On Tue, May 30, 2017 at 3:29 AM, Joa Ljungvall <[email protected]> wrote:
> Dear all,
>
> I'm revisting an old problem involving solving laplace and poision for a
> moderatly complicated geometry. Here I would like to start with a 3D mesh
> with
> few nodes and refine based on a kerry estimator. I've tried this before
> (2009-2010) with libmesh and ran into the problem that I could not figure
> out a good way to snap new nodes to the boundary of my geometry. Last time
> around I gave up since I created meshes that had all types of problems.
>
> Now my question, If I make my own MeshRefinment class and rewrite the
>
> Node * libMesh::MeshRefinement::add_node ( Elem & parent,
> unsigned int child,
> unsigned int node,
> processor_id_type proc_id
> )
> {
>
> ....
> }
>
> method, more in particular the lines
>
> 135 // Otherwise we need to add a new node, with a default id and the
> 136 // requested processor_id. Figure out where to add the point:
> 137
> 138 Point p; // defaults to 0,0,0
> 139
> 140 for (unsigned int n=0; n != parent.n_nodes(); ++n)
> 141 {
> 142 // The value from the embedding matrix
> 143 const float em_val = parent.embedding_matrix(child,node,n);
> 144
> 145 if (em_val != 0.)
> 146 {
> 147 p.add_scaled (parent.point(n), em_val);
> 148
> 149 // If we'd already found the node we shouldn't be here
> 150 libmesh_assert_not_equal_to (em_val, 1);
> 151 }
> 152 }
>
> so that if the created node is on a surface I snap it to the boundary
> by moving it along the normal of the surface of the parent element. Could
> that
> work without any strange side effects? This is the old thread of
> discussion we
> had at the time
>
> https://sourceforge.net/p/libmesh/mailman/libmesh-users/
> thread/20091014130140.GM20082%40joa.me.uk/#msg23751829
I don't think deriving your own MeshRefinement object is the right
approach... not saying that it couldn't be made to work, just that the
class was not really written with that use case in mind, so you would
probably run into a lot of gotchas.
If it's possible for you to use quadrilateral/hexahedral elements, I think
you might have better luck doing what you are trying to do using deal.II.
Since the 2009 email thread, they have added support for linking deal.II
with OpenCascade [0], including IIRC the ability to snap adaptively refined
elements' nodes to the underlying "true" geometry.
[0]: https://www.dealii.org/8.5.0/external-libs/opencascade.html
--
John
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
