Thanks in part to user feedback and detective work by Bernd Machenschalk and others we have located and are now fixing a significant bug in our search code

Thanks in part to user feedback and detective work by Bernd Machenschalk and others we have located and are now fixing a significant bug in our search code. The search code works in three stages. The first two stages search instrument data from two different periods of time. The third stage searches for consistency (coincidence) in the output from these two time periods. The bug is in this final third stage. For a few of the input data sets, this third coincidence step (which should take only a minute or so) takes much longer because the computation time increases from N log N to order N-squared. The results are still correct but the computation can last for hours. Xavier Siemens is rewriting this routine to guarantee N log N behavior; we hope to have a revised application ready for testing within a few days.