Presentation is loading. Please wait.

Presentation is loading. Please wait.

UK egamma meeting, Sept 22, 2005M. Wielers, RAL1 Status of Electron Triggers Rates/eff for different triggers Check on physics channels Crack region, comparison.

Similar presentations


Presentation on theme: "UK egamma meeting, Sept 22, 2005M. Wielers, RAL1 Status of Electron Triggers Rates/eff for different triggers Check on physics channels Crack region, comparison."— Presentation transcript:

1 UK egamma meeting, Sept 22, 2005M. Wielers, RAL1 Status of Electron Triggers Rates/eff for different triggers Check on physics channels Crack region, comparison with isem Recent performance improvements

2 UK egamma meeting, Sept 22, 2005M. Wielers, RAL2 Performance of e25i trigger  =80.0% with b/EC crack included,  =75.2% with crack excluded In 3.4 (4.6)% of events iPatRec finds track, but not SiTrack (IdScan) LVL2 rejects jet events which EF can’t reject, not very nice… Eff %Rates L kHz L2 Calo kHz L2 ID Hz L2 Match Hz EF Calo Hz EF ID Hz EF Match Hz No L Hz SiTrackIDScan Eff %Rates L kHz L2 Calo kHz L2 ID Hz L2 Match Hz EF Calo Hz EF ID Hz EF Match Hz

3 UK egamma meeting, Sept 22, 2005M. Wielers, RAL3 Performance of e25i trigger Cluster composition (number of events) Note: no traceback to conversion in CBNT truth in di-jets though this works correctly in single photon events … ??? However, most of the direct and quark brem photons in final sample are very likely conversion (check done on single  ’s) Implies EF cuts not very efficient on those as we didn’t require cut on b-layer hit Cut on B-layer hit would ‘compensate’ this effect at the price of ~3% loss in efficiency (~ loss in eff from tracking at L2) It’s a matter how you select your events SiTrackNo L2 W  e 24%15% Z  ee8%4% e from b,c decays8%  from direct or quark brem10%18% other50%55%

4 UK egamma meeting, Sept 22, 2005M. Wielers, RAL4 2e15i and e60 2e15i e60 Too many ‘missing’ tracks at L2, so cut not applied For e15i more missing tracks as for e25i For e60 I wanted to have the best eff… Eff %Rates L KHz* L2 Calo Hz EF Calo Hz EF ID88.28 Hz EF Match87.34 Hz Eff % (for 1e) Rates L kHz L2 Calo93.4n/a EF Calo90.7 EF ID83.4 EF Match81.4 Not enough stat to quote rate: 2 evts after L2Calo, 1 after EFCalo, 0 afterwards… few Hz as found in DC1 good guess LVL1 rate given for e25i+e60 ~90% for higher energies Needs to be seen if useful, still many cuts applied, would prefer higher threshold with less cuts for very high energies

5 UK egamma meeting, Sept 22, 2005M. Wielers, RAL5 Check on physics samples Z  ee W  e Efficiencies given after kinematical cuts 2e in |  | 15 GeV for Z  ee 1e in |  | 25 GeV for W  e Effect of e60 on these samples small you wouldn’t use them in analysis as error on eff/turn-on higher than gain Should find out if we don’t want more likely higher threshold which allow relaxing all cuts and knowing we can select very robustly very high energetic electrons 2e15i67.2% e25i92.9% e6020.4% all94.8% 2e15i--- e25i79.6% e606.9% all80.3%

6 UK egamma meeting, Sept 22, 2005M. Wielers, RAL6 Comparison with IsEM In case you wonder why we don’t use IsEM at EF step Don’t apply LVL2 cuts as those were tuned w.r.t. other set of EF cuts LVL1+EF gives for 82.3% efficiency (outside crack) a rate of 600Hz! Shows that DC1 cuts are not optimal! Trying to tune trigger w.r.t. these cuts not too useful…

7 UK egamma meeting, Sept 22, 2005M. Wielers, RAL7 80% efficiency in b/ec crack? Well I did “quick and dirty job” and disabled calo shapes track-cluster match cuts Ended up with 78.5% overall efficiency (compared to 72.2%)  Rate: 51Hz  50% increase in rate! We need to understand better this region and how it should be handled

8 UK egamma meeting, Sept 22, 2005M. Wielers, RAL8 Recent changes in L2 tracking General Use LVL2 cluster position as input for tracking compared to LVL1 position Problem with SCT SP’s fixed IDScan (  IDSCAN ) RoI size in phi reduced by a factor of 2 Less ‘bad’ SP’s picked up Z-finder looks for tracks from first vertex, if this fails look for tracks from 2 nd vertex,… up to the 4 th one SiTrack (  SiTrack ) Use SP’s from larger region than default Results in early rejection of some SP’s Default improves timing and should be used for high lumi Set via jobOptions

9 UK egamma meeting, Sept 22, 2005M. Wielers, RAL9 Performance in SiTrack New version has ~1% better track finding efficiency In 2.6 (3.4)% of events iPatRec finds a track, but SiTrack not in (10.0.1) IDScan New version has ~4% better track finding efficiency! In 1.3 (4.6)% of events iPatRec finds a track, but IdScan not in (10.0.1) w/o L2 tracking on: eff=84.2% Eff % Eff % L196.7 L2 Calo L2 ID L2 Match EF Calo EF ID EF Match Eff % Eff % L196.7 L2 Calo L2 ID L2 Match EF Calo EF ID EF Match

10 UK egamma meeting, Sept 22, 2005M. Wielers, RAL10 Performance in Rate/efficiencies re-calculation Got hold of ~70% of events which pass EF cuts and re- processed them My new files contain currently 1-2 events more, 1 st and last Now I find again conversion in truth??? Using SiTrack at L2 and ‘old’ selection cuts For an eff of 81.3%: R = 40 Hz Using IDscan at L2 and ‘old’ selection cuts (optimised using SiTrack) For an eff of 83.0%: R = 50 Hz  New optimisation for an 80% efficiency will (very likely) yield in ‘old’ rate ~34Hz


Download ppt "UK egamma meeting, Sept 22, 2005M. Wielers, RAL1 Status of Electron Triggers Rates/eff for different triggers Check on physics channels Crack region, comparison."

Similar presentations


Ads by Google