RANDOM_MAP

        [GREG2\]RANDOM_MAP  [Arg1  [Arg2]]  [/NEIGHBOURS  Nb]   [/TRIANGLES]
    [/EXTRAPOLATE] [/BLANKING Bval]

    This command is a general purpose interpolation  task  which  uses  data
    from  the X, Y and Z arrays to fill in a Regular Grid array representing
    the same surface Z=f(X,Y) by an interpolation process. X, Y  and  Z  are
    arrays defined by the COLUMN command (see HELP COLUMN).  It triangulates
    the data for interpolation. If a triangulation already exists, and  only
    the   Z  array  has  been  changed,  RANDOM_MAP  does  not  make  a  new
    triangulation.

    Usually, the user-provided Z=f(X,Y) surface is REGULARLY  SAMPLED,  even
    if  NOT  COMPLETELY  SAMPLED.  In this case, one SHOULD USE directly the
    RGDATA command to create a Regular-grid map, since RANDOM  is  known  to
    perform awkwardly with regularly sampled data.

    The interpolated grid can be  defined  by  the  arguments.  If  Arg1  is
    "RGMAP",  the  interpolated  grid  will  be  exactly coincident with the
    current Regular Grid array, else Arg1 and Arg2 are used to indicate  the
    number of pixels in X and Y directions and Arg2 = Arg1 if not specified.

    Note that several hypothesis are made about the geometry of triangles in
    the  plot  page space (hence the geometry may be changed by changing the
    LIMIT or the BOX_LOCATION). You should thus use  limits  which  gives  a
    reasonable  representation  of  what  are the neighbours of a given data
    point. Moreover, this is an interpolation process. Hence  if  the  basic
    sampling theory is not respected by your choice of input data points and
    output grid, spurious oscillations may occur in the resulting grid.