Generate a convolution function to remove the 2-position chop function
from raster scans SCULIB_2POS_CONFN
For the case where UNBAL = 1, the two beams are of equal strength:-
The ideal convolution function would be the one whose FT was the inverse of the FT of the chop function. Unfortunately, the chop function FT has zeroes in it, at which the inverse FT will tend to infinity. The ideal convolution function, therefore, does not exist. This problem is avoided by generating a function whose FT is the same as the ideal function except at the problem points, where it is set to zero. This function consists of a series of delta functions separated by the chop spacing, the central 2 points being half the chop spacing on either side of the centre of the function.
The function is normalised such that convolving this function with an original chop function of 2 delta functions of unit height will give a delta function of height 1.
Don’t understand what happens when UNBAL 1, but the code has left doing the same as NOD2.
The convolution must cover the map even when the centre of the function is at the left or right hand extremity of the map. Hence the convolution function must be twice the length of the raw map.
Since the raw data is not sampled such that the chop spacing is an integer number of samples, the actual convolution function must be rebinned onto the sample mesh by sinc interpolation.
This routine is essentially a rewritten version of CONF22 from the RESTOR program in the NOD2 package by Haslam (1974) Astron. Astrophys. Suppl. 15 p333