We explored the statistical properties of short-term X-ray variability using long-exposure {it XMM-Newton} data during high X-ray variability phases of blazars S5 0716+714 and PKS 2155-304. In general, hardness ratio shows correlated variations with the source flux state (count rate), but in a few cases, mainly the bright phases, the trend is complex with correlation and anti-correlation both, indicating spectral evolution. Stationarity tests suggest the time series as non-stationarity or have trend stationarity. Except for one, none of the histograms fit resulted in a reduced-(chi^2 sim 1) for a normal and log-normal profile but a normal profile is favored in general. On the contrary, the Anderson-Darling test favors lognormal with a test-statistic value lower for log-normal over normal for all the observations, even if out of significance limits. None of the IDs show linear RMS-flux relation. The contrary inferences from widely used different statistical methods indicate that a careful analysis is needed while the complex behavior of count rate with hardness ratio suggests spectral evolution over a few 10s of kilo-seconds during bright phases of the sources. In these cases, the spectrum extracted from whole observation may not be meaningful for spectral studies and certainly not a true representation of the spectral state of the source.
تحميل البحث