Program Execution

• At the beginning you will have the option to run the code from already existing kburst result files. This bypasses the time consuming calculations and simply replots the graphs. Make sure all result files you give it have the same source direction and the same energy ranges. Kburst will not check this for you. By default, the result files are named 'resultxxxx', where xxxx is the four digit viewing period number. The user may rename them to a more meaning name for future use.

• If you do not use an previously existing result file, an input panel will appear. The user is ask to input the source direction, the minimum and maximum energies, and the earth zenith cutoff angle. If a minimum energy is input followed by a carriage return, a new earth zenith cutoff angle will be calculated automatically.

  An option is also available to run in automatic mode. Using this option, the program will ask you for the maximum angle between the source and the pointing direction that you wish to consider. It then searches through the EGRET timeline file to find all viewing periods where the source was within that angle of the pointing direction. The program then does the analysis once for each individual viewing period.

  If you choose the manual mode you must type in the viewing periods you wish to analyze. You may enter multiple viewing periods or a single viewing period. You may also enter sub-intervals of viewing periods by typing in the starting and ending times of the interval you wish to analyze although periods shorter than one orbit will be interpreted as an orbit in length. Kburst will prompt you for all of this information.

• Kburst now spawns the fortran program 'xpose.exe' to calculate the EGRET exposure for the given pointing direction. The fortran program 'exprd' is then spawned to convert the xpose.exe output from binary to an ASCII format. Kburst next runs the fortran code ' kb' which breaks the exposure output up into orbits and then checks the summary data base to see how many counts were recorded that orbit from the desired direction.

  One of the most difficult jobs for the kb code is to correctly break up the xpose output into orbits. There are no clear unambiguous signals in the xpose output file to indicate with 100% accuracy the beginning and ending of orbits. This problem is solved as follows:

  A file is maintained called 'vp_occul.dat'. As mentioned above, this file is used by the kb code to determine orbit boundaries. For each CGRO viewing period, there is an entry that contains the year, the day and fractional day of the exact center of an occultation somewhere in the viewing period and finally the orbit period in minutes.

  IMPORTANT: This part of the program needs to be maintained! A new entry must be added to the file for each new viewing period. Future updates to this file may be supplied by EGRET team member David Kendig djk@egret.gsfc.nasa.gov.

• Kburst will now allow the user to plot the results. The first graph to be plotted is a visual display of how the flux changes with time. The user may select to plot in units of Truncated Julian Days or in regular calendar days. The data may be binned by either time or by counts. Points are plotted at the end of the time interval where the counts were recorded. The user may then include the chi-squared test for time variability.

• The next graph shows the Kolmogorov-Smirnov test for time variability. This test works by comparing the cumulative number of counts expected from a steady source (equal to the average flux times the exposure for each orbit) to the cumulative number of counts actually observed. For a steady source these lines will overlap closely; while for a time variable source the line for the observed counts will show significant changes in slope, rising above or falling below the line for the expected counts. Kburst calculates a test statistic by taking the maximum difference in counts between these two lines and dividing by the square root of the total number of counts. This test statistic is then converted into a significance. The mathematics for this test is based on the original program by John Mattox. The Kolmogorov significance and the chi-squared significance are generally not equal. Some sources show up as being time variable to a relatively high significance in one test but not in the other. Which significance you should use in any given case should be decided by the particulars of that case.

• The last graph to be plotted is a histogram of the unlikely orbits. The probability for each orbit is calculated using standard Poisson statistics, by comparing the actual number of counts received that orbit to the number of counts expected based on the average flux. All orbits with a probability of 0.1 or less are plotted on this histogram; so for a steady source we would expect the histogram to contain a total number of events equal to 10% of the number of orbits. Kburst outputs the probability and time of occurrence of the single most unlikely orbit. To get the significance of this 'best burst' candidate simply multiply the probability by the number of orbits.

• After all plots have been displayed, the user is given the option to make hard copies of the plots using the last values entered for plotting.