16 Summary

In DX the basic entity used to represent a data cube or a particle (or catalogue) dataset is called a field (this usage is similar to the use of the term ‘field’ in physics, for example, ‘velocity field’, but is different from the usual use in computing). Each field consists of a set of components; arrays holding the positions of the data points and a data array containing the dependent variable. The data array may contain either scalar or vector data. For particle (or catalogue) data, where the positions of the points are not related to each other, the field need only contain the positions and data components. However, for gridded data there is an additional connections array which defines the relation between neighbouring points in the grid and controls how interpolation is performed.

In DX gridded data may be either position dependent or connection dependent.

Position dependent
The data array contains instantaneous samples of the underlying field computed at the position of each grid point.
Connection dependent
The grid points delineate the corners of a cell. The data array contains the value inside this cell.

The practical difference between the two cases is that for position dependent data new values for positions intermediate between grid points are interpolated, whereas for connection dependent data they are not (the single data value applies to the entire cell). Position dependent data are probably more common in astronomy.