Temporal analysis

Exploratory and interactive examination of variability of sources or background is easily done within IDL's Event Browser: Event Browser Manager: Click Analysis to define and fit a model. To establish the probabillity that any significant variations are present, we recommend displaying the Distribution Function (do not smooth first), adopting a linear model (default), and using the probability reported from the Kolmogorov-Smirnov one-sample test. Unless gaps are present in the dataset, this should be a reliable test for variability.

WARNING! The "light curve" produced by EB is a simple histogram of the event timestamps. GTI tables are NOT considered; thus dropped CCD frames (from telemetry saturation) will corrupt the plot.
WARNING! EB does NOT know anything about the intrinsic frame time of the data. If you choose (or let EB choose) a histogram bin size that is not >> the CCD frame time you will get serious aliasing effects.

--- ooo ---

Programs from Arnold Rots at the CXC for barycentric correction are available from D. Sanwal at ~divas/Barycor. The procedures are complex -- read the README file in advance. The ephemeris used by CXC is DE405 with the ICRS coordinate system; this should be satisfactory unless microsecond accuracy is needed.

CIAO has a tool lightcurve, which can also be run from the CIAO tool FirstLook. CXC has a number of caveats about ACIS timing data which should be examined.