ons as rfn
    >>> dt = np.dtype([('a', 'i4'), ('b', 'f4,u2'), ('c', 'f4', 2)])
    >>> a = np.arange(20).reshape((4,5))
    >>> a
    array([[ 0,  1,  2,  3,  4],
           [ 5,  6,  7,  8,  9],
           [10, 11, 12, 13, 14],
           [15, 16, 17, 18, 19]])
    >>> rfn.unstructured_to_structured(a, dt)
    array([( 0, ( 1.,  2), [ 3.,  4.]), ( 5, ( 6.,  7), [ 8.,  9.]),
           (10, (11., 12), [13., 14.]), (15, (16., 17), [18., 19.])],
          dtype=[('a', '<i4'), ('b', [('f0', '<f4'), ('f1', '<u2')]), ('c', '<f4', (2,))])

    r