tt+X production

at ATLAS and cms

Nicolas TONON (DESY)

on behalf of the ATLAS and CMS Collaborations

55 th Rencontres de Moriond QCD, 30 th March 2021

2introduction

~140 fb⁻¹ collected per experiment during Run2Most top processes in precision differential regime

Challenging measurements with complex final states➢ Benefit from advanced identification techniques

tt+HF a crucial handle on modeling & backgrounds➢ Probe different mass scales, test of QCD

tt+V production direct probe of top quark couplings➢ Modified in many BSM models➢ Backgrounds to other Top/Higgs analyses

+HF

3σ(tt+z) Measurements

Direct probe of t-Z coupling

ttZ inclusive & differential➢ 2015–2018 data / 139 fb⁻¹

ttZ inclusive & differential & EFT➢ 2016–2017 data / 77.5 fb⁻¹

3ℓ and 4ℓ (e,μ) channels Events categorized by # of leptons and (b)jets OSSF pair |mℓℓ – mZ| < 10 GeV (in 3ℓ SR) Dominant backgrounds :

t(t)X, WZ, ZZ Nonprompt leptons ↔ Data-driven estimation

(tight-to-loose ratio method)

JHEP 2003 (2020) 056 ATLAS-CONF-2020-028

4σ(ttz) Measurements

➢ Lepton ID, WZ and t(t)X normalizations

→ In agreement & more precise (8–10 %) than NLO+NNLL prediction (10 %)

➢ Parton shower, tZ/WZ/ZZ modeling, b-tagging

Eur. Phys. J. C 79, 249 (2019)

WZ/ZZ+light jets normalizations free in fit Simultaneous fit to 14 event categories

139 fb⁻¹ 77.5 fb⁻¹ JHEP 2003 (2020) 056

ATLAS-CONF-2020-028

5

Differential measurementsversus pt(Z) and cosθ*(Z, ℓZ

-)➢ Parton-level

Constrain anomalous couplings and EFT operators

Differential measurements versus 9 observables (↔ Z, tt, njets)➢ Parton- & particle-level

→ Agreement with NLO+NNLL predictions J. High Energ. Phys. 2019, 39 (2019)

pT(Z)

pT(Z)|ΔΦ(ℓ,ℓ)|

σ(ttz) MeasurementsJHEP 2003 (2020) 056

ATLAS-CONF-2020-028

6σ(tt+γ) Measurements

Direct probe of t-γ coupling

ttγ and tWγ inclusive & differential➢ eμ channel➢ 2015–2018 data / 139 fb⁻¹

ttγ inclusive & differential & EFT➢ ℓ + jets channel➢ 2016–2018 data / 137 fb⁻¹

Signal modeled at LO in QCD→ Include photons from

ISR/top/top decay products with Madgraph

Inclusive measurements more precise than predictions

10.1007/JHEP09(2020)049 CMS-PAS-TOP-18-010

7Atlas σ(ttγ) – 2ℓ Combined measurement of ttγ+tWγ production

➢ eμ OS (pT>25) / = 1 γ (pT>25) / ≥ 2 j (pT>25) / ≥ 1 bj (ε~85 %)➢ High purity, backgrounds estimated from simulation

Include (non-)resonant + interference + off-shell effects

139 fb⁻¹

Inclusive fit to scalar pT sum of all [ℓ+γ+j+MET] :

tWγttγ

Parton-level unfolding to 6 observables sensitive to ttγ modeling, BSM effects, tt spin correlations

All results in agreement with predictions

J. High Energ. Phys. 2018, 158 (2018)J. High Energ. Phys. 2019, 188 (2019)

pT(γ)

→ Comparison with NLO prediction for pp → bWbWγ :

➢ Parton shower, ISR, luminosity

10.1007/JHEP09(2020)049

t

t

t

W

8cms σ(ttγ) – +jetsℓ 137 fb⁻¹

= 1 e or μ (pT>25) / = 1 γ (pT>20) / ≥ 3 j (pT>30) / ≥ 1 bjet (ε~70 %) Constrain main backgrounds in-situ with dedicated sidebands Simultaneous fit to 12 SR + 34 CR categories

ttγ modeling, background estimation, JES

pT(γ)

3j, e 3j, μ ≥ 4j, e ≥ 4j, μ

Particle-level unfolding to pt(γ), |η(γ)|, ΔR(ℓ,γ)

All results in agreement with predictions

Constrain 2 EFT operators impacting t-γ/Z vertices

CMS-PAS-TOP-18-010

EFT

9Direct eft search in t(t)x Target {ttℓℓ/ttℓν/tℓℓq/ttH/tHq} processes simultaneously in {2ℓ SS/3ℓ/4ℓ + bjets} channels

Constrain 16 EFT operators impacting top vertices

Signal event weights parameterized with all Wilson coeffs.➢ Full detector simulation of SMEFT samples

Categorization on objects multiplicities & properties→ Enhance sensitivity to different processes / operators

41.5 fb⁻¹ JHEP 2103 (2021) 095

→ After fit of 16 Wilson

coeffs. to data

10σ(tt+bb) measurements

➢ 2015–2016 data / 36.1 fb⁻¹

2ℓ, ℓ+jets JHEP 04 (2019) 046 JHEP 07 (2020) 125

σ(ttbb) measurements crucial to… ➢ Improve ttbar modeling, compare different

generators, test QCD (ISR flavor, etc.) Large background to ttH(bb), 4tops, and more Unique challenges → flavour-tagging is key !

b-tagging, limited sample size, JES, etc.

➢ 2016 data / 35.9 fb⁻¹

2ℓ, ℓ+jets

All hadronic PLB 803 (2020) 135285First differential result @13 TeV

eμ and ℓ+jets channels, fiducial σ in ≥ 3,4bjets phase spaces Analysis agnostic to origin of extra b

➢ Substract non-tt background from MC➢ tt+(LF/b/c) from fit to 3rd (4th) highest btagging scores➢ Substract ttH/V(bb)

1136.1 fb⁻¹ ATLAS σ(tt+bb) – 2 , ℓ +jetsℓ

More precise than theory predictions Compare w/ many generators → Overall under-prediction Compatible within uncertainties

Precision

13 %

17 %

Several observables unfolded at particle-level→ Some mostly sensitive to b (not) from top

JHEP 04 (2019) 046

12

2ℓ and ℓ+jets channels, results in fiducial and full phase-spaces Measure : σ(ttjj), σ(ttbb)/σ(ttjj) → σ(ttbb)

Extra jets assignment :➢ 2ℓ ↔ jets with 3rd/4th highest btagging scores➢ ℓ+jets ↔ top quark kinematic fit

35.9 fb⁻¹ cms σ(tt+bb) – 2 , +jetsℓ ℓ

2ℓ ℓ+jets

Good overall agreement with predictions Some under-prediction for σ(ttbb)2ℓ

Fit to 2D distribution of btagging scores of extra jetsHighest score among ‘extra jets’

2nd h

ighe

st s

core

12/13 % precision on σ(ttbb) in ℓ+jets/2ℓ

85/23 %

40/12 % for ttjj/bb

ttbb ttbj

ttLFttcc

JHEP 07 (2020) 125

13

≥ 8j, ≥ 2bj → branching ratio, reconstructible final state

Multiple MVA techniques to reduce huge QCD background

35.9 fb⁻¹ cms σ(tt+bb) – all jets

SR SR

Fit distributions of 2 highest btagging scores of ‘extra jets’➢ Include CRs → Data-driven QCD background estimation

Under-predictions from all generators (1–2 σ)

Quark-gluon likelihood Weakly-supervised BDT

+ Jet-assignment BDT

(tt VS QCD)(QCD ↔ + gluon jets)

ttbbttjj

PLB 803 (2020) 135285

14

First measurement of σ(ttcc) c-tagging even more challenging

41.5 fb⁻¹ cms σ(tt+cc) measurements

Exploit c-jet tagger (DeepCSV)→ Calibrated* in CR

Train permutation-NN to assign extra (HF) jets

Train multiclass NN to separate ttcc / ttbb / ttLF

Template fit to 2D distribution of NN response (unrolled)

Measure σ(ttcc), σ(ttbb), σ(ttLF)+ ratios to σ(ttjj)

Consistent with Powheg within 1–2 σ 20 % precision on σ(ttcc), ~ overprediction Some under-prediction for σ(ttbb)

ttbb ttLFttcc

CMS-PAS-TOP-20-003

CMS-PAS-BTV-20-001*

15summary Presented ttZ and ttγ inclusive/differential measurements based on Run2 data

➢ In good agreement, and more precise than theory predictions➢ ttV processes have entered precision regime → More EFT interpretations

Direct CMS measurement of 16 top-EFT operators simutaneously in t(t)X production Presented tt+HF results in multiple final states, overall consistent with simulations

➢ Under-predictions of ttbb → Calls for more differential studies➢ First CMS measurement of ttcc → Fully-consistent treatment of jet flavours

pT(γ)pT(Z) EFT in t(t)X ttbb incl./diff. ttbb 1ℓ,2ℓ ttbb hadr.

ttcc