Measurement of W and Z boson cross sections in pp collisions at 7 TeV with the ATLAS detector

Transcription

1 Measurement of W and Z boson cross sections in pp collisions at 7 ev with the ALAS detector Ryan Reece University of Pennsylvania On behalf of the ALAS Collaboration [1] photo: naotakem he International Europhysics Conference on High Energy Physics (EPS-HEP 2011) Grenoble, France July 22, 2011

2 he Large Hadron Collider 27 km circumference pp collisions at s = 7 ev instantaneous luminosity cm 2 s ns bunch spacing Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

3 he ALAS Experiment 3000 scientists 38 countries 174 institutions 0 M readout channels tracking calorimetry muon spectrometry massive worldwide grid computing, 1 PB / year Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

4 Z µµ candidate with additional soft pile-up interactions. High p leptons allow us to select the interesting EW events. Conversely, W/Z provide events for understanding high p lepton performance. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

5 he dataset pb > 1 fb [3] ALAS Data Summary June 20: Observation of W lν and Z ll Oct 20: Measurement of W lν and Z ll cross sections Nov 20: Observation of W τ h ν Feb 2011: Observation of Z ττ lτ h Mar 2011: Observation of Z ττ ll July 2011: Measurement of p Z July 2011: Measurement of p W July 20: Update: Measurement of W lν and Z ll cross sections July 2011: Measurement of Z τ τ cross section July 2011: Measurement of W τ ν cross section Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

6 he dataset he results shown pb today use the 20 dataset, > where 1 fb our detector related systematics have improved and the pile-up is less. At the end, I will show a hint of our Z ττ events in the 2011 data. [3] ALAS Data Summary June 20: Observation of W lν and Z ll Oct 20: Measurement of W lν and Z ll cross sections Nov 20: Observation of W τ h ν Feb 2011: Observation of Z ττ lτ h Mar 2011: Observation of Z ττ ll July 2011: Measurement of p Z July 2011: Measurement of p W July 20: Update: Measurement of W lν and Z ll cross sections July 2011: Measurement of Z τ τ cross section July 2011: Measurement of W τ ν cross section Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

7 Ingredients for a cross section σ = N obs N bkg A C dtl N obs : number of observed events in the signal region N bkg : estimated number of background events EW backgrounds are estimated with Monte Carlo, constrained to data with performance scale factors. QCD backgrounds are estimated with data-driven methods. A: kinematic acceptance factor, estimated with generator-level Monte Carlo. C: summarizes reconstruction efficiency, estimated with reconstructed Monte Carlo, corrected with scale factors. dt L: integrated luminosity. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

8 Progress in reducing systematics he main advancements in W and Z cross section measurements since last winter are due to the reduction of systematics. Examples: ag and probe studies lead to many improvements in electron and muon identification and scale factors correcting our simulation thereof. Reduced acceptance uncertanties by switching from using a LO Pythia generator to MC@NLO, p W and pz reweighted to better agree with data (discussed later). Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

9 ag and probe studies Events / 1 GeV Events / 1 GeV ALAS preliminary Data 20, s=7 ev, Ldt 40 pb Data Fit Signal (BWCB) Background m ee [GeV] ALAS preliminary Data 20, s=7 ev, Ldt 40 pb Data Fit Signal (template) Background m ee [GeV] apply ID tight tag e Z probe e ag events with sufficient purity, leaving an unbiased probe object. Measure probe ID efficiency in situ. Constrains the performance of our object identification. Derive scale factors for correcting our simulation. [4] ALAS-PERF Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

10 W lν and Z ll Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP 2011 / 31

11 Event selection Events / 2.5 GeV W lν One e/µ with p > 20 GeV E miss > 25 GeV m (l,e miss L dt = 36 pb ) > 40 GeV Data 20 ( W eν QCD W τν s = 7 ev) e m [GeV] Events / 1 GeV Z ll wo e/µ with p > 20 GeV m ll = GeV Data 20 ( s = 7 ev) Z µµ QCD L dt = 33 pb [6] SDM m µµ [GeV] Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

12 QCD background estimation W eν: template fit to E miss. emplate derived from data with inverted electron ID and isolation. Z ee: template fit to m ll to a sample with looser electron ID, extrapolated to the signal region. W µν: matrix method using track isolation. Z µµ: ABCD method with track isolation in m µµ side-band. Events / 2 GeV e Data 20 ( s = 7 ev) W eν + EW + tt QCD L dt = 36 pb Events / 1 GeV Data 20 ( s = 7 ev) Z ee QCD W eν L dt = 36 pb miss E [GeV] m ee [GeV] Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

13 Cross section results σ [nb] stat. sys. lumi. acc. σ W BR(W eν) = σ W BR(W µν) = σ Z/γ BR(Z/γ ee) = σ Z/γ BR(Z/γ µµ) = L dt = pb Data 20 ( s = 7 ev) total uncertainty exp. uncertainty ABKM09 JR09 MSW σ tot W / σ tot Z/γ* Dominant uncertainty is luminosity. Acceptance uncertainty remains significant due to the extrapolation from the fiducial volume. Detector related uncertainties partially cancel in the ratio. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

14 Cross section results σ [nb] stat. sys. lumi. acc. σ W BR(W eν) = σ W BR(W µν) = σ Z/γ BR(Z/γ ee) = Many σ Z/γ more BR(Z/γ details onµµ) the W= lν0.935 and Z/γ ll0.009 cross section measurements, including differential measurements of dσ/dη and Dominant uncertainty their constraints on proton PDFs will be discussed in Massimiliano is luminosity. Bellomo s talk in the QCD session this afternoon. L dt = pb Data 20 ( s = 7 ev) total uncertainty exp. uncertainty ABKM09 JR09 MSW σ tot W / σ tot Z/γ* Acceptance uncertainty remains significant due to the extrapolation from the fiducial volume. Detector related uncertainties partially cancel in the ratio. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

15 p W and pz Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

16 Z boson p measurement Important for modeling high-p lepton kinematics. At leading order, p W/Z = 0 Non-zero p W/Z is generated through the hadronic recoil of ISR, p R. p Z reconstructed directly from p (µ 1 )+p (µ 2 ), while p W reconstructs pr. Detector and FSR effects removed with a bin-by-bin unfolding. η l < l p > 20 GeV 66 GeV < m ll < 116 GeV % precision per bin. p Z [GeV] [1/GeV] dσ fid /dp 1/σ fid 0.06 ALAS 0.05 Z L dt = pb Combined ee+µµ Data 20 RESBOS PYHIA [7] arxiv: v1 (a) Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

17 W boson p measurement Necessary for a future precision W mass measurement. Detector and FSR effects removed by inverting a response matrix parametrizing the probabilistic mapping of p R to pw. Events / GeV Ldt 31 pb Data W µν Z µµ W τν Jets top Z ττ ] [GeV W / dp dσ fid 1/σ fid Ldt 31 pb -5 Data R p [GeV] Data / RESBOS -6 RESBOS p W [GeV] [8] SDM Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

18 W, Z boson p reweighting he modeling of dσ/dp W/Z can have significant effects on the expected efficiency and acceptance. NLO generators MC@NLO and POWHEG have deficits at high p W/Z. NLO effects are important at high p W/Z because the W/Z is polarized by higher (Data,Prediction) / RESBOS order QCD. p W [GeV] 2 1 Combined Data Stat. Uncert. ALPGEN MC@NLO POWHEG PYHIA RESBOS Ldt 31 pb W lν and Z ll cross section measurements use MC@NLO reweighted to match p W/Z for LO Pythia, which agrees with the data because it has been tuned well to the evatron data. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

19 Z ττ and W τν Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

20 aus in ALAS racks are matched to jet seeds and discriminating variables are calculated from the combined tracking and calorimeter information. 1 or 3-prong signature Narrow clustering of tracks and calorimeter deposits. hree advanced discriminants: p -parametrized cuts, projective likelihoods, boosted decision trees. Arbitrary Units 0.14 W τν+z ττ dijet Monte Carlo dijet data dt L = 23 pb prongs 15 GeV<p <60 GeV [9] ALAS-CONF R EM Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

21 Z ττ lτ h event selection single lepton trigger One tight e/µ with p > 16/15 GeV One tight τ h with p > 20 GeV, 1 or 3 tracks, unit charge tight lepton isolation to reject QCD multijets opposite sign cos φ > 0.15 Muons / 5 GeV 120 Data s = 7 ev Ldt = 36 pb γ*/z ττ Multijet W lν W τν γ*/z ll tt Muons / s = 7 ev Ldt = 36 pb Data γ*/z ττ Multijet W lν W τν γ ll tt*/z p (µ) [GeV] I E / p (µ) Choose low p leptons to accept Z ττ. Combat QCD with tight isolation. [] SDM Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

22 Z ττ: W + jet suppression cos φ = cos[φ(l) φ(e miss)] +cos[φ(τ h ) φ(e miss)] Events / s = 7 ev Ldt = 36 pb Data γ*/z ττ Multijet W lν W τν γ ll tt*/z Quantifies if the E miss is between the decay products. Only dependent on the direction, not the magnitude of the E miss. All channels except µµ require cos φ > cos φ Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

23 Z ττ ll event selection eµ-channel one e with p > 15 GeV one µ with p > GeV opposite sign, tight isolation cos φ > 0.15 E +E miss < 150 GeV (rejects t t) µµ-channel two µ with p > 20 GeV opposite sign, tight isolation BD for rejecting Z/γ ll m µµ = GeV Events / GeV Ldt = 36 pb s = 7 ev Data γ*/z ττ Multijet W lν γ*/z ll tt m(e,µ) [GeV] Events / 5 GeV s = 7 ev Ldt = 36 pb Data γ*/z ττ γ*/z ll W lν Multijet tt Diboson m(µ, µ) [GeV] Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

24 Z ττ background estimation Jets are wider in data than in Monte Carlo τ h ID fake rate is underestimated in Monte Carlo. Normalize W lν +jets Monte Carlo using high m data. Other EW backgrounds estimated from Monte Carlo. QCD estimate is data-driven, scailing SS data by R OS/SS, measured in non-isolated (multijet rich) data sample, correcting for EW contamination with Monte Carlo. Events / 5 GeV s = 7 ev Ldt = 36 pb Data γ*/z ττ Multijet W lν W τν γ ll tt*/z m [GeV] N A = N B NC N D = NB R OS/SS Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

25 Z ττ cross section results σ combined = 0.97±0.07(stat.)±0.07(sys.)±0.03(lumi.) nb σ theory = 0.96±0.05 nb at NNLO Z ττ combined 36pb Z ee/µµ 33-36pb τ µ τ h τ e τ h Stat τ e τ µ Syst Stat Syst Stat Lumi τ τ heory (NNLO) µ µ σ(z ll, 66<m <116 GeV) [nb] inv Dominant systematics τ h energy scale 11% τ h efficiency 8.6% µ efficiency 8.6% e efficiency 30% acceptance 3% luminosity 3.4% Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

26 W τν event selection Number of events / Data 20 ( s = 7 ev) W τ h ν τ EW background QCD background (CD) L dt = 34 pb Number of events / 5 GeV Data 20 ( s = 7 ev) W τ h ν τ EW background QCD background (C) L dt = 34 pb S miss E [GeV] m p (τ h ) = GeV E miss > 30 GeV E S E miss = miss (0.5 > 6.0 GeV) E veto events with leptons with p > 15 GeV data-driven QCD estimate: scale events failing S E miss efficiency measured in inverted tau ID sample. by [11] SDM Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

27 W τν cross section results σ(w τν) = 11.1±0.3(stat.)±1.7(sys.)±0.4(lumi.) nb σ theory =.46±0.52 nb at NNLO ALAS W τ ν τ Data 20 ( s = 7 ev) Stat uncertainty Sys Stat ALAS W e ν e Sys Stat Lumi Prediction (NNLO) heory uncertainty ALAS W µ ν µ σ(w l ν l ) [nb] Dominant systematics τ h efficiency.3% τ h energy scale 8.0% τ h + ME trigger efficiency 7.0% luminosity 3.4% acceptance 2.3% Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

28 W τ h ν tau ID measurement Events Data 20 (34 pb ) τ (W τν) e (W eν) Jet background Before tau ID ag jet + ME events, probe for tau. 7 Looser identification working point L = 34 pb Cuts 1.04 ± 0.06(stat) ± 0.04(sys) (ag&probe) Events Number of tracks Data 20 (34 pb ) τ (W τν) e (W eν) Jet background After tau ID (Looser Cuts working point) Number of tracks apply ID 0 Likelihood 1.02 ± 0.09(stat) ± 0.05(sys) (ag&probe) Boosted decision trees (ag&probe) Cuts 1.00 ± 0.05(stat) ± 0.13(sys) (Cross section) Likelihood (Cross section) 1.05 ± 0.06(stat) ± 0.05(sys) 1.02 ± 0.04(stat) ± 0.16(sys) Boosted decision trees 0.94 ± 0.07(stat) ± 0.13(sys) (Cross section) au identification efficiency scale factor [12] ALAS-CONF Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

29 W τ h ν tau ID measurement Events Events Data 20 (34 pb ) τ (W τν) e (W eν) Jet background Looser identification working point 600 Before tau ID L = 34 pb 400 Cuts 1.04 ± 0.06(stat) ± 0.04(sys) (ag&probe) Many 200 more details on tau performance studies have been discussed Number of tracks Data 20 (34 pb ) τ (W τν) e (W eν) Jet background After tau ID (Looser Cuts working point) Number of tracks ag jet + ME events, probe for tau. in Stan Lai s talk in the Detector session this morning. 0 Likelihood 1.02 ± 0.09(stat) ± 0.05(sys) (ag&probe) Boosted decision trees (ag&probe) Cuts 1.00 ± 0.05(stat) ± 0.13(sys) (Cross section) Likelihood (Cross section) 1.05 ± 0.06(stat) ± 0.05(sys) 1.02 ± 0.04(stat) ± 0.16(sys) Boosted decision trees 0.94 ± 0.07(stat) ± 0.13(sys) (Cross section) au identification efficiency scale factor Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

30 Z ττ lτ h with 730 pb Events / 5 GeV 250 After tight BD ID dt L = 730 pb s = 7 ev data Z ττ (truth-matched) Z ττ (non-truth-matched) W µ ν Multijet Z µ µ tt W τ ν m vis (µ,τ ) [GeV] We now have a substantial control sample of hadronic tau decays. More data-driven efforts in taus to come. h [13] AL-COM-PHYS Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

31 Conclusions ALAS has published (or will soon publish this month) W and Z cross section measurements with the 20 dataset for all lepton flavors: W eν W µν W τν Z ee Z µµ Z ττ We have also published precision measurements of the p W and p Z line-shapes. W and Z performance studies have improved our knowledge of our object identification for e, µ, and τ. he 20 ALAS Standard Model analyses have set an impressive standard for measurements with the 2011 data. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

32 Backup Slides Ryan Reece Penn ALAS W and Z cross sections EPS-HEP / 31

33 References I [1] Photo credit: [2] ALAS public event displays. [3] he ALAS Data Summary. [4] he ALAS Collaboration. Electron performance measurements with the ALAS detector using the 20 LHC proton-proton collision data. ALAS-PERF (publication draft pending internal review). July [5] he ALAS Collaboration. Measurement of the W lν and Z/γ ll production cross sections in pp collisions at s = 7 ev with the ALAS detector. arxiv:.2130v1 [hep-ex]. Oct 20. [6] he ALAS Collaboration. Measurement of the W lν and Z/γ ll production cross sections in pp collisions at s = 7 ev with the ALAS detector. SDM (publication draft pending internal review). July Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

34 References II [7] he ALAS Collaboration. Measurement of the transverse momentum distribution of Z/γ bosons in pp collisions at s = 7 ev with the ALAS detector. arxiv: v1 [hep-ex]. July [8] he ALAS Collaboration. Measurement of the transverse momentum distribution of W bosons in pp collisions at s = 7 ev with the ALAS detector. SDM (publication draft pending internal review). July [9] he ALAS Collaboration. Reconstruction, Energy Calibration, and Identification of Hadronically Decaying au Leptons. ALAS-CONF May [] he ALAS Collaboration. Measurement of Z ττ production cross section in pp collisions at s = 7 ev with the ALAS detector. SDM (publication draft pending internal review). July [11] he ALAS Collaboration. Measurement of the W τν production cross section in pp collisions at s = 7 ev with the ALAS Experiment. SDM (publication draft pending internal review). July [12] he ALAS Collaboration. Measurement of hadronic tau decay identification efficiency using W τν events. ALAS-CONF July [13] he ALAS Collaboration. Approved preliminary plots of m vis for Z ττ lτ h events with 730 pb. AL-COM-PHYS July Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

35 Collider kinematics Ryan Reece Penn ALAS W and Z cross sections EPS-HEP / 31

36 Phenomenology of tau decays τ 17.8% 9.0% e ν e ν τ µ ν µ ν τ 17.4% π π 0 ν τ 25.5% π ν τ.9% π 2π 0 ν τ 9.3% K (Nπ 0 )(NK 0 )ν τ 1.5% π 3π 0 ν τ 1.0% π π π + ν τ π π π + π 0 ν τ 4.6% } leptonic 35.2% 1 prong 49.5% } 3 prong 15.2% Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

37 au identification variables Electrmagnetic radius: R EM = rack radius: R track = Ri <0.4 i {EM 0 2} EEM,i R i Ri <0.4 i {EM 0 2} EEM,i Ri <0.4 i p,i R i Ri <0.4 i p,i Leading track momentum fraction: f track = ptrack,1 Core energy fraction: f core = Electromagnetic fraction: f EM = p τ Ri <0.1 E EM i {all},i Ri <0.4 E EM i {all},i Ri <0.4 i {EM 0 2} EEM,i R j <0.4 E EM j {all},j Cluster mass: m clusters, invariant mass clusters at the EM energy scale. rack mass: m tracks, invariant mass of the track system. ransverse flight path significance: S flight Motivation: taus tend to be collimated more than jets, have a leading track, and often significant neutral pion deposits in the EM calorimeter. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

38 au discriminants Cuts p -parametrized cuts on R EM and R track, and a cut on f track. Projective) likelihood d = ln( LS L B = ( ) N i=1 ln p S i (x i ) p B i (x i) Boosted decision trees (BD) track R Arbitrary Units p [GeV] W τν+z ττ dijet Monte Carlo 3 prongs 15 GeV<p <60 GeV 20 dijet data dt L = 23 pb Likelihood Score Arbitrary Units 0.12 W τν+z ττ dijet Monte Carlo 3 prongs 15 GeV<p <60 GeV dijet data dt L = 23 pb BD Score Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

39 Seeing hadronic taus Number of events Data 20 ( s = 7 ev) W τ h ν τ EW background QCD background (B) L dt = 34 pb Number of events / Data 20 ( s = 7 ev) W τ h ν τ EW background QCD background (BD) L dt = 34 pb Number of tracks R EM Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

40 Electrons in ALAS Seeded by matching calorimeter clusters from a sliding-window algorithm to inner detector tracks. Candidates are selected by: track quality track-cluster matching narrow calorimeter cluster high electromagnetic fraction ight candidates have cuts on E/p and high thesholds hits from the transistion radiation in the R. Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

41 Muons in ALAS Combination of muon spectrometer segments with inner detector tracks. rack combination matching to reject decays in flight. Impact parameter constraints to reject cosmic muons. Ryan Reece Penn ALAS W and Z cross sections EPS-HEP / 31

42 W lν and Z ll systematics reduction Moriond result March 2011 July 2011 update L dt = pb Data 20 ( s = 7 ev) total uncertainty exp. uncertainty ABKM09 JR09 HERA MSW σ / σ W Z/γ* L dt = pb Data 20 ( s = 7 ev) total uncertainty exp. uncertainty ABKM09 JR09 MSW σ tot / σ tot W Z/γ* Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

43 Z boson p measurement Important for modeling high-p lepton kinematics. At leading order, p W/Z = 0 Non-zero p W/Z is generated through the hadronic recoil of ISR, p R. [1/GeV] dσ fid /dp 1/σ fid 0.06 ALAS 0.05 Z η l < 2.4 L dt = pb l Detector and FSR p > 20 GeV 66 GeV < m effects removed with ll < 116 GeV 0 a bin-by-bin p [GeV] unfolding. Z Combined ee+µµ Data 20 RESBOS PYHIA (a) σz i p i = 1 Nobs i Ni bkg p A i C i dtl, Ci = C i MC εid,i data ε ID,i MC εtrig,i data ε trig,i MC Ryan Reece Penn ryan.reece@cern.ch ALAS W and Z cross sections EPS-HEP / 31

44

45

46

47

48

49