Fiber Bundle HDF5 - Fields

A single dataset may be specified in various ways to easy different memory layouts and implicitly given information.

Linear maps

• A field may be defined by explicit specification of each value in a dataset.
• If the values of a field are known implicitely via a linear map from the element index (eg. the coordinates of the points of a uniform grid), a field is a subgroups with the two attributes
  • "origin"
  • "delta"
  The two attributes need to be of the appropriate field type. A data value is computed from the point index via the map

            index |-> origin + index * delta
       

  For multidimensional indices, eg. the indices of a regular grid, this expression is valid in each dimension.

Compound data fields

A data set of elements which are not atomic data types, eg. vectors, may be specified as

• A single dataset made by the appropriate compound type (preferred)
• A subgroup containing datasets, with the group name defining the field name, and the name of the datasets in the subgroup corresponding to the entries in the appropriate compound type. This subgroup requires a reference to the named type. This storage method has the advantage that a compound field can be defined from partially existing components only (ie. others set to zero).

Extensions and Restrictions