Experience with Run 1 SAMUS Level 2 Filter

Dmitri Denisov ( .html by J. Linnemann)

Good Documentation: D0 notes 2486, 2131, 1586 describe algorithms, efficiency, rejection in detail
Code was stable (from ``ideological'' point of view)
- throughout Ia
- new for Ib
- modified again early in Ib; kept through 1c
each change improved efficiency
Algorithm was fast, using simple geometry and no drift times
Cal confirmation added rejection of 10 with 95% efficiency
30% of the data written to tape have a good muon reconstructed offline
L2 filter is first step of tracking RECO
- Makes RECO faster
- simplifies optimization and testing
very nice communication with L2 group

15% of data lost due to errors with releases of EXE online version
- wrong RCP
- incorrect "same-event" test (impossible to test offline)
Major Problem was online monitoring
- it's not very simple, because of many potential problems
- 2-mu events are .001 of 1-mu rate and very sensitive to backgrounds, so we lost all 2-mu events for a week, though most EXAMINE plots looked OK
Even with stable algorithms we had to make many modifications due to external factors:
- data format changes
- general code changes (eg rewrites of utilities)
these changes are very time consuming and produce lots of errors