[Jonas Holdeman]

I am looking for references on data structures and algorithms for adding, deleting, & refining triangular meshes. I am interested in local refinement and moving boundaries. Code examples would be helpful. I want to implement this in Matlab. I find lots of papers on applications but little on how this is achieved.

My specific interest is this: I use vector finite elements, which are the curl of a modified Hermite stream function element, for incompressible flow, so these elements are divergence-free. Many of them have a potential tangential discontinuity at inter-element boundaries (which does not affect the divergence). If an element is complete cubic, say, then interpolation of locally cubic fields is exact and exposes no discontinuity, but not so for higher order variation. If one evaluates the potential discontinuity on triangle edges one could refine a mesh until the flow field is locally cubic so that the field is faithfully interpolated by the elements. But this is just an aside for now.

I might also remind users of this forum that the topic of mesh generation on the CFD Wiki is sorely in need of development.

[FMDenaro]

Originally Posted by Jonas Holdeman

I am looking for references on data structures and algorithms for adding, deleting, & refining triangular meshes. I am interested in local refinement and moving boundaries. Code examples would be helpful. I want to implement this in Matlab. I find lots of papers on applications but little on how this is achieved.

My specific interest is this: I use vector finite elements, which are the curl of a modified Hermite stream function element, for incompressible flow, so these elements are divergence-free. Many of them have a potential tangential discontinuity at inter-element boundaries (which does not affect the divergence). If an element is complete cubic, say, then interpolation of locally cubic fields is exact and exposes no discontinuity, but not so for higher order variation. If one evaluates the potential discontinuity on triangle edges one could refine a mesh until the flow field is locally cubic so that the field is faithfully interpolated by the elements. But this is just an aside for now.

I might also remind users of this forum that the topic of mesh generation on the CFD Wiki is sorely in need of development.

have you searched for papers of Barth and Deconinck? I remember also some issue reported in https://books.google.it/books?id=hGT...page&q&f=false

further references and pseudo-code are here http://www.cs.berkeley.edu/~jrs/papers/2dj.pdf