Howdy,
Cataldo Manigrasso and Luca Heltai have been implemented the possibility to
discretize manifolds for some time on a branch of deal.II. This would
allow, for example, to solve PDEs or boundary integral equations on the
surface of a sphere, or other manifolds embedded in higher dimensional
spaces.
This is implemented by adding a second template argument to many classes:
the first one ("dim") denotes as before the dimensionality of the
triangulation. The second ("spacedim") the dimensionality of the space we
are in. This new, second, template argument has as its default value
spacedim=dim, making all existing cases backward compatible to the
functionality we had before. However, if you want you just explicitly
specify this second template argument (e.g. Triangulation<2,3>) to get
manifolds in higher dimensional spaces.
I think this is cool functionality, and I would like to thank Cataldo and
Luca for their work!
Before merging this branch, I've taken the opportunity to revamp a good
part of the iterator/accessor subsystem. The drawback of our previous
implementation where we had explicit specializations of classes like
TriaObjectAccessor for 1d, 2d, and 3d objects was that it was hard to find
the documentation of a function because it could be in a number of places.
These classes have now all been merged into single templates, i.e. there
are now only classes TriaAccessor, DoFAccessor, and MGDoFAccessor, along
with CellAccessor, DoFCellAccessor, and MGDoFCellAccessor. Documentation
should as a consequence be a bit simpler to find now.
Best
Wolfgang
-------------------------------------------------------------------------
Wolfgang Bangerth email: [EMAIL PROTECTED]
www: http://www.math.tamu.edu/~bangerth/
_______________________________________________
