1. 1 Status of Cosmic Analysis Malcolm Ellis Imperial College London Wednesday 10 th March 2004

2. 2 Outline Status at Osaka Data-taking March 2004 Clustering Pattern Recognition Light Yield Efficiency Question mark over geometry/decoding/PR Further Work

3. 3 Status at Osaka Makoto and I had received correct VLPC decoding and confirmed full decoding scheme. Tracks were successfully built and first light yield measurements with tracking look promising. Agreed need for more data…

4. 4 Data-Taking March 2004 Paul set up new cassette (number 110) and modified trigger system Alan and I attached waveguides Data taking resumed on Friday (5 th ) Weekend spent doing timing scans and debugging DAQ (but data still useable for tracking and LY studies!) Date/Time of event added from Run 214 (taking now…) to check whether DAQ rate is flat in time.

5. 5 Data-Taking Rate Over last 24 hours, taken approximately 2300 triggers. Of these 302 (13%) have a reconstructed track. Note, a simple estimate of rate of good triggers gives ~60%, so factor of 4 to be understood… Based on 10 more days of data-taking, we should add another 23000 triggers, and hence 3000 tracks. This gives an estimated total of 5500 tracks by the collaboration meeting! All plots in this talk are from sum of data taken up to and including Run 213 (finished just before lunch today). This comes to 33365 triggers and 2655 tracks.

6. 6 Clustering I now build clusters first. All hits above 1.3 PE are considered. A hit on its own, or two neighbouring hits above 1.3 PE are merged into a cluster For a cluster to be retained, its total light must be more than the usual 2.6 PE. Clusters are then the basis for making points and tracks…

7. 7 Pattern Recognition Once clusters are built, triplets are built in Station B. All unused hits in B (and of course A and C) are built into duplets. Points which are in a reasonable radius (or in the case of triplets, small residual) are kept for track building.

8. 8 Track building All combinations of points are used to build a track The combination with the lowest chi2 is kept. The track is then re-fitted using a Least Squares fit. Still room for improvement to increase track-finding efficiency slightly

9. 9 Extrap to Trigger Scintillator

10. 10 Light Yield We have planes with different amounts of 3HF: –5000 ppm:A/X, A/V, B/X, C/W –3500 ppm: B/W –2500 ppm:B/V Plane W on Station C is a mixed plane and won’t be considered.

11. 11 5000 ppm 3HF

12. 12 3500 ppm 3HF

13. 13 2500 ppm 3HF

14. 14 Efficiency Same algorithm as reported at Osaka, except now clusters are used, instead of single hits Results: –Plane X (5000 ppm) – 93.44 ± 0.62 % –Plane W (3500 ppm) – 95.94 ± 0.50 % –Plane V (2500 ppm) – 98.10 ± 0.36 % 2500 ppm seems best of the three.

15. 15 Geometry/Decoding OK? In general, scanning through events, they look good and can be broken down into 4 categories: –No hits at all, so cosmic didn’t go near the tracker –Some hits, which usually have a point in one or two trackers, but not all 3 –Hits and a track –7 or more hits, but no track…

16. 16 No Hits

17. 17 Some Hits – Grazing Track?

18. 18 Good Track!

19. 19 ??????

20. 20 Further Work Now have gain data from channel-by-channel fit to LED data done by Juan for last year’s and this year’s data. I will incorporate this into code so that more accurate PE is estimated. Need to repeat Light Yield as function of space studies done at Osaka. Extend Efficiency calculation to all planes Check potential problem in geometry/decoding/PR Prepare and run G4MICE and present results at collaboration meeting…