PFA update Lei Xia ArgonneHEP m RMS 90
- Slides: 14
PFA update Lei Xia, Argonne-HEP m. RMS 90 calculation m. PFA at Z-pole m. First look at 200 Ge. V (qqbar uds events) m. First look at 500 Ge. V (qqbar uds Si. D workshop, SLAC Oct 26 -28 events)
RMS 90 calculation m. My calculation is available at the following locations: q. CVS: org. lcsim/contrib/Xia. Lei/rms 90_order. java qhttp: //www. hep. anl. gov/lxia/rms 90_order. java m. Everyone is welcome to use/test it – please let me know if you meet any problem q. Just need to change one line in the program l ICloud 1 D Input. Cloud = (ICloud 1 D) aida. Master. Tree. find("/aida-tree/1 d. Cloud"); m. RMS 90 definition q. Any 90% of the events that gives the smallest rms q. Slightly different from Mark Thomson’s (rms of the smallest region that contain 90% of all events) Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
PFA outline Calorimeter Hits Clustering Algorithm Tracker Hits Track finding Algorithm Calorimeter Clusters Reconstructed Tracks Photon Identification EM Clusters Hadron Clusters Track-cluster matching ‘Neutral’ Clusters Matched Clusters Charge fragment identification Neutral Clusters EM sampling fraction Ephoton E/p check Fragments Hadron sampling fraction Eneu-had 0 0 Si. D workshop, SLAC Oct 26 -28 Ptrack Total event energy Lei Xia, ANL
PFA at Z-pole: sidaug 05_np No cut Barrel event Mean 87. 4 Ge. V RMS 7. 33 Ge. V RMS 90 4. 56 Ge. V [49%/sqrt(E)] Mean 88. 2 Ge. V RMS 6. 63 Ge. V RMS 90 4. 28 Ge. V [46%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) Detector model: Si. Daug 05_np (none projective cells) PFA: no change from Vancouver, except adding E/P check parameters for clustering, etc. are not tuned yet… Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
PFA at Z-pole: sidaug 05 No cut Barrel event Mean 87. 3 Ge. V RMS 7. 13 Ge. V RMS 90 4. 44 Ge. V [48%/sqrt(E)] Mean 88. 0 Ge. V RMS 6. 45 Ge. V RMS 90 4. 18 Ge. V [45%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) Detector model: Si. Daug 05 (projective cells) This is just a sanity check, since higher energy samples are only available for projective detectors Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
PFA at higher CM energies m Data sample: qqbar (uds events) q At 200, 500 Ge. V q With Si. Daug 05 m PFA: no change m Some changes in the program, just to run new data file q Explicitly used beam energy at one place q Need to handle some new ‘Runtime. Exception’s, that didn’t occur with Z-pole events m Just a first look… q Was able to run only a small data sample (clustering takes very long time ~(N[hits])2) q Results should not be treated as a measure of PFA or Si. D performance (PFA not tuned, event at Z-pole) Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
First look at 200 Ge. V No cut Barrel event Mean 214. 2 Ge. V RMS 28. 9 Ge. V RMS 90 19. 6 Ge. V [134%/sqrt(E)] Mean 216. 1 Ge. V RMS 27. 3 Ge. V RMS 90 19. 4 Ge. V [132%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) Something in my PFA stopped working at this energy! Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
First look at 500 Ge. V No cut Barrel event Mean 526. 6 Ge. V RMS 74. 5 Ge. V RMS 90 49. 2 Ge. V [214%/sqrt(E)] Mean 531. 9 Ge. V RMS 59. 4 Ge. V RMS 90 46. 4 Ge. V [201%/sqrt(E)] (cos(theta) < sqrt(2)/2) This is significantly worse than just summing up calorimeter energy Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
What’s wrong? m. Noticed the track-cluster association was quite loose q. OK for Z-pole events (clusters are relatively far apart) q. Should use a tighter criteria for higher energies Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
200 Ge. V with tight track-cluster association No cut Barrel event Mean 201. 3 Ge. V RMS 19. 1 Ge. V RMS 90 11. 6 Ge. V [82%/sqrt(E)] Mean 200. 4 Ge. V RMS 16. 8 Ge. V RMS 90 10. 9 Ge. V [77%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) PFA need to be tuned/modified for higher energy Much better performance should be possible Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
500 Ge. V with tight track-cluster association No cut Barrel event Mean 461 Ge. V RMS 56. 4 Ge. V RMS 90 33. 4 Ge. V [156%/sqrt(E)] Mean 462 Ge. V RMS 39. 9 Ge. V RMS 90 30. 1 Ge. V [140%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) PFA need to be tuned/modified for higher energy Much better performance should be possible Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
Back to Z-pole m. Do we really need loose track-cluster association at Z-pole? m. Try the tight criteria for Z-pole events as well Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
Back to Z-pole: sidaug 05_np No cut Barrel event Mean 87. 0 Ge. V RMS 6. 03 Ge. V RMS 90 3. 81 Ge. V [41%/sqrt(E)] Mean 87. 5 Ge. V RMS 5. 12 Ge. V RMS 90 3. 54 Ge. V [38%/sqrt(E)] (cos(theta[Q]) < sqrt(2)/2) Got it last night – so, very preliminary! Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL
Summary m PFA performance at Z-pole significantly improved from VLCW’ 06 q New result need to be confirmed q Plenty of room for further improvement m PFA constructed at Z-pole doesn’t (automatically) work at higher energy q Tuning/modification is necessary q PFA performance should be significantly improved m My PFA code q Will put some latest version into CVS q Code has been messed up by small modifications over time q A major re-write is expected – this time will try to follow PFA template whenever possible Si. D workshop, SLAC Oct 26 -28 Lei Xia, ANL