Here we list the structure of a general N-dimensional binned dataset, and then consider the various substructures in turn.
DATASET <#D> See comment below on
TYPE
TITLE <_CHAR*80>
DATA_ARRAY(m,n,...) <_REAL> Could be any numeric
type
LABEL <_CHAR*80>
UNITS <_CHAR*80>
VARIANCE(m,n,...) <_REAL>
QUALITY <QUALITY>
QUALITY(m,n,...) <_UBYTE>
BADBITS <_UBYTE>
AXIS(N)
DATA_ARRAY <p_array>
LABEL <_CHAR*80>
UNITS <_CHAR*80>
NORMALISED <_LOGICAL>
WIDTH <s_array>
MORE <EXTENSION>
ASTERIX <EXTENSION>
HEADER <EXTENSION>
INSTRUMENT <EXTENSION>
SORT <EXTENSION>
PROCESSING <EXTENSION>
BGND_SUBTRACTED <_LOGICAL>
CORRECTED <EXTENSION>
DEAD_TIME <_LOGICAL>
VIGNETTING <_LOGICAL>
BARYCENTRIC <_LOGICAL>
LIVE_TIME <EXTENSION> Only for data binned
up from event data
PSF <EXTENSION> Only for image data
ENERGY_RESP <EXTENSION> Only for spectral data
HISTORY <HISTORY>
Note that the only component which is mandatory for a binned dataset is the DATA_ARRAY, though some particular applications will be unable to function sensibly without further information. In general software should default as many of the other components as possible, where they are not present.
A package of subroutines is available to allow easy access to components in standard binned datasets. See ASTERIX document PROG_003 for details.