On Tue, Oct 17, 2017 at 5:01 PM, Jed Brown <[email protected]> wrote:

> Tobin Isaac <[email protected]> writes:
>
> > Which is something so rote that creating DTFD to convert the pointwise
> > residuals/jacobians into stencils would go along way towards unifying
> > DMDA and DMPlex in a single useful interface.
>
> On a related topic, I have reason to add Fornberg's algorithm for
> computing finite difference weights to DMDT.  (It's more stable than the
> usual Vandermonde approach.)  I'd imagine some users would appreciate
> that to take some of the mystery/effort out of the design of higher
> order FD methods.
>

That would be good.

Is there a reason not to do all DMDA computation with p4est? We could
prescribe
the refinement level and have the forest make the decomposition.

   Matt

-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>

Reply via email to