Reducing systematic errors in the selection of signals from two-photon mass spectra

When selecting signals from two-photon invariant mass spectra the signal significance can be optimised by selecting signal candidates from a narrow mass window around the signal peak. However, such tight cuts can lead to large systematic errors. In this study systematic errors due to inaccuracies in the Monte Carlo simulation of the peak width are studied for three selection methods. It is shown that in the presence of such inaccuracies a wider selection window is required to reduce systematic errors. The Ranking selection method is found to be much less sensitive to systematic errors of this type.