Discovery , identification , and study of variability in gamma - ray point sources

We present preliminary results of a statistical analysis obtained with a sample of blazars observed at the Perugia Astronomical Observatory since 1992. We briefly show how these statistical results can be useful to discriminate faint variable sources against the background noise. This technique, together with the more traditional ones, may be used to discover and identify high-energy point sources. 1. Statistical analysis of variability Many of the expected high-energy sources are variable, like blazars which emit signals that often appear to vary chaotically with time. Does such signal result from the superposition of many random events (randomly generated shocks in outflowing gas; randomly appearing reconnections of magnetic field lines; etc.)? Or they are governed by some global physical mechanism described by non-linear differential equations, which shows deterministic chaos (coherent variability of clouds of electrons; instability of a shock at the base of the jet forming region; etc.)? Finally, is it possible to find same kind of periodicity, at least in particular intervals of time (hot spots in the rotating accretion disk; helicoidal movements of the jet; etc.)? For a reasonable answer to these questions, we can use-for example-one or more of the following statistical analysis: • correlation, autocorrelation, power spectrum estimation, with FFT: future high energy missions working in scanning mode (like GLAST) probably will provide data well sampled at evenly spaced intervals, that can be processed with the Fourier Transforms; • correlation, autocorrelation, power spectrum estimation, with the Lomb-Scargle periodogram, (Lomb 1976, Scargle 1982) for irregularly or incompletely sampled data; • correlation and autocorrelation with the Discrete Correlation Function (DCF) described by Edelson & Krolik (1988), useful for unevenly spaced data; • first order Structure Function (SF), because there is a simple correspondence between power laws in the frequency space Fourier analysis, and the SF analysis in the time domain (see, e.