Flattened version of AMR blueprint?

[nselliott]

I think it may be useful to create a second way to describe AMR meshes with the blueprint. This would not replace what has already been done but would be another option. The existing format contains all of the domains on all of the levels regardless of whether they are covered by finer domains. What I envision as the flattened version would only describe the finest or "visible" portions of the mesh at any particular location in space. It would still use structured topologies.

It would be straightforward to add code to SAMRAI to describe one of these hierarchies with this version of the blueprint. Once we have that, this would be easier than the current AMR blueprint to transform into a polyhedral blueprint.

I think this kind of blueprint would also be easily visualizable, though you wouldn't be able to hide fine domains and see the underlying coarse domains the way you can with full structured AMR viz formats.

[cyrush]

@nselliott -- is this already a subset of what is admissible? If the domains are fully covered at finer levels they can be omitted and the nest sets can provide less entries?

As long as the presented domains and nest sets are consistent seems like it might already fit the current blueprint?

[nselliott]

@cyrush You might be right--there would be no nestsets, and it should work as long as all of the domains of varying resolutions are constructed to fit together and cover the problem space.