Use of existing multi-resolution infrastructure
Disclaimer
What follows is just "chicken scratchings" for the
LDRD Final Report.
Why Important
-
Multi-res. support "for free" ("support" has broad applicability)
-
This infrastructure used for computations and development
-
Little, if any, support for multi-res. in existing commercial or common
vis. packages
-
Use existing data files - no copying or translating
-
Multi-res. vis is "hard" partly because data structures for multi-res.
are "hard/complex" - don't reinvent the wheel
Problems
-
Storage format/layer not "solidified/stable" - will it every be?
-
Compile issues
-
2 groups with different goals and difference research are extending the
tools/libraries
-
Cross language issues
Why these libraries aren't enough for vis.
-
Data structures needed to be expanded for vis. (e.g. parenting, adjacency)
-
"Primitive" functions lacking for vis. (e.g. slices, interpolation)
-
Documentation and organization chaotic - release engineering
-
Compile time flags are a problem (e.g. 0D vs. 1D vs. 2D vs. 3D, debugging
on or off) causing barriers (e.g. Rn region extraction)
-
Need a wholestic view of data (not just collections of grids at various
levels of refinement)