Search for heavy neutral resonances in vectorboson fusion in pp collisions at 𝒔𝒔 = 13 TeV

with the ATLAS detector at the LHC

October 27, 2017

Guangyi Zhang1,2

1University of Science and Technology of China2Institute of Physics, Academia Sinica

PhD thesis defense

Supervisor: Liang Han1, Suen Hou2

2/48

Outline

Search for new resonances (R) in qq → Rqq → ℓ+𝜐𝜐ℓ-�̅�𝜐qq (ℓ = e, µ)

using 3.2 fb-1 data collected in 2015

Search for new resonances (X) in qq → Xqq → WWqq → e𝜐𝜐µ𝜐𝜐qq

using 36.1 fb-1 data collected in full 2015 and 2016

Introduction

Physics analyses

Summary

Standard model

Vector boson fusion

LHC and ATLAS detector

3/48

Introduction

Electromagnetic force Weak force

Strong force Gravitation (not in the SM)4/48

Standard Model Standard Model (SM)• Most successful and well-tested theory

• Elementary particles: leptons, quraks (spin-½ fermions)

=> constituents of matter gauge bosons (W/Z, γ, g), spin-1

=> force mediators Higgs boson, spin-0

=> origin of mass

describing the elementary particles and their interactions

• Fundamental forces: electromagnetic, weak, strong forces gravitation not described in the SM

• Origin of mass: Higgs Mechanism, EWSB particles acquire mass via interactions

with Higgs field (υ ≠ 0)

nb

JHEP: 0811.010 (2008)

5/48

Vector boson fusion In the SM, vector boson fusion (VBF) is an important class of processes

• provide a unique means to directly examine the EWSB mechanism

W+W- scattering/fusion without a SM Higgs

W+W- scattering/fusion with a SM Higgs

nb

JHEP: 0811.010 (2008)

unitarity violated

unitarity restored

“u” Mandelstam variable

+

6/48

Vector boson fusion VBF is very sensitive to phenomena beyond the SM

?

• Unknown issues about Higgs boson discovered at the LHC:

fully or only partially unitarizes the VBF amplitude ?

is the coupling H→VV exactly the one that SM predicted ?

(low measurement precision ~20 %)

• New resonance needed:

Higgs partially unitarizes the VBF amplitude => new resonances

Any non-SM HVV coupling => new physics

If new resonance has no/weak coupling to fermions

=> VBF is a leading search channel

7/48

Vector boson fusion Characteristics of VBF

• Relatively small production cross sections

=> Experimental studies of VBF are feasible only at the LHC so far

• Fully-leptonic channels have lower SM background

=> Considered in the analyses

• VBF event topology at the LHC

two charged leptons (ee, μμ, eμ), 𝐸𝐸𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚, two forward jets (tagging jets)

VBF signal characteristics:

Large mjj (>500 GeV)

Large Δηjj (>2.4)

8/48

Large Hadron Collider (LHC)

proton-proton (pp) collisions• World’s largest and most powerful particle accelerator

• Purpose: Higgs boson, new physics (eg. dark matter ), etc.

• Two separate rings: 26.7 km, 45-175m underground

• Four major detectors: ATLAS, CMS, ALICE and LHCb

• pp collisions: 2835×2835 bunches, bunch spacing

25 ns (7.5 m), 1011 protons/bunch

• Designed luminosity and center-of-mass energy:

ℒ = 1034𝑐𝑐𝑐𝑐−2𝑠𝑠−1, 𝑠𝑠 =13 TeV (2015-2018), 14 TeV (2021-2037)

9/48

ATLAS detector

44 m

25 m

7000 t

ATLAS detector: a general purpose detector at the LHC

Muon spectrometer (|η|<2.7)-precise tracking and triggering on muons-precision-tracking systems: MDT, CSC-trigger systems: RPC, TGC

Inner detector (|η|<2.5):- precise measurement of trackingand vertices

- Pixel, SCT, TRT

EM calorimeter (|η|<3.2):-e/γ measurement-LAr-Pb accordion

Magnet system:-solenoid magnet (barrel), 2 T-toroid magnets, 0.5 T (barrel), 1 T (end-cap)

HAD calorimeter (|η|<4.9):-hadronic measurements-scintillating tiles-steel (central), LAr-Cu/tungsten (forward)

10/48

Search for new resonances (R) inqq → Rqq → ℓ+𝜐𝜐ℓ-�𝝊𝝊qq (ℓ = e, µ)using 3.2 fb-1 data collected in 2015

ATLAS conference note: ATLAS-CONF-2016-053

Proceeding paper: EPJ Web of Conferences 137, 08016 (2017)

My contributions: All physics analysis work

11/48

Introduction Analysis goal:

• Search for new resonances (R) in VBF qq → Rqq → ℓ+𝜐𝜐ℓ-�̅�𝜐qq (ℓ = e, µ)three decay channels: ee, μμ, eμ

Signal model:

Γ0=g2m3/64πυ2

σ: scalar isoscalarφ: scalar isotensorρ: vector isovectorf: tensor isoscalart: tensor isotensor

• Benchmark model: EW chiral Lagrangian (EWChL) with K-matrix unitarization• New resonances: only couples to vector boson, thus mainly produced via VBF• Free parameters: coupling RVV (g=2.5) & mass [200, 500] GeV

scalar

vector

tensor

12/48

Signal Signal definition:

Signal sample

• New resonance

• SM EW qq→ℓ+𝑣𝑣ℓ-�̅�𝑣qq

• Interference

SM continuumsample

• SM EW qq→ℓ+𝑣𝑣ℓ-�̅�𝑣qq

= Signal (New resonance + interference)

Signal Xsec vs. resonance mass

-

Using Whizard+Pythia8 to generate both samples

SM EW continuum

Signal samples

13/48

Data/MC samples Data samples:

• 25 ns data in 2015, Luminosity = 3.2 fb−1

MC corrections:

• Lepton energy/momemtum scale/resolution• Lepton Reco/ID/Iso/Trig effSF• Jet energy scale/resolution, b−tag effSF• Pile-up reweighting

MC samples:

• 𝑡𝑡 ̅𝑡𝑡: Powheg• Wt: Powheg• Z+jets: MadGraph (QCD) and Sherpa (EW)• diboson: Sherpa (QCD) and Whizard (EW)• Zγ: Sherpa • ttV: MadGraph• SM Higgs: Powheg (ggH and VBF)

14/48

Event selection Event selections for signal region (SR):

15/48

Background estimation

• W+jets• QCD

• Z+jets

• 𝑡𝑡 ̅𝑡𝑡• Wt• ttV• Zγ+jets• diboson (WW/WZ/ZZ)• SM Higgs (ggH, VBF)Backgrouds of

ll'+ETmiss+2jets

SM processes that can produce events with

two OS leptons

SM processes that haveone or two leptons from

jets (faked)

MC prediction

Data driven(Matrix method)

Strategy of background estimation:

Validation regions (VRs) :

16/48

Dominant background sources :• For ee/μμ channel: Z+jets & 𝑡𝑡 ̅𝑡𝑡• For eμ channel: 𝑡𝑡 ̅𝑡𝑡

Background estimation

• Selection criteria listed on slide 14 is assumed unless otherwise specified

17/48

Z pT reweighting for the Z+jets prediction:

• Some discrepancy is found between data & MC for Z pT distribution • Reweighing function is derived by using a polynomial fit for the spectrum

(Data-Non-Z+jets) /Z+jets• Cut 1-9 in slide 14, |mℓℓ-mZ| < 25 GeV• This reweighting function used in both VRs and SR

Reweighting fit function Before reweighting After reweighting

Data vs. prediction in Z+jets VR

18/48

Data vs. prediction in Z+jets VR Z+jets VR: |mℓℓ-mZ| < 25 GeV, no mjj cut

Reasonable agreement of data and the SM prediction observed in Z+jets VR

MTWW

Njets

19/48

Data vs. prediction in 𝒕𝒕�̅�𝒕 VR 𝑡𝑡 ̅𝑡𝑡 VR: Nb-jets > 1, no mjj cut

Good agreement of data and the SM prediction observed in 𝑡𝑡 ̅𝑡𝑡 VR

MTWW

Nb-jets

20/48

Fake background estimation (matrix method)Matrix Method: a data-driven method to estimate fraction of jets misidentified as leptons.

Relation for the event numbers in the different subsamples:

• r1, r2 (f1, f2): the real (fake) rates evaluated for leading(1) and sub-leading(2) leptons

Fake contribution estimation:

• Measure the real rate of electron: tag-probe method is used to extract it from di-el data sample

• Measure the fake rate of electron: obtained from a fake enriched data sample, MC subtraction.

• Measure NTL, NLT, NTT, NLL, then invert the matrix to get the fake contribution--NRF, NFR, NFF

• The matrix method is applied event by event

This matrix method depends on two parameters:

• Real rate: probability for a real lepton identified as a loose lepton to pass tight lepton selection;

• Fake rate: probability for a real jet identified as a loose lepton to pass tight lepton selection.

21/48

Fake background estimation (matrix method) Real rate of electron vs. el_pt, el_η:

Real rate

Fake rate

Fake rate of electron vs. el_pt, el_η:pT η

pT η

22/48

Data vs. prediction in low-mjj VR

Reasonable agreement of data and the SM prediction observed in low-mjj VR

low-mjj VR: mjj < 500 GeV, validate the overall background estimation

MTWW

mjj

23/48

Data vs. prediction in the signal region Signal region (SR):

• Based on all selections on slide 14

No significant data excess above the SM background prediction is observed in SR

24/48

Data vs. prediction in the signal regionee-channel µµ-channel

eµ-channel

-- Due to two neutrinos in the final state,

𝑀𝑀𝑇𝑇𝑊𝑊𝑊𝑊 is a useful discriminating variable:

-- No significant excess beyond the SM

background predication is found

MTWW

25/48

Systematic uncertainties Experimental uncertainties(%) on the backgrounds in the signal region:

Theoretical uncertainties on the production Xsec of the backgrounds

Additional shape systematic uncertainties for two dominant backgrounds

(Z+jets, 𝑡𝑡 ̅𝑡𝑡) are included.

Experimental uncertainties on signal considered (JES/JER, b-tagging, 𝐸𝐸𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚etc.)

26/48

95% CL upper limits

σ

No significant excess above the SM background expectation is observed.

95% CL upper limits are derived on σ×Br for new resonances (σ, φ, ρ, f and t)

Number counting as inputs to set limit due to limited signal statistics

The frequentist method (CLs), is used to compute 95% CL upper limits

ρ

φ

27/48

95% CL upper limits f

t Observed 95% CL exclusion limitsin 200-500 GeV mass region:

380 − 220 fb (σ particle)460 − 240 fb (φ particle)330 − 270 fb (ρ particle)340 − 260 fb (f particle)310 − 260 fb (t particle)

mR < 230 (300) GeV for ρ (f) excluded

28/48

Summary for analysis 1

• SM backgrounds are carefully studied using MC simulation and

data-driven (matrix method) and validated in several VRs.

• all the possible uncertainties are well evaluated

• no significant data excess above the SM prediction is observed

• First 95% CL upper limits derived, mR < 230 (300) GeV for ρ (f) resonance excluded

Search for new resonances (R) in qq → Rqq → ℓ+𝜐𝜐ℓ-�̅�𝜐qq (ℓ = e, µ) using 3.2 fb-1 data

Need to update:

• Published on:

EPJ Web of Conferences 137, 08016 (2017)

ATLAS-CONF-2016-053

• Small data set (3.2 fb-1)

• Signal samples – low statistics

• Three decay channels (ee, μμ, eμ) studied, but the most sensitive one is eμ

29/48

Search for new resonances (X) inqq → Xqq → WWqq → e𝜐𝜐µ𝜐𝜐qqusing 36.1 fb-1 data (2015+2016)

Paper: submitted to Eur. Phys. J. C, arXiv:1710.01123 [hep-ex]

CERN Preprint URL: CERN-EP-2017-214

My contributions: signal Monte Carlo samples generation, signal

acceptance and predictions, event selection optimization, systematic

uncertainties, etc.

30/48

Introduction Analysis goal:

• Search for new resonances (X) in VBF qq → Xqq → WWqq → e𝜐𝜐µ𝜐𝜐qq

Main differences with the previous analysis:

• Data: 3.2 fb-1 in 2015 => 36.1 fb-1 in full 2015+2016

• Signal model: EWChL model => several individual signal models covering scalar,

vector and tensor resonances

• Signal mass range: extended from 500 GeV up to 3000 GeV

• Decay channel: ee, μμ, eμ => eμ (most sensitive one and easier to handle the SM bkg)

• Event categories: VBF Njet ≥ 2 => VBF Njet ≥ 2 + VBF Njet =1 (gain more sensitivities)

• Event selection and background estimation: optimized and updated accordingly

31/48

Signal models Signal models: new scalar, vector and tensor resonances introduced

• Heavy Higgs with a Narrow Width Approximation (NWA, scalar)

• Heavy Higgs with a Large Width Assumption (LWA, scalar)

• Georgi-Machacek (GM, scalar) model

width = 4 MeV (same widths for different heavy Higgs masses) mass = [200, 3000] GeV

width = 5%, 10% and 15% of heavy Higgs mass mass = [200, 3000] GeV

new scalar resonance: 𝐻𝐻50 (does not couple to fermions) parameters:

o 𝐻𝐻50𝑉𝑉𝑉𝑉 coupling, proportional to 𝑠𝑠𝑠𝑠𝑠𝑠𝜃𝜃𝐻𝐻 (= 0.4)o mass = [200, 1000] GeV

32/48

Signal models• Heavy Vector Triplet (HVT, vector) model new vector resonance: 𝑍𝑍′ parameters:

o 𝑍𝑍′VV coupling 𝑔𝑔𝑉𝑉 (=1)o mass = [300, 1000] GeVo to suppress the non-VBF contributions, assume that 𝑍𝑍′ bosons

does not couple to fermions

• Effective Lagrangian Model (ELM, tensor)

new tensor resonance (spin-2): T parameters:o TVV coupling 𝑓𝑓𝑚𝑚 (=1)o mass = [200, 1000] GeVo T resonance doesn’t couple to fermions

33/48

Event selection Event selections for signal regions (VBF Njet = 1 SR, VBF Njet ≥ 2 SR)

34/48

Background estimation Backgrounds:

• Top and WW (dominant backgrounds):

Normalization factors obtained from simultaneously fitting top and WW contributions to data in control and signal regions

• W+jets: data driven method - “fake-factor” method

• Z+jets, non-WW diboson, Higgs production: small contribution, MC simulation

Background estimation

• Top, WW, non-WW diboson, Z+jets, W+jets, Higgs production

WW in VBF 2J category which was from MC prediction(small contribution, diffcult to isolate a kinematic region with high purity of WW)

35/48

Control regions Definition of Top and WW control regions (CRs)

• some cuts inverted, removed or loosened to gain more data

statistics and higher purity

WW background:

• VBF Njet = 1: WW CR• VBF Njet ≥ 2: WW MC simulation

Top background:

• Top CR: combined VBF Njet = 1and VBF Njet ≥ 2 to gain morestatistics

36/48

Data vs. prediction in the control regions Top control region:

WW control region:

NF_Top_VBF = 1.12−0.12+0.13

NF_WW_VBF1J =1.0 ± 0.2

37/48

W+jets estimationW+jets estimation

• use a data-driven method - “fake factor method” to estimate its contribution

• two basic components: W+jets control sample and the fake factor

W+jets control sample (𝑁𝑁id+anti−id):

o selected from the data using the same event selections as SR but requiringone pair of id+anti-id leptons

o non-W+jets contributions (eg. Top and WW) 𝑁𝑁id+anti−idEW subtracted using MC

2.5

W+jets control sample fake factor

barrel end-cap

barrel end-cap

38/48

W+jets estimationW+jets estimation

fake factor ( 𝑁𝑁idNanti−id

):

o measured using a fake-enriched dijet data sample, with EW (W/Z+jets) subtraction o small trigger bias also considered (more in backup)

Electron

Muon

39/48

Data vs. prediction in the signal region Signal region:

No significant data excess above the SM background prediction observed

VBF 1J SR VBF 2J SR

mT

40/48

Systematic uncertainties Uncertainties of Top background (%):

• Only dominant uncer. shown, others included in “Total”• Single top: theoretical uncer. on single-top-quark production

Uncertainties of WW background (%):• Only dominant uncer. shown, others included in “Total”

Shape uncertainties of Top and WW:• mT shape dependence for the PDF uncertainty in the SRs considered

41/48

Systematic uncertainties Uncertainties of W+jets background:

• mainly arise from jet flavor composition, EW subtraction, stat. uncertainty, etc.

• VBF 1J SR: 32%, VBF 2J SR: 35%

Uncertainties of signal (NWA):

• mainly arise from parton shower, PDF, QCD scales, etc.

• VBF 1J SR: 5.1% - 9.0%, VBF 2J SR: 3.3% - 8%

Uncertainties of other background:

• smaller contributions on the uncertainties

• experimental uncertainties, normalized to high-order Xsec. prediction

42/48

95% CL upper limits No data excess above the SM prediction is observed. 95% CL upper limits are derived on σX×Br(X→WW) The frequentist method (CLs), is used to compute 95% CL upper limits

NWA LWA

GMValues above 1.3 pb at 200 GeV

and above 0.006 pb at 3 TeV for

NWA and 15% LWA are excluded

Scalar resonances

43/48

95% CL upper limits

Current sensitivity not sufficient to exclude the VBF signals from

GM, HVT and ELM models

HVT ELM

Vector resonance Tensor resonance

44/48

Summary for analysis 2

• SM backgrounds are carefully studied using simultaneous fit, fake factor

method and MC simulation, and validated in specific CRs.

• all the possible uncertainties are properly evaluated

• no evidence of such heavy neutral resonances is found

• 95% CL upper limits set on σX×Br(X→WW) of new scalar, vector and tensor

resonances predicted by several individual signal models

Search for new resonances (X) in qq → Xqq → WWqq → e𝜐𝜐µ𝜐𝜐qq using 36.1 fb-1 data

• Published on:

Eur. Phys. J. C, arXiv:1710.01123 [hep-ex] (submitted)

45/48

ATLAS MDT gas work (install, repair and maintain) MDT (Monitored Drift Tubes) gas system repairs and maintenance:

Installation of MDT gas system for BME chambers:

Obtained the ATLAS authorship and took a lot of shift work assigned to USTC

46/48

Publications Publications

• “Search for heavy resonances decaying into WW in the e𝜐𝜐μ𝜐𝜐 final state in pp collisions at 𝑠𝑠 = 13 TeV with the ATLAS detector”Submitted to Eur. Phys. J. C, arXiv:1710.01123 [hep-ex]

• “Search for heavy resonances in vector boson fusion”EPJ Web of Conferences 137, 08016 (2017)

• “Search for heavy neutral resonances in vector boson fusion in pp collisions at𝑠𝑠 = 13 TeV with the ATLAS detector at the Large Hadron Collider”

ATLAS-CONF-2016-053

• “Multi-Boson Simulation for 13 TeV ATLAS Analyses”ATL-PHYS-PUB-2017-005

Paper in ATLAS EB review• “Measurement of 𝑍𝑍𝑍𝑍 → ℓ+ℓ−𝜐𝜐�̅�𝜐 production in proton-proton collisions at 𝑠𝑠 =

13 TeV with ATLAS detector”ATL-COM-PHYS-2016-1801 (aim at Eur. Phys. J. C)

• “Search for heavy resonances in vector boson fusion at 𝑠𝑠 = 13 TeV with theATLAS detector at the LHC”, The 9th Joint Meeting of Chinese PhysicistsWorldwide, Tsinghua University, Beijing, China, 17-20 July, 2017http://meetings.csp.escience.cn/dct/page/1(2017 APS-OCPA Outstanding Conference Poster Award) 47/48

Conference talks and posters Conference talks

• “Search for heavy resonances in vector boson scattering”, XII Quark Confinement and the Hadron Spectrum, Thessaloniki, Greece, Aug. 29 - Sep. 03, 2016.https://indico.cern.ch/event/353906/contributions/2257680/

• “VBF/VBS Resonances”, ATLAS Beyond the Standard Model Higgs and Exotics Joint Workshop, Grenoble, France, 11-15 April, 2016.https://indico.cern.ch/event/465157/

• “Search for heavy neutral resonances in vector boson fusion qq → ℓ𝜐𝜐ℓ𝜐𝜐qq with the ATLAS detector”, Second China LHC Physics Workshop, Peking University,Beijing, China, 16-19 December, 2016http://indico.ihep.ac.cn/event/6062/contribution/66

Conference posters

48/48

Summary

• Search for new resonances (R) in qq → Rqq → ℓ+𝜐𝜐ℓ-�̅�𝜐qq (ℓ = e, µ)

using 3.2 fb-1 data collected in 2015

• Search for new resonances (X) in qq → Xqq → WWqq → e𝜐𝜐µ𝜐𝜐qq

using 36.1 fb-1 data collected in full 2015 and 2016

Two physics analyses performed using pp collision data recoded at

𝑠𝑠 = 13 TeV with the ATLAS detector at the LHC

4 publications, 1 paper under ATLAS review, 3 conference talks,

1 conference poster (outstanding award)

and many important ATLAS group talks (eg. unblinding & approval talks)

ATLAS MDT gas system installation, repair and maintence

49

Backup

50

Mandelstam variables (analysis 1) Definitions of Mandelstam variables (s, t, u)

• 𝑠𝑠 ≈ 2𝑝𝑝1 � 𝑝𝑝2 ≈ 2𝑝𝑝3 � 𝑝𝑝4• 𝑡𝑡 ≈ −2𝑝𝑝1 � 𝑝𝑝3 ≈ −2𝑝𝑝2 � 𝑝𝑝4• 𝑢𝑢 ≈ −2𝑝𝑝1 � 𝑝𝑝4 ≈ −2𝑝𝑝3 � 𝑝𝑝2

s-channel

t-channel

u-channel

Relativistic limit (large p)

• 𝐸𝐸2 = 𝐩𝐩 � 𝐩𝐩 + 𝑐𝑐02 ⇒ 𝐸𝐸2 ≈ 𝐩𝐩 � 𝐩𝐩

⇒ 𝑠𝑠 = (𝑝𝑝1 + 𝑝𝑝2)2= 𝑝𝑝12 + 𝑝𝑝22 + 2𝑝𝑝1 � 𝑝𝑝2≈ 2𝑝𝑝1 � 𝑝𝑝2

where 𝑝𝑝12 = 𝑐𝑐12, 𝑝𝑝22 = 𝑐𝑐2

2

• 𝑠𝑠 = (𝑝𝑝1 + 𝑝𝑝2)2=(𝑝𝑝3 + 𝑝𝑝4)2

• 𝑡𝑡 = (𝑝𝑝1 − 𝑝𝑝3)2=(𝑝𝑝4 − 𝑝𝑝2)2

• 𝑢𝑢 = (𝑝𝑝1 − 𝑝𝑝4)2=(𝑝𝑝3 − 𝑝𝑝2)2

51

Feynman diagrams of VBF (analysis 1) NonVBF-EW process

VBF-QCD process

52

Effective theories and Unitarity (analysis 1)

53

Adding a resonance (analysis 1)

K-matrix

54

K-matrix unitarization (analysis 1)

55

K-matrix unitarization (analysis 1)

56

Object definition (analysis 1)Selection Electron MuonpT (GeV) >25 >25|η| <2.47, veto 1.37-1.52 <2.5| d0/σ(d0) | <5 <3| z0 sinθ | (mm) <0.5 <0.5ID TightLH (if Et_el<300 GeV)

MediumLH(if Et_el>300 GeV)Medium

Isolation passTightIso passTightIso

Selection JetJet type AntiKt4EMTopoJets

pT (GeV) >30(>50 if 2.5<|η|<4.5)

|η| <4.5

JVT JVT > 0.64 if |η| < 2.4 and pT < 50 GeV

Jet quality Not badjet

Jet flavor tagger MV2c20 (85% efficiency)

Selection METMETContainer MET_Reference_Anti

Kt4EMTopoObjets used to rebuild MET

Electrons, muons, jets

METSoftTerm Track soft terms

MET(GeV) >35

57

Triggers and signal acceptance (analysis 1) Single-lepton triggers:

Signal acceptance times efficiency:

58

Lepton centrality and frecoil (analysis 1) Lepton centrality ζ :

frecoil:

• Measures the strength of the recoil system relative to the dilepton system

• ϛ in VBF topology tends to be positive

• To reduce the background from strongproduction of double vector boson processes(ϛ > -0.5)

•

• Useful to reject the Z/γ* → ℓℓ background• frecoil < 2

59

Real rate of electron (analysis 1)

Selection Tight Electron Loose ElectronpT (GeV) >25 >25|η| <2.47, veto 1.37-1.52 <2.47, veto 1.37-1.52| d0/σ(d0) | <5 <5| z0 sinθ | (mm) <0.5 <0.5ID TightLH

(MediumLH, if Et_el>300GeV)LooseLH

Isolation passTightIso No isolation requirement

Tight/Loose electron definitions:

Cut flow:• Basic event pre-selection, event cleaning/GRL cut/PV cut.

• Single-electron trigger chain(e24_lhmedium_L1EM20VH || e60_lhmedium || e120_lhloose)

• Exactly two LOOSE electrons with opposite sign

• mZ -10GeV < mee < mZ +10GeV

• At least one Tight lepton as Tag

• For the probe lepton, real rate = NTight/NLoose

60

Fake rate of electron (analysis 1)

Selection Tight Electron Loose ElectronpT (GeV) >25 >25|η| <2.47, veto 1.37-1.52 <2.47, veto 1.37-1.52| d0/σ(d0) | <5 <5| z0 sinθ | (mm) <0.5 <0.5ID TightLH

(MediumLH, if Et_el>300GeV)LooseLH

Isolation passTightIso No isolation requirement

Tight/Loose electron definitions:

Cut flow:• Basic event pre-selection, event cleaning/GRL cut/PV cut.

• use single-electron trigger chain (e24_lhloose_L1EM20VH || e60 _lhloose || e120_lhloose)

• At least one LOOSE electrons

• reject events with two loose electrons in Z mass window(|Mee-MZ|<20GeV)

• reject events with two or more tight electrons

• MET<25GeV

• Fill histrograms, fake rate = NTight/NLoose

61

Signal models (analysis 2) Signal models: new scalar, vector and tensor resonances introduced

• Heavy Higgs with a Narrow Width Approximation (NWA)

• Heavy Higgs with a Large Width Assumption (LWA)

• Georgi-Machacek (GM) model

width = 4 MeV (same widths for different heavy Higgs masses) mass = [200, 3000] GeV

width = 5%, 10% and 15% of heavy Higgs mass Mass = [200, 3000] GeV

extend Higgs sector with additional one real and one complex triplets scalar potential determined to keep the ratio of charged to neutral currents as

in SM (ρ parameter = 1)

62

• Georgi-Machacek (GM) model reorganize the physical states to have

• 2 singlet neutrals (ℎ,𝐻𝐻)• triplet (𝐻𝐻3+,𝐻𝐻30,𝐻𝐻3−)• fiveplet (𝐻𝐻5++,𝐻𝐻5+,𝐻𝐻50,𝐻𝐻5−,𝐻𝐻5−−)

parameters• All the H5VV couplings are proportional to 𝑠𝑠𝑠𝑠𝑠𝑠𝜃𝜃𝐻𝐻 (fraction of the gauge

boson masses 𝑐𝑐𝑊𝑊 and 𝑐𝑐𝑍𝑍 generated by the vev of the triplets)• resonance mass: [200, 1000] GeV• H5 scalars does not couple to fermions

• Heavy Vector Triplet (HVT) model parameterizes the couplings of the HVT bosons to the SM gauge bosons and

Higgs with 𝑐𝑐ℎ𝑔𝑔𝑉𝑉, to the fermions with 𝑔𝑔2𝑐𝑐𝐹𝐹/𝑔𝑔𝑉𝑉 new resonances (𝑍𝑍′, 𝑊𝑊′) parameters

• XVV coupling 𝑔𝑔𝑉𝑉 (=1), resonance mass [300, 1000] GeV• to suppress the non-VBF contributions, assume that HVT bosons

does not couple to fermions (𝑐𝑐𝐹𝐹=0)

Signal models (analysis 2)

63

• An an effective Lagrangian model (ELM)

introduce general spin-2 fields 𝑇𝑇𝜇𝜇𝜇𝜇 (singlet) and 𝑇𝑇𝑗𝑗𝜇𝜇𝜇𝜇(triplet)

spin-2 singlet case considered, its effective Lagrangian

a form factor introduced to multiply with the amplitudes topreserve unitarization

parameters• characteristic energy scale: Λ = 1.5 TeV• cut-off energy scale and suppression power: Λ𝑓𝑓𝑓𝑓 = 3 TeV, 𝑠𝑠𝑓𝑓𝑓𝑓 = 4• variable coupling parameters: 𝑓𝑓1 = 𝑓𝑓2 = 𝑓𝑓5 = 1• resonance mass: [200, 1000] GeV• Spin-2 resonances doesn’t couple to fermions

Signal models (analysis 2)

64

GM model (analysis 2)https://cds.cern.ch/record/2002500/files/LHCHXSWG-2015-001_2.pdf

65

GM model (analysis 2)https://cds.cern.ch/record/2002500/files/LHCHXSWG-2015-001_2.pdf

66

HVT Model (analysis 2)https://arxiv.org/pdf/1402.4431.pdf

67

Spin-2 Model (analysis 2)https://arxiv.org/pdf/1211.3658.pdf

68

Spin-2 Model (analysis 2)https://arxiv.org/pdf/1211.3658.pdf

69

Triggers and mT definition (analysis 2) Nominal triggers

Triggers for selecting dijet sample in W+jets estimation

mT definition

70

Data/MC sample for analysis 2 Data samples:

• Full data in 2015+2016, Luminosity = 36.5 fb−1

MC samples:

• Signal samples

NWA: Powheg + Pythia8 LWA: MG5_aMC@Nlo + Pythia8 GM/HVT: MG5_aMC@Nlo + Pythia8 ELM: VBFNLO + Pythia8

• Background samples

Top: Powheg + Pythia8 WW: Sherpa 2.2.1 Non-WW diboson: Sherpa 2.2.1 Z+jets: Sherpa 2.1.1 Higgs: Powheg + Pythia8

71

Signal acceptance times efficiency (analysis 2)

72

W+jets estimation (analysis 2)

• selected using the single-lepton prescaled triggers with the low-pT

thresholds of 12 (14) GeV for electrons (muons)

Dijet sample for fake factor measurement:

73

W+jets estimation - trigger bias (analysis 2) A small trigger bias:

• Introduced when the anti-lepton fires the triggers while the id leptondoes not fire the triggers.

• Triggered fake factor:using the nominal unprescaled single-lepton triggers to select dijet sample

Nominal fake factor: applied to the most of events (92%) in the W+jets control sample

Triggered fake factor: only applied to the events (8%) where the anti-lepton fires the triggers and the id lepton does not fire the triggers.

74

W+jets estimation - CR (analysis 2) VBF 1J category

VBF 2J category

75

W+jets estimation (analysis 2)

76

W+jets estimation (analysis 2)

77

mT shape uncertainties of Top and WW (analysis 2)

78

mT shape uncertainties of NWA signal (analysis 2)

79

Correlations among nuisance parameters (analysis 2) NWA m = 800 GeV:

correlation coefficients > 0.4 are shown

80

ATLAS MDT gas system work Off-detector MDT gas system

• Supplies a steady gas flow

• Baseline of MDT gas mixture:

Ar ~93%, CO2 ~ 7%, H2O~0.075%

• Chamber volume: ~725000 nl

• Operating pressure: 3 bar

• Running in a loop, driven by a pump

• Fresh gas input is 10% per day,

10% of gas is flushed per day.

• 15 distribution racks , each one

contains 16 ~ 24 gas channels

• Controlled by GCS.https://atlasop.cern.ch/twiki/bin/view/Main/MDTGasSystemOverview

Off-detector MDT gas system

81

ATLAS MDT gas system work On-detector MDT gas system

https://atlasop.cern.ch/twiki/bin/view/Main/MDTGasSystemOverview

On-detector MDT gas system

82

ATLAS MDT gas system workMDT leak measurement

Leak rate = 104 mbar/d

Procedure:

-- Rack channel closed-- Measuring temperature

and pressure-- Calculating temperature

corrected pressure (normalized to Tn=20℃)

Calculating gas loss usingp1V0/T1=pnVn/Tn

-- p1 : measured (uncorr.) pressure-- V0: channel gas volume-- T1: measured temperature-- Vn: norm volume(3*V0)-- Tn: norm temperature(293K)-- Pn: derived normalized

pressure

83

ATLAS MDT gas system workMDT EO gas leak repairs (2014-2015)

2015

2014

EO gas leaks improved a lot.

1: gas-jumpers made of NORYL (old), 2 : gas-jumper made of POCAN (new)

EO A side:Philipp Fleischmann, Guangyi ZhangEO C side:Anatoli Kozhin ,Vladimir Gushchin

In the most case, gas leaks ofEO are caused by cracks of gasjumpers made of NORYL

84

ATLAS MDT gas system workMDT Barrel gas leak repairs (2014-2015)

BOL5A01_ML2BOL2A03_ML1BIS6C14_ML1

MDT Barrel gas leak measurement (2014-2017)

BME4A13_ML1 BMF2A14_ML1 EIL2A11_ML2 BIL4A01 ML1

85

ATLAS MDT gas system workMDT EO gas leak measurement and repairs (2016-2017)

Gas connections and leak measurement for all new MDT BMG chambers (2017):

Before repairs(2016)

After repairs(2017)

86

Data taking

87

LHC / HL-LHC Plan