Direct and Indirect Searches of New Physics in Multi ... fileC.Maiani 28.02.2014 séminaire LPNHE Je...

Direct and Indirect Searches of New Physics in Multi-Bosons Final States at ATLAS

Camilla MaianiCEA Saclay

28.02.2014

séminaire LPNHE

C.Maiani 28.02.2014 séminaire LPNHE

Je Me Présente

★ Master thesis at Rome University “La Sapienza”

‣ title: Development of the muon isolation algorithm at ATLAS

‣ supervisors: prof. Carlo Dionisi, prof. Stefano Giagu

★ Ph.D. thesis at Rome University “La Sapienza”

‣ title: J/ψ→µ+µ- cross-section and B-lifetime determination at ATLAS

‣ supervisors: prof. Carlo Dionisi, prof. Stefano Giagu, doct. Marco Rescigno

★ Post-Doctorate at CEA Saclay within Samira Hassani’s ERC-DIBOSON project

‣ project goal: diboson production for SM measurements and new physics searches

at ATLAS

Oct 2008 - Jan 2012

2

Jan 2008 - July 2008

Feb 2012 - Feb 2014/2016


★ Indirect new physics searches using diboson production at ATLAS‣ cross-section and Triple Gauge Coupling (TGC) measurement in Wγ‣ first anomalous Quartic Gauge Couplings (QGC) measurement in

Wγγ★ Direct new physics searches using diboson production at ATLAS

‣ model independent searches for new resonances decaying to Wγ★ Other ways: using the Higgs-candidate as a probe at ATLAS

‣ spin-parity measurement in the H→ZZ(*)→4ℓ decay channel

Presentation Outline

3

Introducing Triple Gauge Couplings (TGC)

C.Maiani 28.02.2014 séminaire LPNHE4

Triple gauge couplings are a direct consequence of the SU(2) x U(1) structure of the electroweak sector

★ Indirect searches★ Direct searches★ Other ways: using the Higgs

Defining an effective Lagrangian:

ρT

γγ

W

New physics could:→ modify allowed triple gauge couplings → introduce new ones...

Measurement of TGCs → study of di-boson production: high stat, clean measurements → gives access to new physics in the high energy range


TGC Sensitivity to New Physics

5

How do new physics processes relate to aTGCs?→parameters

measurable in Wγ(→ Z’, W’)

WZ analysis


WZ analysis

★ Analyzing ATLAS 2011 dataset★ Signal:

‣ Wγ → ℓνγ★ Backgrounds estimated from data

‣ W+jets, γ+jets★ Other backgrounds (from MC)

‣ Drell-Yan, WW/WZ/ZZ, top


Wγ Analysis Overview

6

→ ABCD method

★ Main systematic uncertainties‣ luminosity ~3.9%‣ photon identification ~6%‣ jet energy scale ~2-3%‣ EM scale and resolution ~1.5-3%‣ will improve with more stats!


aTGC extraction: pT(γ) > 100 GeV, Njet = 0

ex-fid cross-section measurement [pb]Njet ≥ 0: 2.77 ± 0.03 (stat.) ± 0.33 (syst.) ± 0.14 (lumi.)

Njet ≥ 0 (MCFM): 1.96 ± 0.17

Njet = 0: 1.76 ± 0.03 (stat.) ± 0.21 (syst.) ± 0.08 (lumi.)

Njet = 0 (MCFM): 1.39 ± 0.13

The likelihood function is a Poissonian depending on Nobs and Nexp:

TGC likelihood functions can have more than one minimum

Limits extraction๏ using negative log-likelihood ratio๏ applying frequentist procedure to look for 95% CLs interval


Limits Extraction

7

→ Maximizing the likelihood is not enough!

nobs < nexp

0 Δkγ

nobs > nexp

0 Δkγ

(1-p

)

(1-p

)

0.950.95


→ developed a tool used in other SM EW analyses

N

i

s

(�tot

W�

, {xk

}) = �

tot

W�

· A · C ·Z

L(t)dt · (1 +nX

k=1

x

k

S

i

k

)

(p0 + p1 ⇤ �� + p2 ⇤ �� +p3 ⇤ �2� + p4 ⇤ �� ⇤ ��+p5 ⇤ �2

�) · A · CSM cross-sec


Wγ Observed TGC Limits

8

compatible with

SM predictions

1D limits correlation


★ My role:‣ Wγ background data driven estimates‣ W/Zγ cross-section extraction‣ W/Zγ TGC limits setting→ publications: Phys. Rev. D 87, 11 (2013)→ one conference talk

★ All channels studied with 2011 data @ 7 TeV

‣ no deviations found wrt the Standard Model

★ Sensitivity is still low!

‣ the channel with highest statistics Wγ gives Δκγ < 0.4 and λγ < 0.05

‣ the “interesting” range is a factor 10 away

★ Improvements expected soon

‣ first results on quartic Gauge couplings!

‣ analyzing full 2012 data sample

‣ combining channels sensitive to the same couplings

★ Need to run at 13 TeV (→ higher sensitivity) and 100 fb-1 to probe

the interesting region


Perspectives on aTGC and aQGC Measurements

9

→ Δκγ ~ 0.01 and λγ ~ 0.001

→ 20.3 fb-1 @ 8 TeV

→ 4.7 fb-1 @ 7 TeV

→ 2 to 3 years of data taking


→ I am contact editor of Wγγ analysis


★ Goal: perform model-independent searches for new resonances decaying to W(ℓν)γ or Z(ℓ)γ final states

Direct, Model Independent Searches

10

★ Strategy:‣ Study background composition in “SM” dominated region‣ Perform signal+background fit on mT(Wγ) for different signal massese‣ Extract exclusion/discovery limits by using ATLAS frequentist approach


★ My role:‣ Main author of Wγ analysis: from background estimates to statistical treatment‣ Editor of the publication (PLB in preparation)

https://twiki.cern.ch/twiki/pub/AtlasProtected/StatisticsTools/Frequentist_Limit_Recommendation.pdf

https://twiki.cern.ch/twiki/pub/AtlasProtected/StatisticsTools/Frequentist_Limit_Recommendation.pdf


Background Composition

11


Checking data-MC agreement, mT(Wγ) is blind

ATLAS Work-in-Progress




Mass Modelling for Signal Extraction

12

Performing unbinned extended maximum likelihood mass fit

Signal pdf: Gaussian + Crystal Ball (CB)

Background pdf: ๏ baseline: sum of two exponentials

L =

Y

categories

Pois(Nsig

+ Nbkg

)·Y

events

fsig · PDFsig(mT (W�)) + (1� fsig) · PDFbkg(mT (W�))

empirical


Wγ signal pdf: ๏ empirical model optimized on benchmark MC samples๏ extrapolation for masses in between๏ very different resolutions at low and high mass and between decay channels


Signal Modelling Overview

13

µνγeνγ


ATLASWork-in-Progress



Background Modelling Overview

14

Wγ background pdf: ๏ using background expectations (from MC and data driven estimates) to optimize the background shape

NB: the parameters used in the final limits extraction will be those taken from a direct fit on data




★ Main systematics:‣ normalization: luminosity, γ ID, γ isolation‣ shape: small uncertainty on resolution

★ Final steps of the analysis on-going:‣ unblinding of 2012 data


Expected Limits @ 8 TeV

15

]2) [GeV/cγ(WTm200 400 600 800 1000 1200 1400 1600

) @ 9

5% C

Lγ)

ν W

(l→

BR

(X×

ext/,

fidσ

-210

-110

1

10γ) ν W(l→ Ta

Expectedσ 1 ±Expected σ 2 ±Expected

InternalATLAS

= 8 TeVs, -1Ldt = 20.3 fb∫γνl

→ expected next week


]2) [GeV/cγ(WTm200 400 600 800 1000 1200 1400 1600

) @ 9

5% C

Lγ)

ν W

(l→

BR

(X×

ext/,

fidσ

-210

-110

1

10Signal injection @700 GeV

γ) ν W(l→ Ta

Expected

σ 1 ±Expected

σ 2 ±Expected

InternalATLAS

= 8 TeVs, -1Ldt = 20.3 fb∫γνl

expected for background only hypothesis signal injection at mT(Wγ) = 700 GeV

2011 exclusion~ 700 GeV

★ My role:‣ development of the MELA framework on H→ZZ(*)→4ℓ‣ understanding of detector acceptance and selection effects‣ contribution to systematic uncertainties estimate‣ extraction of signal/background hypotheses separations‣ editor of the spin-parity part of ATLAS-CONF-2013-013


Higgs-Candidate as a Probe to Look for New Physics

16

The discovery of a new Higgs-like particle opens a number of measurements★ significance, mass and couplings★ spin-parity

These allow to test the new particle → is it a SM Higgs? Is it new physics (composite Higgs)? Does it point to new physics (CP violation terms, supersymmetric partners)?

in summer 2012 a new Higgs-like particle is found


→ publications: ATLAS-CONF-2012-169, ATLAS-CONF-2013-013, Phys. Lett. B 726, 1–3 (2013)

→ one talk at LHC France

https://cds.cern.ch/record/1499628




http://www.sciencedirect.com/science/article/pii/S0370269313006369




H→ZZ(*)→4ℓ is very clean, and the full decay kinematic is measured

Defining a 1D discriminant from all the observables sensitive to JP:

In MELA approach: m1, m2, cosθ*, ϕ1, cosθ1, cosθ2, ϕ


mZ*H→ZZ(*)

mZ*H→ZZ(*)

Observables and Separation Power

17

http://arxiv.org/pdf/1208.4018v1.pdf

production and decay angles definition

red: 0+

blu: 0+hgreen: 0-h


→ using full theoretical description of signal final states

→ including corrections for detector/selection effects



→ excluding 0-, 1+, 1- at > 95% CL→ data prefers the SM Higgs hypothesis


Separations in H→ZZ(*)→4ℓ

18


obs log(L(H0)/L(H1))

p-value assuming H0

JP-MELA discriminant:

Test one hypothesis (H0) against another one (H1)

★ assuming that the spin-parity is 0+

★ testing against non-SM hypotheses: 0-, 1±, 2m±

★ spin-2: varying ggF/qq̅ production fraction

P(H0)

P(H0) + P(H1)JP-MELA =

0+ vs 0-


Study of Spin-2 Admixtures

19

★ for spin-2: gluon-gluon fusion, qq̅ production mechanisms or an

admixture of the two are allowed★ testing different spin-2 production mechanisms, fqq̅ = 0, 25, 50, 75, 100%

combining with γγ and WW excluding graviton-inspired spin-2+ model at 99.9%

JP-MELA


[%]qqf

0 25 50 75 100

)) 1)/L

(H0

log(L(

H

-10

0

10

20

30

40 ATLAS Preliminary

4l→ ZZ* →H -1Ldt = 4.6 fb∫ = 7 TeV: s

-1Ldt = 20.7 fb∫ = 8 TeV: s

γγ →H -1Ldt = 20.7 fb∫ = 8 TeV: s

νeνµ/νµν e→ WW* →H -1Ldt = 20.7 fb∫ = 8 TeV: s

Data

Signal hypothesis

+ = 00H

PJ

+ = 21H

PJ

Spin 0

σ1

σ2

[from ATLAS-CONF-2013-040]

https://cds.cern.ch/record/1542341/files/ATLAS-CONF-2013-040.pdf

https://cds.cern.ch/record/1542341/files/ATLAS-CONF-2013-040.pdf


Conclusions and Plans

20

All these results (plus the huge work done on SUSY, exotics, SM precision

measurements) indicate that the Standard Model stands his ground★ no indications of new physics★ extraordinary result: discovery of new Higgs-like particle!

There are open issues still★ dark matter: new matter? new force?★ naturalness and fine tuning problems → new physics at the TeV scale

My plans for the near future★ pursue searches for new physics in preparation of the 13 TeV data★ participate to the ATLAS trigger upgrade: plan to join the Saclay effort on the

New Small Wheels in the spring

we just reached the TeV scale, but haven’t quite explored it yet


Backup Slides


Research Activities

★ Muon isolation trigger algorithm studies at ATLAS‣ isolation algorithm design and commissioning

★ J/ψ→µµ studies in p-p and lead-lead collisions‣ J/ψ→µµ first observation and non-prompt to prompt production separation

‣ B average lifetime measurement in inclusive B→J/ψ X→µµ X decays

‣ first J/ψ→µµ studies in lead-lead collisions

★ Wγ cross-section measurement and aTGC limits extraction‣ Wγ signal extraction

‣ W/Zγ cross-section measurement

‣ anomalous Triple Gauge Coupling limits extraction in W/Zγ★ First Higgs spin parity measurement at ATLAS

‣ H→ZZ(*)→4ℓ

★ Current activities and plans

2008

2010

2012

first LHC collisions!

Higgs boson discovery

2013 22

Master thesis

Ph.D.

Post.Doc.

Aug2012


Wγ 2011


List of (a)TGC Parameters per Diboson Channel

24

coupling parameters channel

WWγ λγ, Δκγ WW, Wγ

WWZ λZ, ΔκZ, Δg1Z WW, WZ

ZZγ h3Z, h4Z Zγ

Zγγ h3γ, h4γ Zγ

ZγZ f40Z, f50Z ZZ

ZZZ f40γ, f50γ ZZ

SM allowed

SM forbidden


Form Factors in aTGC Measurements

25


Current Studies on Form Factors (ZZ)

26


W/Zγ Cross-Section Measurement

27

★ using a maximum likelihood fit to extract the total cross-section of W/Zγ production processes

number of expected signal and background events..

free parameter

N

i

s

(�tot

W/Z�

, {xk

}) = �

tot

W/Z�

· A · C ·Z

L(t)dt · (1 +nX

k=1

x

k

S

i

k

) N

ib({xk}) = N

ib(1 +

nX

k=1

xkB

ik)

The probability that the number of expected signal and background events gives the number of events observed Niobs is regulated by a Poissonian function:

gaussian term to constrain the xk nuisance parameters

�ln[L(�, {xk

})] =2X

i=1

�ln

⇣e

�N

i

exp

(�tot

W/Z�

,{x

k

}) · N i

exp

(�tot

W/Z�

, {xk

})N

i

obs

(N i

obs

)!

⌘+

X

k

x

2k

2

sum on the muonic/electronic channels

★ fit allows to take into account all correlations between systematic uncertainties

when combining channels


Jet Multiplicity in Wγ

28

→ more jets at higher photon ET


aTGC Effective Lagrangian

29

LHC has opened new era of Quartic Gauge Coupling (QGC) measurements:๏ Vector Boson Scattering of particular interest → confirm unitarization๏ search for new resonances in the multi-TeV range

Wγγ paper in preparation:๏ fiducial Wγγ production cross-section๏ first QGC limits at ATLAS

QGC extraction: ๏ same techniques as for TGCs

QGC

...many other vertices

Work-in-Progress


First Look at Quartic Gauge Couplings: Wγγ

30

→ 20.3 fb-1 @ 8 TeV


★ My role:‣ contact person of the analysis and editor of the publication‣ responsible for the cross-section measurement

→ publication on PLB forseen before March 2014


Wγ Resonances 2012


Mass Fit @ 200 GeV

Transverse Mass (GeV)ail120 140 160 180 200 220 240 260

Even

ts /

( 0.7

5 G

eV )

0

20

40

60

80

100

Transverse Mass"aiA RooPlot of "l Transverse Mass"aiA RooPlot of "l

Transverse Mass (GeV)ail100 120 140 160 180 200 220 240 260

Even

ts /

( 0.8

GeV

)

0

10

20

30

40

50

60

70

80

Transverse Mass"aiA RooPlot of "l Transverse Mass"aiA RooPlot of "lµνγ eνγ

X2/NDoF = 259.235/141 = 1.83855

X2/NDoF = 111.546/94 = 1.18666

32


Mass Fit @ 1.6 TeV

Transverse Mass (GeV)ail1000 1200 1400 1600 1800 2000

Even

ts /

( 5.7

5 G

eV )

0

50

100

150

200

Transverse Mass"aiA RooPlot of "l Transverse Mass"aiA RooPlot of "l

Transverse Mass (GeV)ail1000 1100 1200 1300 1400 1500 1600 1700

Even

ts /

( 3.9

5 G

eV )

0

50

100

150

200

250

Transverse Mass"aiA RooPlot of "l Transverse Mass"aiA RooPlot of "lµνγ eνγ

X2/NDoF = 216.273/154 = 1.40437

X2/NDoF = 200.218/134 = 1.49416

CB+Gauss

33


Spurious Signal Study 2

34

methodology

Compare the spurious signal yield to the background yield uncertainty★ bumps correspond to places where the background model doesn’t

describe perfectly the expected shape (→ Asimov dataset)

µνγ channel Asimovhyperbola

boundary effect

pdf overshoots data

pdf overshoots data

data overshoots pdf

data overshoots pdf

first point[broad signal peak!]


Systematic Uncertainties on Signal Normalization

35

All systematic uncertainties related to reconstruction effects that affect the signal normalization have been estimated

[trigger systematic missing in Wγ]


H→ZZ Spin-Parity


Higgs Boson Production and Decays

37


* “Production” angles:θ* of the first boson

* “Decay” angles:Φ is the angle between the two bosons decay planes in the Higgs rest frame

Φ1 is the angle between the first boson decay plane and the Higgs direction in the Higgs rest frame

θ1 and θ2 of the negative leptons in the corresponding boson rest frame

Angles Definitions

38

m1, m2, cosθ*, ϕ1, cosθ1, cosθ2, ϕ

production and decay angles definition in

H→ZZ(*)→4ℓ

The H→V1V2 process is fully characterized by the three masses involved and 5 of the 6 production/decay angles:


Higgs Spin-Parity: Analyses Method Outline

39

define a 1D discriminant which allows to separate between spin-parity hypotheses★ using masses, angles★ taking into account the detector/selection effects

(always) test one hypothesis (H0) against another one (H1)★ assuming that the spin-parity is 0+

★ testing against non-SM hypotheses: 0-, 1±, 2m±

★ spin-2: variating ggF/qq̅ production fraction using log-likelihood ratio to extract separations:

test for generic spin-2 impossible→graviton-like tensor with minimal couplings model

q = log

L(H0)

L(H1)

obs log(L(H0)/L(H1))

★ generating pseudo-experiments → qexp distribution

→ extracting p-values→ 1-CLs(H1) = 1- p(H1)/(1-p(H0))

p-value assuming H0

likelihood depending from µ and ε:

ε=0 → L(H0)ε=1 → L(H1)


PDF(m1, m2, cosθ*, φ1, cosθ1, cosθ2, Φ | m4l ) x P(m4l)∝H→ZZ(*)→4ℓ

40

J=0, ≥2f(fz1,fz2, cosθ*)

+ interference terms (for J≥2)

J=1, ≥2

J≥2

9 complex amplitudes (most

general case)

J=0 only 3 amplitudes → 4

independent parameters describe

the most general hZZ structure

Amplitudes = F(m1,m2, couplings, ΛNP)

http://prd.aps.org/pdf/PRD/v81/i7/e075022






Model: p.d.f.’s from the paper

41

J=0 J=1 J=2 ZZ*

cosθ*

ϕ1

red: J+mgreen: J+hblue: J-







Model: p.d.f.’s from the paper

42

J=0 J=1 J=2 ZZ*

cosθ1

cosθ2

ϕ


red: J+mgreen: J+hblue: J-





Spin-parity Observables

43


JP-MELA Discriminant Shape

44

★ JP-MELA discriminant defined as:

P(H0)

P(H0) + P(H1)JP-MELA = p(Hi) → probability density function of Hi hypothesis

including detector/selection effects[P = P(m4ℓ, m1, m2, cosθ*, ϕ1, cosθ1, cosθ2, ϕ)]

Hi = 0±, 1±, 2±m

0+ vs 0-

★ expected distributions computed using JHU MC (signal, ZZ background) and data (reducible background)

0+ vs 2+m


Signal Detector Effects Modeling

45

๏ Right-paired candidates‣ Z candidates mass smearing‣ 1D angular observables acceptance corrections‣ Study of acceptance correlations

๏ Wrong-paired candidates → same flavour/charge leptons from the two Z bosons switched

‣ Z candidates mass modeling‣ 1D angular observables PDFs‣ Study of correlations

PDFS = fRP PDFRP (m4l, pT , ⌘,mZ1, mZ2, ~⌦|g1, ..g10, fz0, ..fz2) · AccRP (~⌦)

+(1� fRP )PDFWP (m4l, pT , ⌘,mZ1, mZ2, ~⌦) acceptance correction

RP = right-paired candidatesWP = wrong-paired candidates

H

Z1

Z2

l1+

l2+

l1-

l2-

Z1reco

Z2reco

right-paired fraction from PowHeg MC


φ-3 -2 -1 0 1 2 30

0.01

0.02

0.03

0.04

0.05

0.06

truth

reco 4e + 4mu

reco 2e2mu + 2mu2e

Large effect in 4µ and 4e channels‣ Reconstruction effect: Z2 mass below cut‣ Physics effect: fermions interference?

Wrong-Paired Angular PDF Example: Φ

46

4µ/4e channels

mixed channelsPowHeg 0+: 0.184 ± 0.004

JHU 0+ : 0.171 ± 0.003 0+ JHUJHU 0- : 0.170 ± 0.003 0- JHUJHU 2+ : 0.109 ± 0.003 2+ JHUJHU 2- : 0.247 ± 0.003 2- JHU

systematic uncertainty


Separations in H→ZZ(*)→4ℓ

47

p0(0-)=0.0022

(exp 0.0011)

p0(0+)=0.40

CLs =99.7%

0+ vs 0-

p0(1+)=0.0028

(exp 0.0031)

p0(0+)=0.51

CLs =99.5%

0+ vs 1+

p0(2+)=0.113

(exp 0.064)p0(0+)=0.381

CLs =81.8%

0+ vs 2+

p0(2-)=0.11

(exp 0.003)p0(0+)=0.08

CLs =88.5%

0+ vs 2-

p0(1-)=0.027

(exp 0.001)p0(0+)=0.11

CLs =96.9%

0+ vs 1-

→results from both methods in good agreement→ excluding 0-, 1+, 1- at > 95% CL→ data prefers the 0+ hypothesis

0+ vs 0- [JP-MELA] 0+ vs 1+ [BDT]


Spin-parity Systematic Uncertainties

48

•  Systematic uncertainties to be considered are the same as the ones

included in the HCP/council measurement (since the method didn’t change) •  Two types of systematic uncertainties: of shape and of normalization •  Normalization systematics: -  luminosity -  Higgs mass shift → bin migration -  electron energy scale -  ZZ background normalization -  reducible background normalization •  Shape systematics: -  variation of wrong-paired candidates fraction -  electron energy scale

common to BDT and MELA

→ MELA only → common to BDT and MELA

C.Maiani 28.02.2014 séminaire LPNHE50

Heavy Higgs Exclusion Projections

[Djouadi and Quevillon, arXiv:1304.1787]

http://arxiv.org/abs/1304.1787

http://arxiv.org/abs/1304.1787

Direct and Indirect Searches of New Physics in Multi ... fileC.Maiani 28.02.2014 séminaire LPNHE Je...

Documents

Transcript of Direct and Indirect Searches of New Physics in Multi ... fileC.Maiani 28.02.2014 séminaire LPNHE Je...