The functions described in this chapter will read or write data to
a file in the HDF5 format. These contain
datasets together with a set of attributes for each dataset.
Datasets are arranged in a heirarchical name space similar to Unix
file system. Each namespace is called a group. Datasets are
stored in a group in an area of the file called a dataspace.
- Datasets consisting of a vector (
[]i64,[]f64, etc). - Datasets consisting of a 2-d array (
[][]u8,[][]i16, etc). - Datasets consisting of a 3-d array (
[][][]u32,[][][]f32, etc). - Any number of attributes for each dataset (
int,[]f64,string, etc). Attributes can be scalars or vectors. These are often metadata of the dataset describing how it was acquired or created.
- Writing to groups other than
/(which is the default) - Datasets of arrays of strings.
- Compound data structures (akin to
struct). - Images or tables.
- Compression.
- Distributed datasets (pointers to other HDF5 files).
- Parallel reading or writing.
See the HDF5 website for documentation and examples in C or Fortran.