OGC Hierarchical Data Format Version 5 (HDF5®) Core Standard

Value that identifies a location.

An instance of an array type where each element or appropriate group of elements may be referenced randomly and independently of the others [ISO2382]. A non-degenerate array has a rectilinear shape described by its rank (dimensionality) and extent. There are two degenerate cases: empty arrays (so-called null arrays), and arrays that consist of only a single element (scalars).

Composite type whose components are of the same data type [ISO2382].

A variable capable of assuming values of an array type.

Data type, each data object of which consists of a single non-decomposable data value. [ISO2382]

Either of the digits 0 or 1 when used in the binary system [ISO2382].

String that consists of a number of bits, treated as a unit, and usually representing a character or a part of a character [ISO2382]. The number of bits in a byte is fixed for a given data processing system. The number of bits in a byte is usually eight (8).

Data type that has a structure composed of the data structures of one or more data types and that has its own set of permissible operations [ISO2382].

Reinterpretable representation of information in a formalized manner suitable for communication, interpretation, or processing [ISO2382].

Unit of data that is considered in context to be indivisible [ISO2382].

Graphical and/or lexical representation of data, specifying their properties, structure and inter-relationships [ISO11179].

One or more computers, peripheral equipment, and software that perform data processing [ISO2382].

Defined set of data objects of a specified data structure and a set of permissible operations, such that these data objects act as operands in the execution of any one of these operations [ISO2382].

A graph where the edges are represented by ordered pairs of vertices and an implied edge direction (or orientation) from the first to the second vertex.

A finite sequence of directed graph edges (arcs, lines) where for all edges but the last the edge’s second vertex is identical to the next edge’s first vertex.

Any concrete or abstract thing that exists, did exist, or might exist, including associations among these things [ISO2382].

Conversion of data into a series of codes [ISO11179].

Named set of records stored or processed as a unit [ISO2382].

A non-empty, finite set of vertices (nodes) and a set of edges (arcs, lines) that connect pairs of vertices. An edge can also connect one vertex to itself.

A named array variable ascribed to an HDF5 object, its owner or assignee.

An immutable kind of HDF5 object that is used to store an HDF5 datatype definition, which can be referenced by multiple HDF5 datasets and HDF5 attributes.

A mapping of an HDF5 object graph into storage such as one or more files in a file system, a buffer in memory, or an object store.

An array variable linked to one or more HDF5 groups.

An HDF5 datatype whose values represent references to a subset of elements of an HDF5 dataset. Logically, such a reference can be thought of as a pair which consists of a reference to an HDF5 dataset and a boolean array variable of the same shape as the dataset where boolean true/false might indicate the inclusion/exclusion in/from the subset.

A description of the data type of a data element including its storage layout as a pattern of bits.

A member of the HDF5 data model.

An HDF5 object that is a collection of zero or more HDF5 links.

A named, unidirectional association between an HDF5 group and an HDF5 object.

An HDF5 group, dataset or committed datatype; an HDF5 entity that can be linked to zero or more HDF5 groups and assigned zero or more HDF5 attributes.

A rooted, directed multigraph whose edges consist of HDF5 links, and whose vertices consist of all HDF5 objects which can be reached by traversing the edges starting from the HDF5 root group.

An HDF5 datatype whose values represent references to HDF5 objects in an HDF5 container.

A finite-length sequence of HDF5 link names. HDF5 path names are used to locate and refer to HDF5 objects.

Concrete module created from a generic module by generic instantiation [ISO2382].

Data about data or data elements, possibly including their data descriptions, and data about data ownership, access paths, access rights and data volatility [ISO2382].

A graph with multiple edges connecting the same two vertices.

Predetermined course of events that occur during the execution of all or part of a program [ISO2382].

Syntactic unit that conforms to the rules of a particular programming language and that is composed of declarations and statements or instructions needed to solve a certain function, task, or problem [ISO2382].

Artificial language for expressing programs [ISO2382].

A positive integer denoting the number of dimensions (dimensionality) of an array variable.

Set of data elements treated as a unit [ISO2382].

A graph with a single distinguished vertex, referred to as the root, and where all its other vertices can be reached by at least one directed path from the root.

The normative provisions in this specification are denoted by the URI: http://www.opengis.net/spec/HDF5/data-model/1.0.

Any partial URI that may appear in this document’s requirements and conformance tests shall be considered relative to this base.

All the diagrams included in this document use the Unified Modeling Language (UML) static structure diagrams, as described in Subclause 5.2 of OGC Web Services Common [OGC 06-121r8].

The following abbreviations are being used throughout this document:

An HDF5 link connects an HDF5 group, its source, with another HDF5 object (group, dataset, or committed datatype) and is the means by which group membership is established. The object that a link connects to is referred to as the link’s target.

An HDF5 link has a name associated with it. The name is a Unicode string that identifies the link within its source HDF5 group. A link’s name is used to resolve the link, i.e., to locate the link’s target. Because of this link names within one HDF5 group must be unique although there is no limit on how many links can have the same source HDF5 group and/or the same target. It is possible for HDF5 links to "dangle", i.e., their targets can be undefined or unavailable.

HDF5 links are the mean of traversing an HDF5 object graph, as depicted in Figure 2. Each vertex of the graph represents either an HDF5 group, dataset, or committed datatype. The identifiers associated with each HDF5 object in the graph are only to aid explanation in this text and are not part of the data model. However, an implementation would need to adopt some form of identification in order to distinguish different HDF5 objects.

One of the HDF5 groups in the graph is distinguished from the other groups because all the other objects can be reached by following one or more HDF5 links only starting from it. This HDF5 group has a special name: root group, and the data model allows only one such group to exist.

The names of the sequence of HDF5 links which lead to a specific HDF5 object form that object’s HDF5 path name. One HDF5 object can have more than one path name. HDF5 path names starting from the root group are called absolute, while those starting from any other HDF5 group are called relative. For example, these are all HDF5 absolute path names to reach the HDF5 dataset id:951: (a, c, c), (a, c, d, p), and (a, f, p). Few examples of HDF5 relative path names to the same dataset are: (f, p) from the id:100 group, and (d, p) or (c) from the id:591 group.

Cyclical path names (the start and end group are the same) are possible. Two examples in Figure 2: the HDF5 absolute path (a, f, b) from the root group, and the relative path (loop) of the id:591 group.

The HDF5 Group is a container of zero or more HDF5 links and has a functionality akin to a file system directory. With the exception of the root group, any other group of an HDF5 object graph is the link target of one HDF5 group.

An HDF5 object linked into a group is said to be a member of that group. The same HDF5 object can be a member of one or more groups. An example of this in Figure 2 is the id:951 dataset which is a member of both id:591 and id:684 groups.

An HDF5 datatype describes an individual data element of an HDF5 dataset or attribute. It specifies the set of possible values a data element can have and the set of operations that can be performed on it. The data model recognizes 10 different classes of HDF5 datatypes that fall into two major categories: atomic and composite, as depicted in Figure 3. Expected set of data values for each datatype class is listed in Table 3.

An HDF5 dataset or attribute has an HDF5 datatype associated with it. The datatype may either be private to that dataset or attribute, or be linked into an HDF5 object graph and be available for reuse by other HDF5 datasets and attributes. A linked HDF5 datatype is referred to as committed datatype. An example of committed datatype in Figure 2 is the id:802 object with the HDF5 absolute path (a, c, my uint).

An HDF5 dataspace describes the dimensionality (rank) and extent (the size of each dimension) of the data array associated with an HDF5 dataset or attribute object. The data model recognizes three different types of HDF5 dataspaces which properties are listed in Table 2.

The HDF5 Dataset is the main entity of the data model for storing data. An HDF5 dataset is comprised of a multidimensional array, an HDF5 datatype describing the array’s data elements, and an HDF5 dataspace specifying the array’s rank and extent.

HDF5 groups, datasets, and committed datatypes may be assigned zero or more HDF5 attributes. Each HDF5 attribute is private to the assignee HDF5 object.

An HDF5 attribute is a multidimensional array with a name. The array requires an HDF5 datatype and an HDF5 dataspace, the same as an HDF5 dataset.

An HDF5 attribute is functionally similar to an HDF5 dataset with the following differences:

HDF5 attributes support the self-descriptiveness of the data encoded with the HDF5 data model by holding user-created metadata.

The core requirements every implementation of the HDF5 data model must fulfill are listed in Table 1. Their normative definitions are given in the subsequent subclauses.