Optimum filters from experimentally measured noise in high resolution nuclear spectroscopy

We introduce a general method for calculating the optimum filter for high-resolution nuclear spectroscopy starting from an experimentally measured noise spectrum. The method is based on fitting the experimental noise, recorded in a finite bandwidth, by a truncated Laurent series approximation. The procedure of evaluating the noise power spectral density and of synthesizing the corresponding optimum filter is fully developed and applied to a typical example.