File format#
The LUE physical data model is a file format that is implemented on top of HDF5. HDF5 is a data model itself, as well as a library of functions with an associated API. More information about HDF5 can be found on the HDF5 support portal.
The LUE data model is a convention for organizing information about large collections of objects in an HDF5 file. In HDF5 information is stored using groups, datasets, and attributes. Groups can contain groups and datasets. Datasets contain multidimensional arrays. Groups and dataset can contain attributes.
In the next figure, the layout of the LUE physical data model is shown.
![digraph physical_data_model {
graph [
fontname="Arial"
fontsize=14
margin=0.1
nodesep=0.1
ranksep=0.3
bgcolor="transparent"
]
node [
fontname="Arial"
fontsize=14
fontcolor="white"
shape="oval"
style="filled"
]
edge [
fontname="Arial"
fontsize=14
]
dataset [
label="dataset"
fillcolor="#88C0D0"
]
universes [
label="universes"
fillcolor="#d08770"
]
universe [
label="universe"
fillcolor="#d08770"
]
phenomena [
label="phenomena"
fillcolor="#b48ead"
]
phenomenon [
label="phenomenon"
fillcolor="#b48ead"
]
object_id [
label="object-id"
fillcolor="#434c5e"
]
collection_property_sets [
label="collection property-sets"
fillcolor="#bf616a"
]
property_sets [
label="property-sets"
fillcolor="#bf616a"
]
property_set [
label="property-set"
fillcolor="#bf616a"
]
object_tracker [
label="object tracker"
fillcolor="#434c5e"
]
active_set_index [
label="active set index"
fillcolor="#434c5e"
]
active_object_index [
label="active object index"
fillcolor="#434c5e"
]
active_object_id [
label="active object id"
fillcolor="#434c5e"
]
properties [
label="properties"
fillcolor="#a3be8c"
]
property [
label="property"
fillcolor="#a3be8c"
]
time_domain [
label="time domain"
fillcolor="#5e81ac"
]
space_domain [
label="space domain"
fillcolor="#5e81ac"
]
value [
label="value"
fillcolor="#ebcb8b"
]
dataset -> universes [
label="1"
]
universes -> universe [
label="*"
]
universe -> phenomena [
label="1"
]
dataset -> phenomena [
label="1"
]
phenomena -> phenomenon [
label="*"
]
phenomenon -> collection_property_sets [
label="1"
]
collection_property_sets -> property_set [
label="*"
]
phenomenon -> property_sets [
label="1"
]
phenomenon -> object_id [
label="1"
]
property_sets -> property_set [
label="*"
]
property_set -> object_tracker [
label="1"
]
object_tracker -> active_set_index [
label="1"
]
object_tracker -> active_object_index [
label="1"
]
object_tracker -> active_object_id [
label="1"
]
property_set -> time_domain [
label="1"
]
property_set -> space_domain [
label="1"
]
property_set -> properties [
label="1"
]
properties -> property [
label="*"
]
property -> value [
label="1"
]
}](../../_images/graphviz-b25eda13acb71cb6a72de4dc112c66a22096056a.png)
The API for storing information in a LUE dataset reflects this layout. See for example the next example which
uses the LUE Python package and in which a new dataset is created. The object returned by
lue.data_model.create_dataset supports adding a universe to the universes collection and a phenomenon to the
phenomena collection. The object returned by lue.data_model.Dataset.add_phenomenon supports adding
collection property-sets and property-sets.
TODO