Imperfect Ambiguity Resolution

There are a number of reasons why TCA fails to do a perfect job:-

• An event that is a part of an association may fail to trigger the detector.

• Unrelated events may mimic events that are part of an association.

• the TCAR may be too restrictive. For example, the setting of the time gate is likely to involve some degree of compromise between association detection efficiency and the time TCA takes to run and the memory it requires.

Loss of a few associations may be easy to correct for, but spurious associations can be much more problematic. Setting:-

$tca_ambig_req 1

will help if a good definition of association quality can be framed in the TCAR, but it will not prevent events turning up as secondaries in other associations. In particular it is possible, depending on the conditions imposed by the TCAR bank, for form two associations from the same two events, with each taking its turn as the primary. The only certain to prevent events being used more than once, without being arbitrary, is to record the quality of all associations and then pick the best, non-conflicting set. Unfortunately, that involves either keeping all events in memory at once or many passes over data, neither of which are viable procedure. So the best that can be done is to pay close attention to the TCAR and to try to reduce spurious ambiguity to a minimum. As TCA only measures ambiguity with respect to the primary, the choice of event type that takes on this role should be the one for which the fewest spurious candidates will be found. TCA lists the number of candidates of each type found so may be able to provide useful information in this respect.