have been modified) or (better) submatrices could increment state in
MatAssemblyEnd() only when they have been modified on some process.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110315/45e38aa7/attachment.html>