sieve-dev [petsc-maint #54303] Support for unstructured simulations

[Matthew Knepley]

On Sat, Oct 9, 2010 at 11:32 AM, Michel Laurent <laurent.michel at epfl.ch>wrote:

> Hi,
>
> I have heard from some friend who participated to one of your workshops
> that you have now support for unstructured meshes and simulations in your
> library (I have even found a pdf presentation on Google about that). It
> seems like one can use the fieldsplit preconditioner now with unstructured
> problems (which wasn't the case a few months ago when I asked).
>
> My question is: where can I find documentation about that? I mean, about
> fieldsplit applied to unstructured problems, AND the way petsc users are
> supposed to deal with unstructured meshes with your library.
>

I will answer in two parts:

  1) FieldSplit

This is somewhat well-documented and stable. It is very easy to do. Instead
of saying which
dofs in which DA belong to a "field", you gives a set of IS objects, listing
the dofs in each "field".
After that, everything is the same. Should be very easy. We use this
strategy in the PyLith <http://www.geodynamics.org/cig/software/pylith>
package to solve our systems.

  2) Unstructured meshes

This is not well-documented. There is a tutorial presentation and a
repository of code for it. A few
people have used this, but it is nowhere near the level of clarity and
robustness that the rest of
PETSc has. However, it can do that mesh handling and if you are willing to
be an alpha user it
might be right for you. I suggest:

  a) Looking at the tutorial and our
paper<http://portal.acm.org/citation.cfm?id=1611487>

  b) Look at some sample code. I attach a really simple example here in the
C interface. There is
      a more expressive C++ interface (that PyLith uses) which has no
documentation at all.

  c) Do requirements gathering for your code. If you know what you need, I
can easily tell you
      whether the library supports it. There is no consensus for what should
be supported by
      unstructured mesh packages, unlike say linear algebra.

  Thanks,

     Matt

P.S. The sample code reads an ExodusII mesh which i also attached.

Thanks guys for your work, it's amazing,
>
> L.
>
> --
> Laurent Michel
> PhD Candidate
> EPFL MATHICSE
>
>


-- 
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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/sieve-dev/attachments/20101009/33f36f5f/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TestMesh.c
Type: application/octet-stream
Size: 3175 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/sieve-dev/attachments/20101009/33f36f5f/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Square.gen
Type: application/octet-stream
Size: 2852 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/sieve-dev/attachments/20101009/33f36f5f/attachment-0001.obj>