12 * The simplest approximative Field. In future we want to replace it by pair: DofHandler + vector of dof values.
13 *
14 * In current setting:
15 * This should implement mapping: (ElementAccessor.INDEX) -> value
16 * Currently INDEX is global index of element (any dimension, both bulk and boundary part). In future INDEX should
17 * probably consist of MESH_LEVEL (unique mesh, dimension, bulk/boundary, level of refinement) and INDEX within this level.
18 * We want to make memory optimization since usually field lives either on boundary or on bulk part and some bulk fields live only on some dimension(s).
19 * This can be achieved by two level indirection table of mesh_levelscontaining tables for indexes. We should test if the performance penalty is not to big.
20 *
21 * Currently, we just use one vector for bulk and one for boundary elements.
22 *
23 * TODO:
24 * - move raw access resolution functions from FieldValues_ into FieldElementwise
25 * - allow elementwise int or FieldEnum data with optimal storage buffer, this needs
26 * templated GMSH reader
27 * - allow initialization of multiple fields by one reader
28 * - allow common storage for more elementwise fields to have values for one element on one place