Limits to the applicability of commonly used pulse-shape figures of merit and possible alternative measures of performance

The ratio, Asub(n)/Asub(γ) of desired to undesired pulses measures the effectivness of a pulse shape discrimination system. Calculations show that this ratio is poorly predicted by commonly used pulse-shape figures of merit. The peak-to-valley ratio is shown to be a better predictor of Asub(n)/Asub(γ), but it also has limitations. A formula is given that provides better predictions and uses parameters easily obtained from the data. The variation of Asub(n)/Asub(γ) as a function of the fraction of the desired pulses accepted is also discussed. (Auth.)