Raw Data Smoothing

The basic goal when smoothing your diffraction data is to improve the signal/noise ratio, in order to be able to determine small peaks even in low-quality diffraction patterns that typically contain a large contribution of noise. Match! uses the well-known algorithm by A. Savitzky and M.J.E. Golay (described in Analyt. Chem. 36, 1627 (1964); corrected by Steinier, Termonia and Deltour in Analyt. Chem. 44, 1906 (1972)) to perform this task.

In order to smooth the data, either select the "Smooth raw data" command from the "Pattern" menu, press the corresponding button in the main toolbar at the top, or press <Ctrl+F4> on your keyboard (<Cmd+F4> on the Mac).

You can run the command several times until the desired degree of smoothing has been reached. However, always keep in mind that every smoothing operation may result in a loss of information (peaks)!

The algorithm can be configured on the "Raw data tab" of the "Options" dialog. Here, to the right of the check box "Smooth raw data", you can adjust the width of the smoothing window (which is by default determined automatically) as well as the order of the corresponding polynom (which is 2 by default).

By default, the width of the smoothing window is determined automatically from the step size and the width of the peaks. However, the best width of the smoothing window may vary, depending both on the step size as well as the peak width. If the width of the smoothing window is not suitable there may be artifacts (typically to the right of strong peaks). In addition, small peaks may "vanish".
As a rule of thumb, the width of the smoothing window should be equal to the number of data points in the full width at half maximum (FWHM) of the peaks (or up to 20% larger).
If the width of the window is too small (i.e. if too few neighbouring data points are used in the calculation), too much noise will remain. On the other hand, if the width of the window is too large, the shape of the peaks will be modified, resulting both in the vanishing of small peaks as well as in a loss of resolution.