UV_INVERT

        Warnings: JP May 30th, 2004.
      - It is not clear that the image noise computed from the visibility
        weights is correct when tapering and/or weighting.
      - It is not clear what is the 4th element of UV_TAPER$.

    Optimized  version of UV_MAP using a .tuv table rather than a .uvt table
    (i.e. in time vs channels order, rather than in the natural channels  vs
    time order). This version is disk-based, and computes the optimal number
    of channels to be processed together based on the specified memory usage
    by  SPACEPPING. It is not memory limited, provided one channel fits into
    virtual memory. This means that this task will be able to  process  ALMA
    data...

    UV_INVERT  makes a map from UV data by gridding the UV data using a con-
    volving function, and then  Fast  Fourier  Transforming  the  individual
    channels.   UV_INVERT  can produce one beam per channel or a common beam
    for all channels. In the latter case, the gridding is  done  only  once,
    thus  neglecting  frequency  change  between channels; this is of course
    much faster.

    UV_INVERT can shift the map center and rotate the image, by shifting the
    phase  tracking  center  and rotating the UV coordinates of the input UV
    table.

    UV_INVERT produces an output LMV cube (X, Y, Velocity) and a BEAM  image
    or cube from  the input  UV table.