CPPM PhD days Measurement of electron efficiency using W → … · 2 The Tag & Probe Method - The tag object: strict requirements to enhance the purity of the sample - The other

Measurement of electron efficiency using W → eν decays.Study of the Higgs production in association with tt quarks

PhD student: Ana Elena DumitriuSupervisors: Cristinel Diaconu (CPPM), Emmanuel Monnier (CPPM),

Alexa Calin (IFIN-HH)

10/7/16 1

CPPM PhD days

2

The Tag & Probe Method- The tag object: strict requirements to enhance the purity of the sample - The other particle serves as a probe for the measurement.

- The efficiency: N selectedN probe

Numerous advantages in using W events:➢ The measurements with W and Z bosons cover both a large kinematic range; ➢ The J/ψ channel completes the measurement at low pT.➢ W provides additional statistics for the low Et range (15-25 GeV) and offers interesting

experimental signatures due to its large statistics ➢ Different systematics from Z

3

W → eν analysisW events topology

● In oder to reduce the background, we test the angular distribution in the transverse plane

● Jets with pT>15,10,5 GeV tested

● Only first 2 leading jets are used && ΔR (e-jet)>0.2

The large suppression by the tight ID algorithm indicate large backgrounds at low PT

ET vs e- menu

Selection• Triggers WTP (prescaled)• MET > 25 (TST)

• |η| < 2.47• mT > 40 GeV

+ wstos < 4, Rη >0.7, RΦ >0.7Probe • rHad and TQ • pT > 15 GeV

ET Probe

All details in backup about the cleaning procedure by using the W topology

4

Trigger tunning in Run II● -ATLAS trigger organised in Level1 +

HighLevelTrigger (HLT)– Use topological variables to reduce the background, use track requirements (trkcut) , use MET (xe) and MET significance (xs)

● Topological variables tuned to reduce trigger rates: tuning at HLT and L1. Recently, algorithms implemented in a new system L1TOPO.

lower rate

10/07/16 5

Trigger menu tuning in offline and online emulation

• Similar efficiency in HLT compared to Emulation with L1 topo.• New trigger proposal includes xs,xe,mT cuts + topological cuts for different ET threshold

- L1_EM12_XS20 for e13_etcut_trkcut_X chains L1 default

- L1_EM15_XS30 for e18_etcut_trkcut_X chains

● Trigger rates reproduced in emulation: rates can be reduced with modest cost in efficiency

6

Efficiency measurement method

Efficiency for each identification algorithm extracted from the ratio of signal electrons εLevel=NLevel/Nprobe (Level=VL,L,M,T) as f(ET)

Example for HLT_xe70Cone30,20

7

Efficiency results example

Systematics of the efficiency measurements to be studied (various bckg template, triggers, subtraction methods, cone size, trigger matching effect, topology dependency etc.)

New trigger proposed for 2016 will collect much more statistics with better background conditions

VLLMT

8

Introduction to 4l channel● My analysis is based on a 4l final state → ttH to 4L affected by a

small branching ratio, but benefit from low background

ATLAS-CONF-2016-058Background composition

https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/CONFNOTES/ATLAS-CONF-2016-058/

9

Optimization studies (L = 10 fb-1)● Study options to reduce the

background (in particular ttbar) contribution:– Different Isolation WP

● Fixed Signal Region and Kinematics cuts optimization

– Different Event Topologies ● As a function of bjet multiplicity (and

algorithm)● As a function of lepton charge/flavor (SFOS)

The main reduction of the background with a small signal loss is provided by IsolationGradient (decided to be further used as the final isolation WP)

● In order to test the nominal cuts, we varied 5 variables:

● MET > {10,20,25,30,35,45,50,55}● |Mll – 91| > Zveto {-1, 10, 13, 15}● M4l> {100, 120, 150};● M4l< {300,350,400,450,500, 600, 10000}● Pt_lep_0 > {25, 30, 35}● Pt_lep_1 > {13, 15, 18, 20}

Kinematic cuts variations

Roofit significance ttH vs all BG

10/07/16 10

3l analysis approach to 4l: The Scaling Factor● Principles:

● 2 categories of fake source:

– Z-like: fake leptons (mostly) coming from light-jets– top-like: fake leptons (mostly) coming from b-jets

● The fundamental idea: select a control sample (CR) of events enriched in the background being estimated, and then use an extrapolation factor to relate these events to the background in the signal region (SR).

● In this talk explore the influence of a b-jet requirement on the Scaling factors determination

b l

V21 (L=22.1 fb-1) V20/04 (L=13.2 fb-1) V20/02 (L=13.2 fb-1) V19 (L=13.2 fb-1)

λbe 0.92 ± 0.08 ± 0.46 ± 0.47 1.00 ± 0.10 ± 0.50 ± 0.51 0.76 ± 0.08 ± 0.38 ± 0.39 0.99 ± 0.11 ± 0.50 ± 0.51

λle 1.18 ± 0.12 ± 0.59 ± 0.60 1.24 ± 0.14 ± 0.62 ± 0.63 1.11 ± 0.13 ± 0.56 ± 0.57 1.35 ± 0.17 ± 0.68 ± 0.70

λbμ 0.86 ± 0.12 ± 0.43 ± 0.45 1.10 ± 0.13 ± 0.55 ± 0.56 0.50 ± 0.10 ± 0.25 ± 0.27 1.01 ± 0.16 ± 0.50 ± 0.53

λlμ 0.71 ± 0.10 ± 0.36 ± 0.37 0.85 ± 0.11 ± 0.42 ± 0.44 0.56 ± 0.07 ± 0.28 ± 0.29 0.77 ± 0.12 ± 0.39 ± 0.40

Scaling Factor method is robust between different MC and data versions

10/07/16 11

4l Fake factor following the Run 1 experience

● Study the fake contributions in SR' and 3 CR enriched in tt and Z+jets:

– CR1: 2 tight leptons + 2 anti-tight electrons

– CR2: 2 tight leptons + 2 anti-tight muons

– CR3: 2 tight leptons + 1 anti-tight electron + 1 anti-tight muon

V20/04 (13.2 fb−1): 4T CR1 CR2 CR3

Data 12 9 5 11

prompt 17.60+-1.53 3.50+-0.62 0.81+-0.31 1.87+-0.47

ttbar-dilep 0.02+-0.02 3.95+-0.64 3.68+-0.68 8.66+-1.01

ttH_Herwig 0.69+-0.04 0.08+-0.01 0.10+-0.03 0.12+-0.03

Robust data-driven method, but statistics is poor so far; it will become feasible with more data

10/07/16 12

Conclusions● Electron identification efficiency measured using W

T&P● Tuning the analysis in 2015/2016, new triggers

implemented in 2016● In work: Performance publication including W T&P

analysis → for Moriond 2017

● Main analyst of ttH->4L for ICHEP2016● Signal selection optimization, fake leptons studies● In work: refined analysis using MVA, publication

with full stat by summer 2017

13

Backup Slides

14

Introduction to electron efficiency studies

The production cross-section times branching-ratio

correction factors to account for the geometrical and kinematicacceptance of the detector

the correction factors to account for the efficiency of the event beingreconstructed in the detector

• ε event

: efficiency of the signal events passing the event preselection cuts• α

reco accounts for the differences between applying the geometrical and kinematic

selection at generator or reconstruction level, i.e. the reconstruction efficiency;• ε

ID : efficiency of an electron passing the identification criteria

• ε trig is the efficiency of an event being triggered;• ε iso is the efficiency of possible isolation cuts.

Precise knowledge of the electron identification efficiency is an essential ingredient for many physics analyses

15

ATLAS experiment and detectorThe LHC is the world's largest and highestenergy particle accelerator. The collider is contained in a circular tunnel, with a circumference of 27 kilometers underground.

ATLAS: One of two general purpose detectors (CMS), used to look for signs of new physics, including the origins of mass and extra dimensions.

Display of a proton-proton collision event recorded by ATLAS on 3 June 2015, with the first LHC stable beams at a collision energy of 13 TeV.

ATLAS detector

https://www.youtube.com/watch?v=bbuQ3drSg9I

https://en.wikipedia.org/wiki/Particle_acceleratorhttps://www.youtube.com/watch?v=bbuQ3drSg9I

16

W T&P in 25ns data

Electron identification

menus

ET vs e- menu

● All e ID menus tested (cut based and likelihood)● Very large background at the probe level observed

ElectronID attempts to make optimal use of the tracking andcalorimetric detectors: shower widths, ratios of various energy deposits, tracking hits, track-cluster matching, etc.

Shower Shape Variables

Rhad E→ T (hadronic cal)/ET (EM cal)Rη Ratio In of cell energies in → η3x7 vs 7x7 celsRΦ Ratio In of cell energies in → Φ3X3/7X7Wstos Total shower width.→TQ (Track quality): Number of hits in the pixel detector 1 && ≥ Number of hits in the pixels and SCT 7≥

backup

17

Δφ distributions for Jets pT>15

ΔΦe-jet

ΔΦjet-MET

MC

MC

Final cuts applied on angular variables:• ΔΦejet0.7

Data

Data

Data

Data

All other kinematic distributions in backup

18

Δφ (METJet eJet) correlation between for Jets with pT>10

Δφ (eJet eMET) correlation between for Jets with pT>10

e18a Most of the statistics at trigger level is in background region. Further cuts on angular variable to reduce the background won't affect the signal (as seen in Tight = Signal Like distribution)

Δφ (METJet) Δφ (METJet)

Δφ

(eJ

et)

Δφ

(eJ

et)

Δφ

(e-M

ET)

Δφ

(e-M

ETt)

Δφ (eJet) Δφ (eJet)

PROBE

PROBE

TIGHT (SL)

TIGHT (SL)

19

Trigger prescales per runs

DAOD EGAM 5 tested: full luminosity used

20

PreScaled Et distributions: Event+Probe Level

Et distribution per trigger bit:✔ 3 groups observed

✔ Different Et behavior

Loose Medium

Tight

High MET

High Et

21

Δφ (METJet eJet) correlation between for Jets with pT>10

Δφ (eJet eMET) correlation between for Jets with pT>10

e13

22

backup

23

backup

10/07/16 24

Trigger Matching Procedure● TM_dR=0.07 ● dR = sqrt(ΔΦ2 + Δη2);

– (probe electron, electron from the HLT_xAOD__ElectronContainer_egamma_Electrons)

● Cuts: Ele.pT/1000. > 13 && |η| < 2.5 && el_nPixHits > 0 && el_nSCTHits > 7

● if(max(dR)

10/07/16 25

Triggers tested

backup

10/07/16 26

W MET triggers Ratio efficiencies CB_M/failVL LH for without (fill) / with (dotted) Trigger Matching, e13 cut, dR=0.07

0.0

10/07/16 27

W e13 triggers Ratio efficiencies CB_M/failVL LH for without (fill) / with (dotted) Trigger Matching , e13 cut, dR=0.07

0.0

28

Top quark polarization and spin correlation

The scope of my PhD thesis is to explore the potential of the spin-correlation properties in the associated Higgs top-pair production at the LHC as a possible tool to improve the separation of the signal from the ttH irreducible backgrounds.

➢ spin correlations could help in disentangling the SM scalar component from a pseudoscalar contribution in the top-Higgs coupling

➢ by reconstructing the individual top systems, the top-quark spin properties can be accessed by measuring angular distributions of the final decay products

➔ Observables in the laboratory frame and in different top quark spin quantization bases are explored, used to measure the coefficient fSM , which is related to the number of events where the t and t spins are correlated as predicted by the SM,

● The measured value of fSM is translated into the spin correlation strength A , which is a measure for the number of events where the top quark and top antiquark spins are parallel minus the number of events where they are antiparallel with respect to a spin quantization axis, divided by the total number of events:

The strength of the spin correlation is:

The factor αi is the spin-analyzing power, which must be between -1 and 1.

As shown in JHEP07(2014)020, spin-correlation features in the ttH production are quite promising for enhancing the signal sensitivity over the irreducible background.

Similar studies will be performed on 2015 data.

WHAT MAKES TOP QUARK INTERESTING?

29

11 Heaviest fundamental particle in the Standard ModelLarger mass ➙ Larger coupling to SM Higgs +

mtop is a fundamental parameter in SM

Allows for Self-Consistency Checks of SM Post Higgs Discovery

44 Hints of new BSM/physics?

Exotic Particles Could Decay Preferentially to Top Quarks

33 Short Lifetime(~10-25 s) Possible reconstruct quarks before

hadronization – Unique among the quarks!

Access to Polarization and Spin Correlations

22 Processes including top are backgrounds for new physics

+ Exotics and SUSY

Good Understanding ➙ Improvements in Searches

PhD topic : ttH spin correlations at LHC

The goal of my PhD thesis is to explore the potential of the spin-correlation properties in the associated Higgs top-pair production at the LHC as a possible tool to improve the separation of the signal from the ttH irreducible backgrounds.

As shown in JHEP07(2014)020, spin-correlation features in the ttH production are quite promising for enhancing the signal sensitivity over the irreducible background. Similar studies will be performed on 2015 data.

➢ spin correlations could help in disentangling the SM scalar component from a pseudoscalar contribution in the top-Higgs coupling

➢ by reconstructing the individual top systems, the top-quark spin properties can be accessed by measuring angular distributions of the final decay products

→ in the chiral limit of vanishing top-quark mass (mtt ≫ mt) t and t spins are highly correlated and parallel to each other along the tt production axis. Top pairs are hence produced in the LR + RL helicity configurations.

→ when the tt is produced in association with a Higgs boson, the top quark and antiquark helicities are also correlated, but Higgs-boson emission from the top-quark final states via Yukawa interactions induces a chirality flip in the top-quark polarization →LL+RR helicity configuration, (LR+RL configuration suppressed by terms of order O(m2t /m2tt).

JHEP07(2014)020

tt tth

t

t

31

Top quark physics

Decay products of the top quark: correlated with its spin the direct coupling of top quark with → W+ opens the window for a direct measurement of the CKM element |Vtb| + one can also predict the Standard Model (SM) allowed helicity states of the produced W+ and check experimentally beyond the SM predictions (new top quark supersymmetric decay modes can be tested, as well as other aspects of the SM limits)

Statistically limited

tth Run I

Run1 Run2

Luminosity increase by a factor 4

Cross section increase by a factor 4

• ttH is a major goal at LHC in the next period, new properties can be studied with higher statistics

32

Introduction to 4l channel● My analysis is based on a 4l final state (small but clear signal) where

the top quark pair decay leptonically or semi-leptonically, and that the Higgs decay either to WW, tt, and ZZ.

➢ SR (baseline) definition:

➢ 4 leptons, pT>10 GeV, total charge=0, at least one of 4 leptons to be trigger matched➢ Z veto : |Mll-91.2|>10 GeV + di-lepton resonances veto : Mll >12 GeV ➢ at least 2 jets , at least 1 b-jet➢ 100

33

Test a custom cut on isolation using cone 0.2

SR Nominal entries + MET>10 GeV + topo20ET/pt 10 GeV + topo20ET/pt 10 GeV + topo20ET/pt 10 GeV + topo20ET/pt 35Zveto_13

10020

MET>35Zveto_13

10020

MET>10Zveto_13

10013

MET>10Zveto_13

10013

10/07/16 34

SG optimization+IsolationBaseline Selection fixed Optimized SR +isoGrad

ttHSG 0.954+-0.114 0.884+-0.110 0.713+-0.092

ttZ 1.165+-0.031 0.821+-0.026 0.602+-0.022

ZZ 0.070+-0.024 0.029+-0.013 0.028+-0.013

ttbar 0.828+-0.436 0.828+-0.436 0+-0

ttbardilep 0.6556+-0.182 0.444+-0.148 0+-0

BG_MC 2.964+-0.47 2.312+-0.464 0.732+-0.045

Signif 0.220 0.238 0.372

Zveto_1310020

Zveto_1310013

Backup

35

Signal and Bakcground samples in optimized SR

36

Significance in optimized SR

37

IsolationGradient per flavor Baseline Selection isolationGradient

ttHSG 0.0418893+-0.0161731

ttbar 0+-0

ttbardilep 0+-0

BG_MC 0.10992+-0.0162984

ttHSG 0.187774+-0.0390074 0.139367+-0.0327263

ttbar 0.592545+-0.367539 0+-0

ttbardilep 0.127163+-0.0816679 0+-0

BG_MC 1.07942+-0.377138 0.286083+-0.0183444

ttHSG 0.28092+-0.0471114 0.238117+-0.0412026

ttbar 0+-0 0+-0

ttbardilep 0.248111+-0.111898 0+-0

BG_MC 0.71041+-0.121675 0.373851+-0.0445861

ttHSG 0.328861+-0.0782478 0.304276+-0.0767198

ttbar 0.235964+-0.235964 0+-0

ttbardilep 0.280338+-0.11849 0+-0

BG_MC 0.903777+-0.265337 0.321169+-0.0248878

ttHSG 0.115098+-0.0539284

ttbar 0+-0

ttbardilep 0+-0

BG_MC 0.160604+-0.0201292

eeee

eeem

eemm

emmm

mmmm

Most of the ttbar events → in eeem/eemm/emmm pairsOption → apply isolation gradient only in the case of Ne=1,2,3

isolationGradient IsolationGradient for eeem/

eemm/emmm

ttHSG 0.781+-0.096 0.838+-0.108

ttbar 0+-0 0+-0

ttbardilep 0+-0 0+-0

BG_MC 1.211+-0.059 1.251+-0.060

Significance (RooFit)

0.31438 0.350

38

4l type/SFOS pairs and jet multiplicity

L = 3.2 fb-1

Attempt to adjust isolation conditions as a function of the events topology/content

4l type

SFOS pairs

10/07/16 39

4l Mass

10/07/16 40

Optimization only using M4l min/max cutBaseline Selection

fixedSR opt using min

M4lSR opt using M4l SR opt using M4l + isoGrad

ttHSG 0.954+-0.114 0.951+-0.114 0.939+-0.112 0.760+-0.094

ttZ 1.165+-0.031 1.163+-0.031 1.019+-0.029 0.736+-0.024

ZZ 0.070+-0.024 0.070+-0.024 0.070+-0.024 0.062+-0.022

ttbar 0.828+-0.436 0.828+-0.436 0.828+-0.436 0+-0

ttbardilep 0.655+-0.182 0.602+-0.174 0.602+-0.174 0+-0

BG_MC 2.964+-0.47 2.908+-0.474 2.722+-0.473 0.907+-0.050

Signif 0.220 0.223 0.232 0.364

Zveto_1010015

Zveto_1010015

Zveto_1010015

Zveto_1010015

10/07/16 41

SG optimization+Isolation, without MET cutBaseline Selection fixed No MET added +isoGrad

ttHSG 0.954+-0.114 0.884+-0.110 0.713+-0.092

ttZ 1.165+-0.031 0.821+-0.026 0.602+-0.022

ZZ 0.070+-0.024 0.029+-0.013 0.028+-0.013

ttbar 0.828+-0.436 0.828+-0.436 0+-0

ttbardilep 0.6556+-0.182 0.444+-0.148 0+-0

BG_MC 2.964+-0.47 2.312+-0.464 0.732+-0.045

Signif 0.220 0.238 0.372

Zveto_1310020

Zveto_1310013

42

Results for nbjet=1 or SFOS option

Small signal loss by applying isolation cuts on a ttbar enrich topology cuts, efficient further background rejection with MET>10 GeV (for non ttbar samples)

Fixed SR SR + IsoGrad=4 for nbjet=1 MV2c20_70

SR + IsoGrad=4 for nbjet=1 MV2c20_77

SR + IsoGrad=4 for (SFOS=2 || SFOS=0)

ttH 0.829+-0.099 0.852+-0.100 0.838+-0.108

BG_ALL 1.348+-0.071 1.421+-0.088 1.251+-0.060

ttbar 0+-0 0+-0 0+-0

Significance 0.325 0.329 0.350

Optimized SR SR Nominal reopt MET SR + IsoGrad=4 for nbjet=1 reopt using MET

SR + IsoGrad=4 for (SFOS=2 || SFOS=0) reopt

using MET

ttH 0.943+-0.114 0.800+-0.097 0.792+-0.106

BG_ALL 2.794+-0.469 1.112+-0.066 1.011+-0.055

ttbar 0.828+-0.436 0+-0 0+-0

Significance 0.228 0.349 0.365MET>10Zveto_1

10020

MET>10Zveto_13

10013

MET>35Zveto_13

10013

L = 10 fb-1

10/07/16 43

Impact of the mH veto on the scaled distributions (with no bjet requirement scale factors) FF no errorSample (with isoGradient)

With no FF applied With FF applied FF-noFF yields Entries

ttW 0.03+-0.01 0.03+-0.01 0.01+-0.00 22

ttZ 0.95+-0.03 0.98+-0.03 0.04+-0.01 1652

Zgam 0.03+-0.03 0.12+-0.12 0.09+-0.09 1

ttH Signal 0.74+-0.10 0.75±0.10 0.01+-0.00 2002

Sample (without isoGradient)

With no FF applied With FF applied FF-noFF yields Entries

ttW 0.07+-0.01 0.08+-0.01 0.01+-0.00 52

ttZ 1.14+-0.03 1.19+-0.03 0.06+-0.01 2007

Zgam 0.03+-0.03 0.12+-0.12 0.09+-0.09 1

ttbar 0.83+-0.44 1.30+-0.73 0.48+-0.33 6

ttbardilep 0.60+-0.18 0.69+-0.20 0.09+-0.03 17

ttH Signal 0.92+-0.11 0.92+-0.11 0.01+-0.01 2634

The table with no isolation gradient is given in order to observe the impact of the scaling factors λ on ttbar (in the nominal selection this contribution is zero)

10/07/16 44

PT dependence of the fakes in the yields, FF no error

Sample (with isoGradient)

With no FF applied With FF nominal applied

With FF fakes pT>12 GeV



ttW 0.03+-0.01 0.03+-0.01 0.04+-0.01 0.04+-0.01 0.04+-0.01

ttZ 0.95+-0.03 0.98+-0.03 0.99+-0.03 1.00+-0.03 1.02+-0.04

Zgam 0.03+-0.03 0.12+-0.12 0.11+-0.11 0.16+-0.16 0.16+-0.16

ttH Signal 0.74+-0.10 0.75±0.10 0.75+-0.10 0.75+-0.10 0.76+-0.10

Sample (without isoGradient)

With no FF applied

With FF nominal applied




ttW 0.07+-0.01 0.08+-0.01 0.08+-0.02 0.08+-0.02 0.09+-0.02

ttZ 1.14+-0.03 1.19+-0.03 1.19+-0.03 1.21+-0.04 1.23+-0.04

Zgam 0.03+-0.03 0.12+-0.12 0.11+-0.11 0.16+-0.16 0.16+-0.16

ttbar 0.83+-0.44 1.30+-0.73 2.00+-1.22 2.39+-1.52 3.86+-2.74

ttbardilep 0.60+-0.18 0.69+-0.20 0.75+-0.22 0.77+-0.22 0.76+-0.23

ttH Signal 0.92+-0.11 0.92+-0.11 0.93+-0.11 0.93+-0.11 0.95+-0.12

Impact on the yields by applying different requirement on the fake pT (when calculation the scale factors) is negligible

10/07/16 45

noFF NOMINAL A B C D0.07+-0.01 0.08+-0.01 0.10+-0.03 0.11+-0.03 0.07+-0.01 0.07+-0.011.14+-0.03 1.19+-0.03 1.22+-0.04 1.21+-0.04 1.19+-0.03 1.17+-0.030.03+-0.03 0.12+-0.12 0.14+-0.14 0.04+-0.04 0.14+-0.14 0.05+-0.050.83+-0.44 1.30+-0.73 2.28+-1.44 2.28+-1.44 1.00+-0.54 1.00+-0.540.60+-0.18 0.69+-0.20 0.91+-0.27 0.95+-0.29 0.56+-0.16 0.57+-0.170.92+-0.11 0.93+-0.11 0.92+-0.12 0.92+-0.12 0.92+-0.11 0.93+-0.11

Central values + statistical error

NOM+ A+ B+ C+ D+ NOM- A- B- C- D-

0 0.032 0.032 0 0 0.014 0.02 0.022 0 0

0.022 0.048 0.062 0.03 0.064 0.024 0.024 0.042 0.035 0.033

0.05 0.07 0.05 0.07 0.05 0.04 0.05 0.02 0.05 0.03

0.55 2.1 2.1 0.51 0.51 0.41 1.2 1.2 0.36 0.36

0.063 0.16 0.15 0.054 0.045 0.054 0.16 0.15 0.054 0.054

0.014 0.033 0.024 0.014 0.014 0.01 0.022 0.014 0.01 0.014

Systematics UP Systematics DOWN

Nominal = No bjet req A = # bjet=0 for Zjets&tt B = # bjet>0 for Zjets && # bjet=0 for tt C = # bjet=0 for Zjets && # bjet>0 for tt D = # bjet>0 for Zjets && # bjet>0 for tt

Statistical and systematic errors

10/07/16 46

noFF NOMINAL A B C D0.07+-0.01 0.08+-0.01 0.10+-0.03 0.11+-0.03 0.07+-0.01 0.07+-0.011.14+-0.03 1.19+-0.03 1.22+-0.04 1.21+-0.04 1.19+-0.03 1.17+-0.030.03+-0.03 0.12+-0.12 0.14+-0.14 0.04+-0.04 0.14+-0.14 0.05+-0.050.83+-0.44 1.30+-0.73 2.28+-1.44 2.28+-1.44 1.00+-0.54 1.00+-0.540.60+-0.18 0.69+-0.20 0.91+-0.27 0.95+-0.29 0.56+-0.16 0.57+-0.170.92+-0.11 0.93+-0.11 0.92+-0.12 0.92+-0.12 0.92+-0.11 0.93+-0.11

Central values + statistical error

NOM A B C D

0.007 0.027 0.027 0 0

0.023 0.045 0.052 0.0325 0.0485

0.045 0.045 0.035 0.06 0.04

0.48 1.65 1.65 0.435 0.435

0.0585 0.155 0.15 0.054 0.0495

0.012 0.0235 0.019 0.012 0.014

0.5*(Systematics UP + Systematics DOWN)

Nominal = No bjet req A = # bjet=0 for Zjets&tt B = # bjet>0 for Zjets && # bjet=0 for tt C = # bjet=0 for Zjets && # bjet>0 for tt D = # bjet>0 for Zjets && # bjet>0 for tt

Statistical and systematic errors

10/07/16 47

Cut Flow in 3L CR 22.1 fb-1 ● CR Z+jets

v19 Data ttbar-dilep Z+jets others

Fakes_e 1381 11.999±1.050 829.724±101.738 253.114±9.888

Fakes_mu 682 8.076±0.875 594.908±86.190 215.397±10.457


Fakes_e 1956 19.7±1.5 1294.3±131.0 411.1±13.6

Fakes_mu 976 13.1±1.2 866.5±113.7 348.7±14.2

Others = SingleTop & Dibosons & NLO+ttV & tribosons & tZ_bis & tWZ & ttH_Herwig & VH & tttt_ttWW & W+jetsSherpaNNPDF30NNLO & tHbjMadGraph+Herwig

V20/02 Data ttbar-dilep Z+jets others

Fakes_e 1389 17.187±1.299 1009.536±110.696 255.268±9.903

Fakes_mu 686 13.010±1.167 826.339±95.970 215.863±10.376


Fakes_e 1525 17.187±1.299 1009.536±110.696 259.657±10.338

Fakes_mu 952 13.010±1.167 826.339±95.970 236.567±11.383

Backup

10/07/16 48

Cut Flow in 3L CR 22.1 fb-1


Fakes_e 136 116.131±3.415 2.733±1.646 17.307±1.332

Fakes_mu 69 55.518±2.322 0.167±0.177 13.038±2.878

Others = SingleTop & Dibosons & NLO+ttV & tribosons & tZ_bis & tWZ & ttH_Herwig & VH & tttt_ttWW & W+jetsSherpaNNPDF30NNLO & tHbjMadGraph+Herwig


Fakes_e 137 149.319±3.881 3.078±1.685 19.731±1.445

Fakes_mu 70 97.739±3.126 0.430±0.366 20.828±3.877


Fakes_e 202 184.2±4.6 3.24±2.16 29.1±2.0

Fakes_mu 96 89.4±3.2 -0.273±0.289 19.3±4.4


Fakes_e 180 149.319±3.881 3.078±1.685 26.850±5.611

Fakes_mu 130 97.739±3.126 0.430±0.366 22.378±3.899

● CR ttBar

Backup

10/07/16 49

FF using 3L information: Scale Factors in v19/20/21 dependency on b-jets multiplicity nJets_OR_MV2c10_70

10/07/16 50

FF using 3L information: Scale Factors in v19/20/21 dependency on PT fake

10/07/16 51

Lambda propagation models• M1

– An event with at least a fake lepton is a fake event– FSFs are applied according to the nature of the fake event

(“light” or “heavy” environment) and the flavour of the leptons

– Events with N expected fakes will be reweighted by• M2:

– Method 1 applied only if is NotIsolated type lepton– Factorize if more than 1 fake lepton

• M3: use lep_isFake to separate fakes FSF applied the same as in Method 1

• Option: apply lambda deduced at high b-jet multiplicity

Backup

10/07/16 52

Summary on the 4l fake estimation v19ttH(herwg) M1 M2 M3Total_Brut 0.654 0.654 0.654

Prompt 0.568 0.568 0.579Fakes 0.09 0.09 0.07

Fakes_Scaled 0.09 0.08 0.07Total_Corrected 0.654 0.649 0.649

Fakes_Scaled bjet>0 0.1 0.15 0.13Total_Corrected bjet>0 0.670 0.719 0.710

TOTAL_BG M1 M2 M3Total_Brut 1.932 1.932 1.932

Prompt 1.654 1.654 1.689Fakes 0.28 0.28 0.24



The fake rates with various methods vary within 13-14%

Backup

10/07/16 53

Summary on the 4l fake estimation v20/02ttH(herwg) M1 M2 M3

Total_Brut 0.689 0.689 0.689

Prompt 0.602 0.602 0.613

Fakes 0.09 0.09 0.08

Fakes_Scaled 0.05 0.05 0.05

Total_Corrected 0.654 0.655 0.663

Fakes_Scaled bjet>0 0.06 0.08 0.07

Total_Corrected bjet>0 0.664 0.680 0.682

TOTAL_BG

Total_Brut 1.633 1.633 1.633

Prompt 1.338 1.338 1.384

Fakes 0.3 0.3 0.25

Fakes_Scaled 0.33 0.31 0.27

Total_Corrected 1.673 1.645 1.650

Fakes_Scaled bjet>0 0.34 0.24 0.2

Total_Corrected bjet>0 1.591 1.497 1.487

The fake rates with various methods vary within 15%

Backup

10/07/16 54

Summary on the 4l fake estimation v20/04ttH(herwg) M1 M2 M3Total_Brut 0.764 0.764 0.764

Prompt 0.660 0.660 0.686Fakes 0.1 0.1 0.08



TOTAL_BGTotal_Brut 1.504 1.504 1.504

Prompt 1.255 1.255 1.284Fakes 0.25 0.25 0.22



The fake rates with various methods vary within 15%

Backup

10/07/16 55

Summary on the 4l fake estimation v21ttH(herwg) M1 M2 M3Total_Brut 1.031 1.031 1.031

Prompt 0.908 0.908 0.927Fakes 0.12 0.12 0.1



TOTAL_BG M1 M2 M3Total_Brut 2.197 2.197 2.197

Prompt 1.834 1.834 1.870Fakes 0.36 0.36 0.33



The fake rates with various methods vary within 13-14%

Backup

10/07/16 56

4l Yields in the 4T and CR v214T CR1 CR2 CR3

Data 19 7 5 7prompt 27.15+-2.05 4.80+-0.82 0.80+-0.33 2.04+-0.59

ttbar-dilep 0.05+-0.05 3.28+-0.61 1.50+-0.40 5.75+-0.87ttH_Herwig 1.09+-0.06 0.10+-0.02 0.05+-0.02 0.08+-0.03

4T CR1 CR2 CR3Data 12 9 5 11

prompt 17.60+-1.53 3.50+-0.62 0.81+-0.31 1.87+-0.47ttbar-dilep 0.02+-0.02 3.95+-0.64 3.68+-0.68 8.66+-1.01ttH_Herwig 0.69+-0.04 0.08+-0.01 0.10+-0.03 0.12+-0.03

4l Yields in the 4T and CR v20/04 Compatible with Run1

Run1 results (20.3 fb−1):

prompt = tZ_bis + tttt_ttWW + W+gamma + tribosons + tWZ + VH + tHbjMadGraph+Herwig + W+jetsSherpaNNPDF30NNLO + NLO+ttV + Dibosons

10/07/16 57

Conclusions I● Electron identification efficiency has been tested using W T&P● Angular distributions illustrate background features → used to reject

bkgrd events

● Isolation studies in ET and η binning for different background templates (LH+CutBased) have been performed → systematics

● Trigger configuration has been proposed for 2016 data taking● Increased statistics in 2016-2017 will allow for a precise differential

measurement, competitive with Z

● Future prospects for W T&P analysis:

– Test the new trigger configuration– Efficiency studies including systematics to be concluded– Internal note illustrating the W T&P analysis → for Moriond 2017– Dedicated talk in the ATLAS EGAM workshop (7-10 November 2016)

10/07/16 58

Conclusions II● Isolation requirement reduced the ttbar and other MC Background sample contributions:

– Isolation variables have been tested for all leptons and in different topologies (SFOS=0/2 and region where nbjet=1)

– Cut variation studied + b-tagging tested

– The isolation conditions are relatively similar; Proposal: IsolationGradient WP

● Systematics of the scaling factors → dependence on the b-jet requirement and fake lepton pT

● Estimation of the fakes provided by using the scaling factor method and Run 1

● Future plans:

– MVA techniques to be tested in order to increase the significance– Improve the fake estimation in the new data sets : scaling factor method + fake factor

by using the Run 1 definitions

● ICHEP 2016 CONF note

– Plans: paper for EPS 2017

Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Slide 24Slide 25Slide 26Slide 27Slide 28Slide 29ttH measurements at LHCSlide 31Slide 32Slide 33Slide 34Slide 35Slide 36Slide 37Slide 38Slide 39Slide 40Slide 41Slide 42Slide 43Slide 44Slide 45Slide 46Slide 47Slide 48Slide 49Slide 50Slide 51Slide 52Slide 53Slide 54Slide 55Slide 56Slide 57Slide 58

Documents

CPPM PhD days Measurement of electron efficiency using W → … · 2 The Tag & Probe Method - The tag object: strict requirements to enhance the purity of the sample - The other