tet(tri) meshing from balanced oc(quad)trees

[mrbuche]

Seems pretty easy (at least a lot easier than dualization so far...)

https://www.cs.umd.edu/users/hjs//pubs/reddyasae01.pdf

I guess you can do it simply based on Delaunay triangulation, and you don't seem to run into the same issues about order-of-doing-things that we run into with dualization.

We could build it and then do a smoothing step (internal nodes) to get better quality and hopefully something more like the unstructured tet meshes we are used to seeing.

If the input is a segmentation and we can get nice interfaces somehow, this could be a useful tool.

And maybe there are a limited number of "templates" for adding tris (tets) along the stair-stepped boundaries of the segmentation in order to smooth it out.

[mrbuche]

have to add some padding since nels have to be powers of 2 to fit the octree scheme. assign Some/None materials to cells. make the padding None, and any other materials CLI asks to be removed from segmentations. remove those cells before meshing.

[mrbuche]

https://doi.org/10.1007/s00366-013-0328-8

[mrbuche]

https://doi.org/10.1002/nme.1620290105

[mrbuche]

seems like you can use more tets in the templates in order to get better aspect ratios. but good smoothing might take care of that afterward anyway? what "templates" does Cubit seems to "use" in practice?

[mrbuche]

https://doi.org/10.1016/j.cad.2012.01.002

[mrbuche]

probably do the simplest (least amount of tets) templates first, then think about these other papers that try to control quality afterward, if we even need to (smoothing might be enough?)

[mrbuche]

https://doi.org/10.1016/0956-0521(94)90014-0

https://doi.org/10.1002/nme.1620201103

https://doi.org/10.1002/fld.3918

[mrbuche]

Strong/weak balancing, as well as pairing/not, should remain options. Since if we ever do tetmeshing from octrees, we probably can use weak balancing and skip pairing.

[hovey]

That sounds good. 👍 On Dec 11, 2024, at 16:02, Michael R. Buche ***@***.***> wrote: Strong/weak balancing, as well as pairing/not, should remain options. Since if we ever do tetmeshing from octrees, we probably can use weak balancing and skip pairing. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were assigned.Message ID: ***@***.***>