Madison, Don H.
Data, in particular that generated through successive computer approximations, may not always follow the smooth curve it is suppose to. Approximations in input data, round off error, and estimations made in the mathematical formulas, are just some of the things that could cause this problem. In any respect, this data must somehow be smoothed or fitted for publication. One way of fitting such data sets is the relaxation method. This method does not rely on the data conforming to a specific mathematical shape, as opposed to other programs that require the data to be fitable by some mathematically generated curve. Instead, the relaxation method uses the values of the points surrounding the current one to determine the new value at that location. This provides a curve independent method for fitting the data.
Lehmann, Michael J., "Relaxation Technique for the Fitting of Numerical Data" (1994). Opportunities for Undergraduate Research Experience Program (OURE). 36.