High resolution cluster method for topological studies of the light curves of gamma-ray pulsars.

The authors describe a non-parametric, binning-free, adaptive method for the detection and the analysis of structures in the light curves of gamma-ray pulsars, based on a clustering technique. They use computer simulated data and a sample of COS-B data relative to the Vela pulsar to show its behavior and robustness. As a main result they find that the method is able to detect microstructures as narrow as the time resolution of the experiment down to very low signal-to-noise values and counting statistics. This important advantage with respect to the usual histogram technique, together with the conceptual simplicity, suggests its use in gamma-ray astronomy, both in the low energy (high counting statistics) and in the high energy range (low counting statistics).