Glossary

Terms used throughout this documentation.

Term	Meaning
Endpoint	The leaf object reached by walking mission, instrument, and quantity, such as spn.kaguya.esa1.energy_flux. Getting it reads no files; work starts with load() or plot()
SopranArray	The array returned by load(): an xarray DataArray with schema, time range, source files, and processing history attached
Schema	Definition of a variable's name, dimensions, units, data type, frame, and aliases (VariableSchema, InstrumentSchema). Written in code; documentation and validation are generated from it
Alias	A short name for the same variable, such as eflux for energy_flux or b for magnetic_field
Time range (TimeRange)	Half-open interval [start, stop); times without a time zone are UTC
View	Immutable analysis conditions bundling a time range, region, and processing settings
Project	Working directory per analysis topic that holds sopran.toml and outputs
Case	Analysis conditions saved by name in sopran.toml
Store	One root directory that keeps source files, intermediate data, and adopted results, separated into layers
Layer	A division of the Store: raw, normalized, features, models, databases, cache, registry
Dataset	One unit of data in the Store with a manifest, schema, catalog, and shards, named by an ID such as kaguya.esa1.counts
Shard	One Parquet file of a dataset; usually one per day
Sidecar	A *.sopran.json next to a source file or shard that records checksums and serves as the source of truth for catalogs
Manifest	A dataset's dataset.json, recording provenance, settings, and inputs
Catalog	catalog.parquet, an index of shard time ranges, row counts, checksums, and status
variant	A storage unit per set of conditions under one dataset ID; its ID is derived from settings and input content
managed	Marks datasets that SOPRAN creates and reuses by itself; the default target of synchronization
Cache	Keeping computed results in the Store for reuse, controlled by cache="use", "refresh", "never"
Download policy	How missing source files are retrieved: missing, never, always
Provenance	Record of the inputs, settings, and processing behind data or figures, kept in metadata, manifests, and figure JSON
Processing history (operations)	Record of operations such as rebin and transform and their arguments, in metadata["operations"]
Quicklook	A saved figure together with a JSON record of how it was made, created by quicklook()
PlotStack	A figure that stacks panels on a shared time axis; corresponds to SPEDAS tplot
Native bins	Time, energy, and viewing bins as recorded by the instrument, before calibration or rebinning
look	Index of a PACE detector viewing direction; not a direction itself, converted to one with the calibration tables
Pitch angle	Angle between particle velocity and the magnetic field (0–180°)
FrameContext	Bundle of SPICE kernels, time scale, and backend used for coordinate transforms
SPICE kernel	Data file for NASA NAIF SPICE with time, frame, ephemeris, or attitude information
Rust extension	The bundled native module sopran._native that handles heavy computation; users never call it directly
Experimental API	Research APIs in sopran.experimental, without compatibility guarantees
Pipeline	Mechanism that arranges retrieval, decoding, and saving stages to process long periods at once