25
E/gamma Triggers Sam Harper RAL E/gamma Meeting 06/09/10

E/gamma Triggers

  • Upload
    deanna

  • View
    44

  • Download
    5

Embed Size (px)

DESCRIPTION

E/gamma Triggers. Sam Harper RAL E/gamma Meeting 06/09/10. Introduction. E/gamma trigger menu is evolving rapidly we are in an exceptional time of the experiment where the LHC luminosity is rapidly increasing we are not fully commissioned we have two new menus left this year - PowerPoint PPT Presentation

Citation preview

Page 1: E/gamma Triggers

E/gamma Triggers

Sam Harper

RAL

E/gamma Meeting

06/09/10

Page 2: E/gamma Triggers

2/8

Introduction • E/gamma trigger menu is evolving rapidly

– we are in an exceptional time of the experiment where the LHC luminosity is rapidly increasing

– we are not fully commissioned

• we have two new menus left this year– 2E31, deploying imminently – 6E31, first proposal earlier today in TSG

• note: in 2E31: all triggers have H/E<0.15 applied– names do not change (this is still being debated…)– Photon filternames go from PhotonEtXCleanedHcalIsolFilter

to PhotonEtXCleanedHEFilter – this is not a new thing, it was originally planed to have H/E

on all EG triggers at L1 but this year it has instead moved to HLT

Page 3: E/gamma Triggers

3/8

EGEle at 2E31

• current ele triggers at CMS, we have:– standard (H/E<0.15 + PM– CaloEleID (sigmaIEtaIEta < 0.014 <0.03)– EleId (dEtaIn,dPhiIn,sigmaIEtaIEta– EleIdIsol (dEtaIn, dPhiIn, sigmaIEtaIEta, isolEm, isolHad,

isolTrk)

Page 4: E/gamma Triggers

4/8

Method

Data:• bkg: Photon10 skim• signal: a skim of W (MET>30GeV) /Z (two eles)

passing WP95 in 361p4 prompt reco

MC:• 381 W->enu RelVal• RECO+ HLT:

– CMSSW_3_8_1:• RECO GT MC: MC3XY_V9, data: GR10_P_Vl• HLT GT MC: MC3XY_V9, data: GR10_H_V8

Page 5: E/gamma Triggers

5/8

CaloID Variables: All SC

X/Y:

X = nr events with RECO value < cut where HLT cut fails

Y = nr events with RECO value < cut

84/56931

(0.0015)

43/36526

(0.0012)

21/47644

(0.0004)

50/25570

(0.0020)

Differences all at per mill level

note: events are required to be “good” η region

Page 6: E/gamma Triggers

6/8

Calo Isol Variables: All SC

X/Y:

X = nr events with RECO value < cut where HLT cut fails

Y = nr events with RECO value < cut

note: events are required to be “good” η region

123/24983

(0.005)

184/23501

(0.008)

328/53727

(0.006)

958/24113

(0.04)

Page 7: E/gamma Triggers

7/8

Track matching Variables: Zs

• note, the matching is pretty bad for all jets

• better for Zs• also note: I

don’t think I have the alignment properly set in endcap

note: events are required to be “good” η region

Page 8: E/gamma Triggers

8/8

Tracker Isolation: Zs

• tracker isolation is not well reproduced in the HLT – I’ve tried to make version which is always smaller although

this is challenging

Page 9: E/gamma Triggers

9/8

Trigger Performance: 2E31 Menu

• Z dependence in startup windows eff:– been investigating, some EB events are actually failing

dphi cuts… – also includes H/E cut

Page 10: E/gamma Triggers

10/8

PM + H/E Efficiency: Ws

Page 11: E/gamma Triggers

11/8

Ele12 EleIdIsol Eff after PM+H/E

• efficiency after L1+ET cut + PM

• I havnt managed to get the new endcap alignments so removed dEtaIn in endcap– this will run as a backup trigger…

Page 12: E/gamma Triggers

12/8

LEleId Efficiency after PM + H/E

• AKA our online tracking efficiency…• better than MC in barrel, endcap worse

– wondering if the endcap missalignment is contributing

Page 13: E/gamma Triggers

13/8

CaloEleId After PM + H/E

• now theres a trigger efficiency I like

Page 14: E/gamma Triggers

14/8

EleId Efficiency

• still have alignment issue in endcap so no point showing it

Page 15: E/gamma Triggers

15/8

Plans for the Future

• GsfTracking in the HLT: A. Holzner• double object triggers

– cross – channel triggers, ie with mu’s, MET etc– single seeded di-pho/ele triggers

• only require 1 obj to pass L1• worked on by R. Covarelli building on work by B. Dahmes

– PAG input vital

• triggering will be even more challenging next year– I encourage every analysis to be thinking about how they

trigger their signal and ensuring its in the menu• single electron triggers with Et<30-40 GeV will require heavy id

Finally EG HLT is undermaned: Please get involved (even if its just making trigger efficiencies for your analysis

Page 16: E/gamma Triggers

16/8

Backup

Page 17: E/gamma Triggers

17/8

Trigger Selection Variables• we have 4 ID variables

– dEtaIn, dPhiIn, sigmaIEtaIEta, H/E

• we have 4 Isolation variables– EM Isolation, HCAL isolation, Trk Pt isolation (Ele), Trk Pt

isolation (Pho)

• I have tried to make these variables either identical or value which is always looser than RECO– this is not always possible– I have very few ways of altering dPhiIn, dEtaIn, they are

what they are

• I consider H/E, sigmaIEtaIEta to be very good, Calo Isolation to somewhat good and dPhiIn, dEtaIn, tracker isolation to be not so good-critiera is solely RECO-HLT agreement

Page 18: E/gamma Triggers

18/8

EGPho at 2E31

• note all E/gamma triggers have H/E<0.15 – 1 exception is a high threshold Photon trigger for safety

• most notable change for exotica is that our uncleaned high PT trigger is gone– it was going to be 500GeV at 6Hz , rate and threshold was

too high to justify and wouldn’t have survived next menu

Page 19: E/gamma Triggers

19/8

6E31

Page 20: E/gamma Triggers

20/8

Photon35

• various options for this trigger:– sigmaIEtaIEta– tighten track isolation variable

• cone size to 0.03, min pt to 1.0 GeV• decreasing veto cone size 0.03

– tighting cuts – going to 0.4 cones

• note I noticed when checking, ele/pho isol is different– pho has 2.5 crystals width, 3.5 cone

Page 21: E/gamma Triggers

21/8

Pho35 Eff

• good efficiency

Page 22: E/gamma Triggers

22/8

Revised Menu

• the dEtaIn safety trigger adds 0.5Hz to a 34.8 Hz• overlap with EGPho is 7.8Hz (22% of EGEle) • I would also like to run a HLT_Ele17_SW_EleId_L1R

trigger– this adds no rate but if people want a single electron trigger

at 17 GeV at 6E31, this is a candidate…

• both triggers are in confdb

Page 23: E/gamma Triggers

23/8

Validation of Paths

• for 25 GeV Z->ee WP95 electrons which pass L1 and are fiduical in EB or EE (total 746)

• EleId paths are affected by dEtaIn missalignment– in theory should be fixed but I have got the right settings?

Trigger N pass Eff

Ele10 738 98.9 ± 0.3

Ele12_EleIdIsol 687 92 ± 1

Ele12_EleIdIsol_NoDEtaInEE 717 96 ± 1

Ele12_EleId 689 92 ± 1

Ele17 738 98.9 ± 0.3

Ele17_CaloEleId 738 98.9 ± 0.3

Ele17_EleId 689 92 ± 1

Ele17_LEleId 722 97 ± 1

Ele22_CaloEleId 738 98.9 ± 0.3

Photon20 744 99.7 ± 0.2

Page 24: E/gamma Triggers

24/8

Proposal: ET thresholds

• physics wants a 15 GeV trigger for 20 GeV electrons– HLT turn on better then this, we propose 17 GeV for 20GeV

– plateau is at 18 GeV• curve driven by difference in ET w.r.t (0,0,0) and w.r.t primary vertex

• at this stage HLT does not distinguish between ele + phos

Page 25: E/gamma Triggers

25/8

Rate reduction

• 2 GeV = ~30% rate reduction: very compelling