Search for Z-Photon Resonances with ATLAS...1 bb LHC HIGGS XS WG 2011 ττ cc gg γγ Zγ WW ZZ 9 The H→Zγ Decay • Z boson is unstable and immediately decays in various ways -

Search for Z-photon resonances with the ATLAS detector

Nathan Readioff

Who I am…

2

• I’m from Wallasey, England • Undergraduate degree taken at the University of Liverpool

- Special four year course combining BSc and Masters- Masters project titled:

• Search for evidence of gauge-mediated supersymmetry breaking in the scenario where the neutrino is long-lived

• Followed by PhD at the University of Liverpool- Search for resonances decaying to a Z boson and a photon in

the ATLAS Detector using data from the Large Hadron Collider - Additional work on the ATLAS ITK upgrade project

Who I am…

2

• I’m from Wallasey, England • Undergraduate degree taken at the University of Liverpool

- Special four year course combining BSc and Masters- Masters project titled:

• Search for evidence of gauge-mediated supersymmetry breaking in the scenario where the neutrino is long-lived

• Followed by PhD at the University of Liverpool- Search for resonances decaying to a Z boson and a photon in

the ATLAS Detector using data from the Large Hadron Collider - Additional work on the ATLAS ITK upgrade project

Introduction

3

• Introduction to the Higgs boson

• Search for H→Zγ decay- First studies using 7 and 8 TeV data (2011+2012)- Analysis refinements - Z mass slices technique- Latest results using 13 TeV data (2015+2016)

• Search for new exotic resonances, X, decaying via X→Zγ- Initial search using 8 TeV data (2012)- First search with early 13 TeV data set (2015)- Latest results with full 13 TeV data set (2015+2016)

4

The Electroweak Force

• The Standard Model describes three out of four interactions in Nature- Electromagnetism- Strong Nuclear Force- Weak Nuclear Force

• Electromagnetic and weak forces are two facets of a single force- Above a unification energy, forces merge into single “electroweak” force- Sheldon Glashow, Abdus Salam and Steven Weinberg shared 1979 Nobel Prize

• The resultant Lagrangian has just one problem: no mass terms- The photon, W and Z vector bosons should all therefore be massless- Photon is massless, - W boson has mass 80.4 GeV- Z has mass 91.2 GeV

• Introduce a doublet of complex scalar fields

• Choose a wine-bottle potential

• The Lagrangian acquires the following terms:

5

�(x) =

✓�

+

�

0

◆=

1p2

✓�1 + i�2

�3 + i�4

◆

V (�) = µ2�†�� †�

�2

The Brout-Englert-Higgs Mechanism

• No terms relating to the photon field - massless photon!

• Terms in W and Z:

• Massive W and Z bosons!• Residual term:

• Scalar field has an associated massive boson!

L = (Dµ�)† (Dµ�)� V (�)

=1

2@µH@µH +

1

4g2

�H2 + 2vH + v2

�W+

µ W�µ

+1

8

�g2 + g02

� �H2 + 2vH + v2

�ZµZ

µ

+ µ2H2 +�

4

�H4 + 4vH3

�

M2WW+

µ W�µ =1

4g2v2W+

µ W�µ 1

2M2

ZZµZµ =

1

8

�g2 + g02

�v2ZµZ

µ

1

2M2

HH2 =

✓1

2µ2 � 3

2�v2

◆H2 =

✓1

2µ2 � 3

2µ2

◆H2 = �µ2H2

,

6

Production of the Higgs at the LHC

• LHC produces the Higgs through several mechanisms- Gluon-gluon fusion (ggF)- Vector boson fusion (VBF)- Production with associated vector boson (WH/ZH)- Production with associated tt ̅pair (ttH̅)

• Percentage mixture varies with √s and mH

• For √s = 8 TeV, mH = 125 GeV:- ggF dominates, with ~87% of the total- VBF is next, with ~7% of the total

g

g

H

g

g

t

H

t̄

q

q̄Õ

q

H

q̄Õ

q

q̄

H

W/Z

ggF VBF WH/ZH ttH̅

[GeV]HM100 120 140 160 180 200

Higg

s BR

+ T

otal

Unc

ert

-310

-210

-110

1

LHC

HIG

GS

XS W

G 2

011

bb

ττ

cc

gg

γγ γZ

WW

ZZ

7

Discovery of the Higgs Boson

• Higgs boson discovery announced in July 2012

[GeV]HM100 120 140 160 180 200

Higg

s BR

+ T

otal

Unc

ert

-310

-210

-110

1

LHC

HIG

GS

XS W

G 2

011

bb

ττ

cc

gg

γγ γZ

WW

ZZ

[GeV]4lm100 150 200 250

Even

ts/5

GeV

0

5

10

15

20

25

-1Ldt = 4.8 fb∫ = 7 TeV: s-1Ldt = 5.8 fb∫ = 8 TeV: s

4l→(*)ZZ→H

Data(*)Background ZZ

tBackground Z+jets, t=125 GeV)

HSignal (m

Syst.Unc.

ATLAS

7


• Higgs boson discovery announced in July 2012• Discovery made through study of the “golden channels”

- H→ZZ- H→γγ

[GeV]HM100 120 140 160 180 200

Higg

s BR

+ T

otal

Unc

ert

-310

-210

-110

1

LHC

HIG

GS

XS W

G 2

011

bb

ττ

cc

gg

γγ γZ

WW

ZZ

[GeV]4lm100 150 200 250

Even

ts/5

GeV

0

5

10

15

20

25


4l→(*)ZZ→H



HSignal (m

Syst.Unc.

ATLAS

7



- H→ZZ- H→γγ

Here I am!

[GeV]HM100 120 140 160 180 200

Higg

s BR

+ T

otal

Unc

ert

-310

-210

-110

1

LHC

HIG

GS

XS W

G 2

011

bb

ττ

cc

gg

γγ γZ

WW

ZZ

[GeV]4lm100 150 200 250

Even

ts/5

GeV

0

5

10

15

20

25


4l→(*)ZZ→H



HSignal (m

Syst.Unc.

ATLAS

7



- H→ZZ- H→γγ

• Channels significant because:- They have relatively low background- All final state particles can be fully reconstructed

Here I am!

H

f

f

f

Z

γ

H

W

W

W

Z

γ

H

W

Z

γ

8

The H→Zγ Decay

• H→Zγ decay process is similar to H→γγ- Similar branching fraction (Zγ: 1.54x10-3, γγ: 2.3x10-3)- Shared set of leading order-Feynman diagrams

• Decay proceeds through loops:

• W loops, and top-quark dominated fermion loops• Modifications to BR(H→Zγ) expected if:

- H is a non-Higgs neutral scalar- H is a composite state- H couples to new particles exchanged in loops:

• colourless charged scalars, leptons, vector bosons

• BR(H→Zγ) and BR(H→γγ) could be affected differently• Measurement of BR(H→Zγ)/BR(H→γγ) could provide

valuable insights into BSM models

[GeV]HM100 120 140 160 180 200

Higg

s BR

+ T

otal

Unc

ert

-310

-210

-110

1

LHC

HIG

GS

XS W

G 2

011

bb

ττ

cc

gg

γγ γZ

WW

ZZ

9

The H→Zγ Decay

• Z boson is unstable and immediately decays in various ways- Hadronic decays lack sensitivity at low mass- Invisible decays are beyond the scope of these studies- Focus given to leptonic decays

• Only Z→ee or Z→μμ decays are studied- Z→ττ is also possible, but τ leptons principally decay hadronically

9

The H→Zγ Decay



• Choose events with final state of eeγ or μμγ

9

The H→Zγ Decay



• Choose events with final state of eeγ or μμγ• The challenge:

- BR(Z→ee + Z→μμ) = 6.6%

9

The H→Zγ Decay




- BR(Z→ee + Z→μμ) = 6.6%• Available statistics dramatically reduced

9

The H→Zγ Decay




- BR(Z→ee + Z→μμ) = 6.6%• Available statistics dramatically reduced

Object Identification and ReconstructionRequire good quality, isolated leptons and photons

Reconstruct Z from opposite sign, same flavour lepton pairs

Require mll > mZ - 10 GeVMerge Z candidate with highest-pT photon

10

The Analysis

• Search is performed for excess events in the invariant mass spectrum, mllγ

• Mass spectrum fitted with a combined Signal+Background model• Signal shape obtained from simulated Monte-Carlo samples• Background shape obtained by fitting a function directly to data

- Choice of function from studies with simulated Monte-Carlo background samples- Normalisation from data

• Signal-to-Background ratio is very small (<few %)• Sensitivity may be enhanced by improving:

- Invariant mass resolution- Signal-to-Background ratio

Object Identification and ReconstructionRequire good quality, isolated leptons and photons

Reconstruct Z from opposite sign, same flavour lepton pairs

Require mll > mZ - 10 GeVMerge Z candidate with highest-pT photon

11

Improving resolution: Event Corrections

• Photon origin set to the Primary Vertex- Photon origin only known from extrapolation in EM calorimeter, but must come from PV- Photon η and ET are recalculated

Lepton

Lepton

Z

Photon

11



Lepton

Lepton

Z

Photon

11



• Leptons can radiate photons - final state radiation- Negligible for electrons, where photon and electron electromagnetic showers overlap - Represents loss of energy in muons, and shifts mll distribution to lower energies- Photons identified within ΔR < 0.08 or ΔR < 0.15 (based on ET) included in mll

11




• Z mass constraint- Vary lepton 4-momenta parameters within uncertainties to improve Mll

- Process recovers momentum lost due to bremsstrahlung- Brings signal peak closer to true Higgs generated mass

11




• Z mass constraint- Vary lepton 4-momenta parameters within uncertainties to improve Mll

- Process recovers momentum lost due to bremsstrahlung- Brings signal peak closer to true Higgs generated mass

[GeV]γeem105 110 115 120 125 130 135 140 145

]-1

[GeV

γee

1/N

dN

/dm

00.020.040.060.08

0.10.120.140.160.180.2

0.220.24

γeemγeeCorrected m

[GeV]γµµm105 110 115 120 125 130 135 140 145

]-1

[GeV

γµ

µ1/

N d

N/d

m

00.020.040.060.08

0.10.120.140.160.180.2

0.220.24

γµµmγµµCorrected m

12

Improving Signal-to-Background Ratio: Event Categorisation

eeγ

μμγ

High PTt

Low PTt

High PTt

Low PTt

√s = 7TeV

eeγ

μμγ

High PTt

Low PTt

High PTt

Low PTt

√s = 8TeV

High ΔηZγ

Low ΔηZγ

High ΔηZγ

Low ΔηZγ

thrust axis

pTZγp

Tt

pTl

pTγp

TZ

Higgs pTt is component of Higgs momentum orthogonal to difference between Z and γ momenta

• Events are separated into orthogonal categories• Each category will have its own:

- signal resolution- signal-to-background ratio

• Events are filtered into ten categories according to:- Centre-of-mass collision energy- Final state- Higgs pTt (> or < 30GeV)- ΔηZγ (> or < 2.0)

12

Improving Signal-to-Background Ratio: Event Categorisation

• Events are separated into orthogonal categories• Each category will have its own:

- signal resolution- signal-to-background ratio

• Events are filtered into ten categories according to:- Centre-of-mass collision energy- Final state- Higgs pTt (> or < 30GeV)- ΔηZγ (> or < 2.0) Tt

Cut on p0 10 20 30 40 50 60

η ∆

Cut

on

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0.14

0.145

0.15

0.155

0.16

0.165

0.17

γZη∆

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

γZη∆

1/N

dN

/d

00.010.020.030.040.050.060.070.080.09

= 125 GeVH

ggF, m

= 125 GeVH

VBF, m

= 125 GeVH

WH, m

= 125 GeVH

ZH, m

= 125 GeVH

ttH, m

MCγZ+

Data

= 8 TeVs ee, →, Z γ Z→H

[GeV]tT

p0 20 40 60 80 100 120 140

[1/3

GeV

]t T

1/N

dN

/dp

4−10

3−10

2−10

1−10

= 125 GeVH

ggF, m

= 125 GeVH

VBF, m

= 125 GeVH

WH, m

= 125 GeVH

ZH, m

= 125 GeVH

ttH, m

MCγZ+

Data

= 8 TeVs ee, →, Z γ Z→H

13

VBF Categorisation

• Potential category for VBF events• Events characterised by two forward jets• Multiple jet and dijet variables considered• Final variables:

- |Δηjj|- |Δφjj,H|- mjj

• Background taken from 5GeV data sidebands• Signal from VBF MC• Multivariate analysis performed to determine optimal cuts

• Ultimately too few events for selection to be practical

q

q̄Õ

q

H

q̄Õ

jjη∆

0 1 2 3 4 5 6 7 8 9 10

jjη∆

dN/d

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0.045Data Sidebands

VBF Signal MC

S+BSSBR:

H,jjφ∆

0 1 2 3 4 5 6 7 8 9 10

H,jj

φ∆

dN/d

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035 Data Sidebands

VBF Signal MC

S+BSSBR:

[GeV]jjm0 200 400 600 800 1000 1200 1400

[1/1

0 G

eV]

jjdN

/dm

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04Data Sidebands

VBF Signal MC

S+BSSBR:

|Δηjj| |Δφjj,H| mjj

Background reduced by factor 100 for visibility

• MC signal samples generated for 7 mass points- mH = 120, 125, 130, 135, 140, 145, 150 GeV

• Signal in each category modelled by summing Crystal Ball and Gauss functions

• Crystal Ball function is a Gaussian distribution with a power law tail- Used to capture the core of the distribution

• Gaussian function used to capture the tail of the distribution

• Model parameters vary linearly with mass point, all mass points fitted simultaneously• Model can then be generated for any arbitrary mass point

14

Signal Modelling

8 TeV, ggF, eeγ High pTt

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

0.5

GeV

0

50

100

150

200

250

300

350

400

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

0.5

GeV

0

20

40

60

80

100

120

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

0.5

GeV

0

100

200

300

400

500

600

GA (x, µ,�) = N exp

� (x� µ)

2

2�

2

!

CB (x, µ,�,↵, n) = N ·

8<

:e

�t2/2 if t > �↵⇣n|↵|

⌘n· e�|↵|2/2 ·

⇣n|↵| � |↵|� t

⌘�nif t �↵

8 TeV, ggF, eeγ Low pTt, High ΔηZγ 8 TeV, ggF, eeγ Low pTt, Low ΔηZγ

15

Background Modelling

8 TeV, ggF, eeγ High pTt 8 TeV, ggF, eeγ Low pTt, High ΔηZγ 8 TeV, ggF, eeγ Low pTt, Low ΔηZγ

Even

ts /

1 G

eV

0

10

20

30

40

50

ee→, Z γ Z→H

1118 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd

20−15−10−5−05

101520

Even

ts /

1 G

eV

0

10

20

30

40

50

60

70

80

ee→, Z γ Z→H

1926 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd

30−20−10−0

102030

Even

ts /

1 G

eV

020406080

100120140160180200220

ee→, Z γ Z→H

4699 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd

40−30−20−10−010203040

[GeV]Hm120 130 140 150 160 170

Bsp

urio

us-s

igna

l/

-0.4

-0.2

0

0.2

0.4

0.6

exp (115-170 GeV)

exppol2 (115-170 GeV)

exppol3 (115-170 GeV)

pol2 (115-170 GeV)

pol3 (115-170 GeV)

pol4 (115-170 GeV)

pol5 (115-170 GeV)

pol6 (115-170 GeV)

-1Ldt = 20.3 fb∫ = 8 TeV, see→,ZγZ→H

LowpTtHigdelEta

• Background is dominated by non-resonant Z+γ events • Small contribution from Z+Jet events

- Events where jet is misreconstructed as a photon

• Background modelled by fitting analytical function to data

• Choice of model influenced by level of spurious signal:- Bias on signal yield from background model choice- Ratio of fitted signal yield to its uncertainty must be < 20%

• Several functions used to model background in different categories:- 4th- or 5th-order Chebychev polynomials, - Simple exponentials, - Exponentials of polynomials

[GeV]Hm120 125 130 135 140 145 150

)γZ→

(HSMσ)/γZ

→(Hσ

95%

CL

limit

on

0

5

10

15

20

25

30ObservedExpected

σ 1±σ 2±

= 7 TeVs, -1 Ldt = 4.5 fb∫ = 8 TeVs , -1 Ldt = 20.3 fb∫

ATLAS

[GeV]Hm120 125 130 135 140 145 150

0Lo

cal p

-210

-110

1

10

0 observed pγ Z→H

0 expected pγ Z→SM H

= 8 TeVsData 2012, -1Ldt = 20.3 fb∫ = 7 TeVsData 2011, -1Ldt = 4.5 fb∫

ATLASσ = 1.6 maxZ = 142.0 GeVHm

σ1

σ2

16

Expected Limits

• Paper was published in March 2014• Phys. Lett. B 732C (2014), pp. 8-27

• No significant excesses identified• Largest excess was 1.6σ at mH = 141 GeV

• Limits on cross-section times branching ratio at 95% CL- √s = 8 TeV : 0.13 to 0.5 pb

• Limits set on the relative signal strength- Divide cross-section by SM expectation- Expected limits vary from 5.1 to 15.8 x SM- Observed limits vary from 3.1 to 17.6 x SM

- At 125.5 GeV:• Expected limit = 9.0 x SM• Observed Limit = 11.0 x SM

17

Improving The Limits: New Event Categorisation

• LHC shut down during 2013-2015• Opportunity to further investigate Run 1 data

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

0

0.001

0.002

0.003

0.004

0.005

0.006

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m85

90

95

100

105

110

8TeV, ggF, mH = 125 GeV, eeγ 8TeV, ggF, mH = 125 GeV, μμγ

17


• LHC shut down during 2013-2015• Opportunity to further investigate Run 1 data• Strong correlation observed between reconstructed Z and Higgs boson masses• Events concentrated at Z and Higgs mass poles

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

0

0.001

0.002

0.003

0.004

0.005

0.006

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m85

90

95

100

105

110


17


• LHC shut down during 2013-2015• Opportunity to further investigate Run 1 data• Strong correlation observed between reconstructed Z and Higgs boson masses• Events concentrated at Z and Higgs mass poles• Exploit this by adding categories for given Z mass bins

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

0

0.001

0.002

0.003

0.004

0.005

0.006

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m85

90

95

100

105

110


0

0.002

0.004

0.006

0.008

0.01

) [GeV]γM(ll116 118 120 122 124 126 128 130 132 134

M(ll

) [G

eV]

85

90

95

100

105

110ATLAS Simulation Work In Progress

18

Z Mass Slices

• Each Z slice has a different mass resolution• Slices nearer the true Z Mass have better resolution, better signal-to-background ratio

- Such slices contribute more to the expected limit- Z slices offer a way of improving sensitivity

• Standard analysis sees all slices merged together- Higgs peak smeared out, poorer resolution

7TeV, ggF, mH = 125 GeV, eeγ 7TeV, mH = 125 GeV, eeγ [GeV]γllm

116 118 120 122 124 126 128 130 132 134 [1

/0.5

GeV

]γll

dN/d

m0

0.002

0.004

0.006

0.008

0.01

0.012

0.014 < 86 GeVll85 < m < 91 GeVll90 < m < 95 GeVll94 < m

Z Mass Slices: Analysis overview

• Analysis follows the same procedure as before

• Several corrections not applied:- Impact of FSR was considered negligible for the purposes of this study- Z Mass constraint removed

• Z-slices should perform the same job• Constraint moves event to peak region, loss of statistics in tail slices• Constraint makes correlations between mll and mllγ more complicated

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

8TeV, mH = 125 GeV, eeγ 8TeV, mH = 125 GeV, eeγ, Z Mass Constraint

0

0.0005

0.001

0.0015

0.002

0.00250.003

0.0035

0.004

0.0045

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

19

eeγ

μμγ

eeγ

μμγ

√s = 7 TeV

√s = 8 TeV

89<mll<90

90<mll<91

87<mll<89

85<mll<87

91<mll<92

92<mll<93

93<mll<95

Z Mass Slices: First Test

• Events categorised based on:- Centre-of-mass collision energy- Final state- Z Mass

• Note tails widen to 2GeV- Boost statistics in tails of mll distribution

20

eeγ

μμγ

eeγ

μμγ

√s = 7 TeV

√s = 8 TeV

89<mll<90

90<mll<91

87<mll<89

85<mll<87

91<mll<92

92<mll<93

93<mll<95

Z Mass Slices: First Test

• Events categorised based on:- Centre-of-mass collision energy- Final state- Z Mass

• Note tails widen to 2GeV- Boost statistics in tails of mll distribution

[GeV]Hm120 125 130 135 140 145 150

)γZ→

(HSM

σ)/γZ→

(Hσ

95%

CL

limit

on

02468

101214161820 Z Slice Observed

Z Slice Expectedσ 1±Z Slice σ 2±Z Slice

Kinematic ObservedKinematic Expected

σ 1±Kinematic σ 2±Kinematic


• Expected/Observed limits evaluated• Expected limits comparable to kinematic

analysis• Observed limit enhanced at mH = 125 GeV• No real improvement though

20

Z Mass Slices: Final Event Categorisation

21

• Events categorised based on:- Centre-of-mass collision energy- Final state- Higgs pTt (> or < 30 GeV)- Z Mass

• 7 TeV has too few statistics to be split further• 8 TeV cannot be further split by ΔηZγ

High PtT

Low PtT

High PtT

Low PtT

eeγ

μμγ

eeγ

μμγ

√s = 7 TeV

√s = 8 TeV

89<mll<90

90<mll<91

87<mll<89

85<mll<87

91<mll<92

92<mll<93

93<mll<95

Z Mass Slices: Signal & Background Models

22

• Signal model remains a Crystal Ball (Core) + Gauss (tails)- Parameters vary linearly with nominal mass- All mass points fitted simultaneously

• Due to low statistics, background modelled as single exponential- More complex functions would be too distorted by local fluctuations

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

1 G

eV

0

20

40

60

80

100

120

140

160

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

1 G

eV

020406080

100120140160180200220

[GeV]γllm100 110 120 130 140 150 160 170

Even

ts /

1 G

eV

0

20

40

60

80

100

120

140

Even

ts /

1 G

eV

0

5

10

15

20

25

30

35

40

45

ee→, Z γ Z→H

588 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd

15−10−5−05

1015

Even

ts /

1 G

eV

0

5

10

15

20

25

30

35

40

ee→, Z γ Z→H

745 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd

15−10−5−05

1015

Even

ts /

1 G

eV

0

5

10

15

20

25

30

35

40

ee→, Z γ Z→H

631 Events

Data 2012 50)× SMσ = 125 GeV,

H (mγ Z→H

[GeV]γllm120 130 140 150 160 170

Dat

a - B

gd15−10−5−05

1015

8TeV eeγ 85<Z<87 Low Ptt 8TeV eeγ 90<Z<91 Low Ptt 8TeV eeγ 93<Z<95 Low Ptt

Z Mass Slices: Expected & Observed Limits

23

[GeV]Hm120 125 130 135 140 145 150

)γZ→

(HSM

σ)/γZ→

(Hσ

95%

CL

limit

on

02468

101214161820 Z Slice & Kinematic Observed

Z Slice & Kinematic Expectedσ 1±Z Slice & Kinematic σ 2±Z Slice & Kinematic

Kinematic ObservedKinematic Expected

σ 1±Kinematic σ 2±Kinematic

= 7 TeVs, -1 Ldt = 4.5 fb∫ = 8 TeVs , -1 Ldt = 20.3 fb∫• Expected/Observed limits evaluated

• No systematic uncertainties re-evaluated

• At mH = 125 GeV:- Expected (Kinematic) = 9.01*SM- Expected (Z Slices) = 8.05*SM

• Z slices reduce limits by ~10%

Z Mass Slices: Possible refinements

24

• The technique introduces multiple categories and associated systematic uncertainties• Final model contains many parameters• Signal and background could possibly be modelled using 2D functions…

0

0.0005

0.001

0.0015

0.002

0.0025

0.003

0.0035

0.004

[GeV]γllm116 118 120 122 124 126 128 130 132 134

[GeV

]ll

m

85

90

95

100

105

110

Search for X→Zγ Decays

25

• LHC restarted June 2015• First studies using 3.2 fb-1 of 13 TeV data completed by December 2015


25

• LHC restarted June 2015• First studies using 3.2 fb-1 of 13 TeV data completed by December 2015• Diphoton channel reported a slight excess at 750 GeV

- Could have been a statistical fluctuation- Could have been a new particle

• Similar excess reported by CMS

[GeV]γγm200 400 600 800 1000 1200 1400 1600

Even

ts /

40 G

eV

1−10

1

10

210

310

410ATLAS Preliminary

-1 = 13 TeV, 3.2 fbs

Data

Background-only fit

[GeV]γγm200 400 600 800 1000 1200 1400 1600

Dat

a - f

itted

bac

kgro

und

15−10−

5−05

1015

ATLAS-CONF-2015-081

[GeV]Xm

200 400 600 800 1000 1200 1400 1600 1800

Loca

l p-v

alue

5−10

4−10

3−10

2−10

1−10

1

ATLAS Preliminary-1 = 13 TeV, 3.2 fbs

Observed

σ0

σ1

σ2

σ3

σ4

ATLAS-CONF-2015-081


25

• LHC restarted June 2015• First studies using 3.2 fb-1 of 13 TeV data completed by December 2015• Diphoton channel reported a slight excess at 750 GeV

- Could have been a statistical fluctuation- Could have been a new particle

• Similar excess reported by CMS• Particles coupling to the photon expected to couple to the Z boson

- A boson X undergoing X→γγ should also decay via X→Zγ

[GeV]γγm200 400 600 800 1000 1200 1400 1600

Even

ts /

40 G

eV

1−10

1

10

210

310

410ATLAS Preliminary

-1 = 13 TeV, 3.2 fbs

Data

Background-only fit

[GeV]γγm200 400 600 800 1000 1200 1400 1600

Dat

a - f

itted

bac

kgro

und

15−10−

5−05

1015

ATLAS-CONF-2015-081

[GeV]Xm

200 400 600 800 1000 1200 1400 1600 1800

Loca

l p-v

alue

5−10

4−10

3−10

2−10

1−10

1

ATLAS Preliminary-1 = 13 TeV, 3.2 fbs

Observed

σ0

σ1

σ2

σ3

σ4

ATLAS-CONF-2015-081

[GeV]φm200 400 600 800 1000 1200 1400 1600

) [fb

]γ)- l+

Z(l

→ φ B

R(

×fidσ

-310

-210

-110

1

10

=8 TeVs, -1 L dt = 20.3 fb∫ATLAS

γ-l+ l→pp

Observed 95% CL upper limitExpected 95% CL upper limit

σ 1 ±Expected σ 2 ±Expected

, parameter set (a) γ) -l+ Z(l→ φ, parameter set (b)γ) -l+ Z(l→ φ

Three-body invariant mass distribution for the e+e-γ final state. The expected signal for a resonance mass of 500 GeV is superimposed.

PLB 738 (2014) 428-447 PLB 738 (2014) 428-447

Run 1 results

26

• An exotics search for high mass resonances decaying to Zγ was performed• Run 1 analysis published in PLB 738 (2014) 428-447• Search for singlet scalar resonance decaying to Zγ, Z→ll (where l = e or μ)

- Used 20.3 fb-1 of 8 TeV data- 1σ excess at 700 GeV down to 0σ at 750 GeV

• Trivial matter to extend Run 2 SM H→Zγ software to search for high mass resonances

http://www.sciencedirect.com/science/article/pii/S0370269314007254



[GeV]γllm520 540 560 580 600 620 640 660 680

Proj

ectio

n of

sig

nal p

df

1

10

210

310

LoαCBσ

Power LawLo-n~m

HiαCBσ

Power LawLo-n~m

CBσ

Gaussian Distribution

X m∆

X→Zγ: Signal Modelling

27

• Signal MC samples generated using narrow width approximation- Higgs signal peak has width ~4.0MeV

• Mass points at: 200, 300, 500, 700, 750, 800, 1000, 1500 GeV• Signal modelled with a double-sided Crystal Ball function

N ·

8>>>><

>>>>:

e�t2/2 if �↵Lo t ↵Hi

e�0.5↵2Loh

↵LonLo

⇣nLo↵Lo

�↵Lo�t⌘inLo if t < �↵Lo

e�0.5↵2Hih

↵HinHi

⇣nHi↵Hi

�↵Hi+t⌘inHi if t > ↵Hi,


27



- All mass points fitted simultaneously- All parameters vary as a function of mass points

N ·

8>>>><

>>>>:


e�0.5↵2Loh

↵LonLo

⇣nLo↵Lo


e�0.5↵2Hih

↵HinHi

⇣nHi↵Hi


Parameter dependence

Par Mass DependenceμCB aμ + bμ*x + cμ*x*x+mX

σCB aσ + bσ * xαlo aαlo + bαlo /(x+cαlo)αhi aαhi + bαhi /(x+cαhi)

X (mX-100)/100


27




N ·

8>>>><

>>>>:


e�0.5↵2Loh

↵LonLo

⇣nLo↵Lo


e�0.5↵2Hih

↵HinHi

⇣nHi↵Hi





X (mX-100)/100

[GeV]γllm700 720 740 760 780 800

Even

ts /

1 G

eV

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

= 749.4CBµ

= 6.0CBσ = 0.8Loα = 1.1Hiα = 9.0Lon = 5.0Hin


27




• Signal efficiency described by an exponential function

N ·

8>>>><

>>>>:


e�0.5↵2Loh

↵LonLo

⇣nLo↵Lo


e�0.5↵2Hih

↵HinHi

⇣nHi↵Hi





X (mX-100)/100

Sign

al e

fficie

ncy

0.3

0.32

0.34

0.36

0.38

0.4

0.42

0.44 ATLAS Simulation Work in Progress-1Ldt = 1.00 fb∫ = 13 TeV: s

Inclusive

(Mass-100)/100 [GeV]2 4 6 8 10 12 14

MC

/ Fi

t

0.95

1

1.05

a+ b · e(c·mX)

[GeV]γllm700 720 740 760 780 800

Even

ts /

1 G

eV

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

= 749.4CBµ

= 6.0CBσ = 0.8Loα = 1.1Hiα = 9.0Lon = 5.0Hin

• Background determined by fitting function directly to data

• Selected background model used in other searches:- multi-jet, photon+jet and diphoton

• Class of functions have form:

• where:

• F-Test:- Identify simplest model for which extra

parameters do not significantly improve fit quality

• Spurious signal- Bias on signal yield from background model

choice- Ratio of fitted signal yield to its uncertainty

must be < 20%

• The simplest function is:

• Spurious signal treated as systematic uncertainty• Parameterised by the envelope of spurious signal at

each mass point

[GeV]γZM200 400 600 800 1000 1200 1400 1600

SδS/

0.3−

0.2−

0.1−

0

0.1

0.2

0.3

DijetDijet1Dijet2

X→Zγ: Background Modelling

28

f

k;d

(x; b, {ak

}) = (1� x

d)bxPk

j=0 aj log(x)

j

,

x =m``�p

s

fk=0;d=1/3(x; b, d, {ak}) = (1� x

1/3)bxa0

• Background determined by fitting function directly to data

• Selected background model used in other searches:- multi-jet, photon+jet and diphoton

• Class of functions have form:

• where:

• F-Test:- Identify simplest model for which extra

parameters do not significantly improve fit quality

• Spurious signal- Bias on signal yield from background model

choice- Ratio of fitted signal yield to its uncertainty

must be < 20%

• The simplest function is:

• Spurious signal treated as systematic uncertainty• Parameterised by the envelope of spurious signal at

each mass point

[GeV]γZM200 400 600 800 1000 1200 1400 1600

SδS/

0.3−

0.2−

0.1−

0

0.1

0.2

0.3

DijetDijet1Dijet2

Even

ts /

( 10

)

3−10

2−10

1−10

1

10

210

310p0 = 5.766 +/- 9.230p1 = -3.3680 +/- 1.354

Data 2015 (382 events)µµ→ee + Z→, Z-1Ldt = 3.2 fb∫ = 13 TeV, s

[GeV]γllm200 400 600 800 1000 1200 1400 1600

Dat

a - B

kg

20−

0

20

X→Zγ: Background Modelling

28

f

k;d

(x; b, {ak

}) = (1� x

d)bxPk

j=0 aj log(x)

j

,

x =m``�p

s

fk=0;d=1/3(x; b, d, {ak}) = (1� x

1/3)bxa0

X→Zγ: Results

29

• Four candidate M(llγ) events with mass > 700 GeV

• Limits set on production cross-section (pp→X) x branching ratio (X→Zγ)- Expected limits range from 24.3 fb to 230 fb- Observed Limits range from 22.9 fb to 296 fb- Largest deviation from background-only hypothesis is ~2σ at ~350 GeV

• Analysis is dominated by statistical uncertainties• Largest systematic uncertainty is on luminosity (±5%)

[GeV]Hm200 400 600 800 1000 1200 1400 1600

) [fb

]γZ

→Br

(X×

X)→

(pp

σ95

% C

L lim

it on

0

50

100

150

200

250

300

350

400ObservedExpected

σ 1±σ 2±

= 13 TeVs, -1 Ldt = 3.2 fb∫

[GeV]Hm400 600 800 1000 1200 1400

0p

2−10

1−10

1

10

= 13 TeVs, -1 Ldt = 3.2 fb∫ 0Observed p

0Expected p

σ1

σ2

X→Zγ: Hadronic Z Decays

30

• Search looks for X→Zγ, Z→qq ̅• Search performed using 3.2fb-1 of 13 TeV data

- Published in PLB• Boosted quark pair merge into single large radius jet• Largest excess is 1.8σ for mX = 1.9 TeV

Even

ts /

20 G

eV

2−10

1−10

1

10

210

310ATLAS qq→, ZγZ→X→pp

-1=13 TeV, 3.2 fbsData

bBackground-only fit,

[GeV]γJm1000 1500 2000 2500 3000

Dat

a - b

10−5−05

10

[GeV]Xm500 1000 1500 2000 2500 3000

) [fb

]γZ

→ B

R(X

×X)

→

(pp

σ95

% C

L lim

it on

10

210

ObservedExpectedσ 1±σ 2±

ATLAS

qq→, ZγZ→X→pp-1=13 TeV, 3.2 fbs

[GeV]Xm500 1000 1500 2000 2500 3000

Loca

l p-v

alue

3−10

2−10

1−10

1σ0

σ1

σ2

σ3

ATLAS


http://dx.doi.org/10.1016/j.physletb.2016.11.005

X→Zγ: Hadronic Z Decays

30

• Search looks for X→Zγ, Z→qq ̅• Search performed using 3.2fb-1 of 13 TeV data

- Published in PLB• Boosted quark pair merge into single large radius jet• Largest excess is 1.8σ for mX = 1.9 TeV• Hadronic analysis better than leptonic for mX>1.5TeV

- Boosted jets merge into single large radius jet- More statistics in this region

Even

ts /

20 G

eV

2−10

1−10

1

10

210

310ATLAS qq→, ZγZ→X→pp

-1=13 TeV, 3.2 fbsData

bBackground-only fit,

[GeV]γJm1000 1500 2000 2500 3000

Dat

a - b

10−5−05

10

[GeV]Xm500 1000 1500 2000 2500 3000

) [fb

]γZ

→ B

R(X

×X)

→

(pp

σ95

% C

L lim

it on

10

210

ObservedExpectedσ 1±σ 2±

ATLAS


[GeV]Xm500 1000 1500 2000 2500 3000

Loca

l p-v

alue

3−10

2−10

1−10

1σ0

σ1

σ2

σ3

ATLAS


[GeV]Xm500 1000 1500 2000 2500 3000

) [fb

]γZ

→ B

R(X

× X

)→

(pp

σ95

% C

L lim

it on

10

210

310ObservedExpectedσ 1±

σ 2±

ATLASγZ→X→pp

-1=13 TeV, 3.2 fbs

Leptonic

HadronicCombined

http://dx.doi.org/10.1016/j.physletb.2016.11.005

After the PhD…

31

• Search for X→Zγ was refined using additional 13 TeV data recorded in 2016- Search using 13.3 fb-1 of 13 TeV data presented at ICHEP 2016- Diphoton excess at 750 GeV had disappeared- X→Zγ found no significant excesses

• Results supplanted by a new paper (currently being submitted to JHEP)• Analysis utilises 33.1 fb-1 of 13 TeV data• Paper features:

- Updated search for X→Zγ, Z→ll decays- First look at H→Zγ, Z→ll in 13 TeV data

Low Mass Search: Improvements on Run1

32

• Search for Standard Model Higgs boson• Event categorisation re-optimised for higher statistics• Categories based on pTt were used in Run 1

- Cuts reoptimised for new data set

• New category for ptγ/mllγ - Introduced for high mass search- Variable stable over wide range of mllγ values

• Optimal cuts identified from 2D scan of variables- ptγ/mllγ = 0.4, pTt = 40 GeV

• New category gives 14% boost in sensitivity

thrust axis

pTZγp

Tt

pTl

pTγp

TZ

Higgs pTt is component of mZγ momentum orthogonal to difference between Z and γ momenta

Selected Events

< 0.82

I II III IV

< 0.4

μμee

> 0.82

> 0.4

V VI

pTt pTt

Lepton Flavour

ptγ/mllγ

VBF BDT

> 40GeV < 40GeV > 40GeV < 40GeV

pTt = 2|pZ

x

p�y

� p�x

pZy

|pZ��

t

[GeV]γ

Tp

0 100 200 300 400 500 600 700 800 900 1000

arb.

uni

ts

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7Data

= 300 GeVX

Signal MC: m = 700 GeV

XSignal MC: m

= 1000 GeVX

Signal MC: m

[GeV]γ

Tp

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

arb.

uni

ts

00.020.040.060.08

0.10.120.140.160.180.2

0.22Data

= 300 GeVX

Signal MC: m = 700 GeV

XSignal MC: m

= 1000 GeVX

Signal MC: m

ptγ

ptγ/mllγ

BDT output1− 0.5− 0 0.5 1

1/N

dN

/d(B

DT

outp

ut)

00.5

11.5

22.5

33.5

44.5

5DataγZ+

Z+jets = 125 GeVHVBF m = 125 GeV

HggF m

PreliminaryATLAS-1 = 13 TeV, 36.1 fbs

< 170 GeVγZ 2, 115 GeV < m≥ jetsN

Low Mass Search: New VBF Category

33

• VBF category not possible using Run 1 data• VBF production characterised by two forward jets

- Large di-jet invariant mass, - High η separation of jets

• Main background- Zγ pair production with at least two jets- Single Z events with at least three jets and one fake photon

• Boosted decision tree combines six correlated variables• Significance maximum for BDT response > 0.82• Category increases sensitivity by 7.7%

q

q̄Õ

q

H

q̄Õ

Variables Mass Dependence

mjj Invariant mass of dietΔηjj Pseudo-rapidty separation of dietΔφZγ,jj Azimuthal angle between Zγ and diet system

PTt Zγ Pt projected perpendicular to Zγ thrust axis

ΔRγ or Z, jMinimum ΔR between one object of the Zγ and

the corresponding leading jetsηZeppenfield |ηZγ - 0.5(ηj1 + ηj2)|

[GeV]γll m

Even

ts /

GeV

0100200300400500600700800

DataBackground fit

20×Signal = 125 GeVHm

PreliminaryATLAS-1=13 TeV, 36.1 fbs

Ttee low p

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

-500

50 [GeV]γll m

Even

ts /

GeV

0

20

40

60

80

100 DataBackground fit



Tt high pµµ

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

-100

10 [GeV]γll m

Even

ts /

GeV

0

200

400

600

800

1000DataBackground fit



Tt low pµµ

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

50−0

50

Low Mass Search: Signal and Background Modelling

34

• All signal samples modelled using a double-sided Crystal Ball function• Test performed to identify background model with minimum “spurious signal”

- Ratio of fitted signal yield to its uncertainty must be < 45%• Background in each category modelled as Bernstein 2nd- or 4th- order polynomial

[GeV]γll m

Even

ts /

GeV

0

2

4

6

8

10

12




VBF-enriched

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

-505 [GeV]γll m

Even

ts /

GeV

0

10

20

30

40




Thigh relative p

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

-100

10 [GeV]γll m

Even

ts /

GeV

01020304050607080 Data

Background fit 20×Signal

= 125 GeVHm


Ttee high p

[GeV]γZm115 120 125 130 135 140 145 150D

ata

- Fit

-200

20

High Mass Search: Signal and Background Modelling

35

• All signal samples modelled using a double-sided Crystal Ball function- Samples generated for CP-even Spin-0 and Spin-2 resonances produced via ggF and VBF

• Background model unchanged from previous analysis• One event category per final state

- Exploits difference between eeγ and μμγ invariant mass resolutions

• Signal resolution varies from:- eeγ : 2.8 GeV (mX = 200 GeV) to 16 GeV (mX = 2.5 TeV)- μμγ : 3.1 GeV (mX = 200 GeV) to 36 GeV (mX = 2.5 TeV)

• Loss of precision in high-pT muon reconstruction

[GeV]γZm1400 1500 1600

]-1 [G

eVγZ

1/N

dN

/dm

0

0.01

0.02

0.03

0.04

0.05 Simulation PreliminaryATLASγ Z→ X →gg

= 1500 GeV Xm = 13 TeVs

-e+ e→Z-µ+µ →Z

[GeV]γZm280 300 320

]-1 [G

eVγZ

1/N

dN

/dm

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

0.18 Simulation PreliminaryATLAS

γ Z→ X →gg = 300 GeV Xm = 13 TeVs

-e+ e→Z-µ+µ →Z

ICHEP 2016 analysis

Even

ts /

20 G

eV

2−10

1−10

1

10

210

310

410

DataBackground fit

-1 = 13 TeV, 36.1 fbs PreliminaryATLAS

[GeV]γZm210×3 310 310×2Si

gnifi

canc

e

4−2−024

Latest Results

36

• No significant excess at low mass- At mH = 125.09 GeV:

• Expected limit at 95% CL = 5.2 x SM• Observed limit at 95% CL = 6.6 x SM

- 55% gain in exclusion sensitivity over Run 1 result- 20% gain through analysis optimisations

Latest Results

36

[GeV]X m

210×3 310 310×2 310×3

BR

[fb

]×

σ 9

5%

CL U

pper

Lim

it on

1

10

210

310

Observed

Expected

1 std. dev.±

2 std. dev.±


γ Z→ X→gg

= 4 MeVXΓ = 0, XJ

[GeV]Xm

500 1000 1500 2000 2500

0Loca

l p

-310

-210

-110

1

10 µµee + ee

µµ


γ Z→ X→pp

σ: 0.8 µµGlobal significance of largest deviation for ee+

σ+1

σ+2

σ+3

• For Spin-0 resonance produced via ggF:- Over range mX = 250 GeV to 2.4 TeV:

• Expected limits: 61 fb to 2.1 fb• Observed limits: 88 fb to 1.8 fb

• For Spin-0 resonance produced via VBF:- Limits up to 4% lower

• Results also interpreted for Spin-2 resonance- For ggF, over range mX = 250 GeV to 2.4 TeV:

- Expected limits: 82 fb to 2.7 fb- Observed limits: 117 fb to 2.4 fb

- For VBF, over range mX = 250 GeV to 2.4 TeV:- Expected limits: 66 fb to 1.7 fb- Observed limits: 94 fb to 1.5 fb

• No significant excess at low mass- At mH = 125.09 GeV:

• Expected limit at 95% CL = 5.2 x SM• Observed limit at 95% CL = 6.6 x SM

- 55% gain in exclusion sensitivity over Run 1 result- 20% gain through analysis optimisations

• Largest mllγ candidates found: - 1.47 TeV (eeγ)- 1.57 TeV (μμγ)

• No significant excess observed• Limits set on σ(pp→X) × BR(X→Zγ) at 95% CL

Local p0 for background-only hypothesisLimits for Spin-0 hypothesis

The Future…

37

• ATLAS is getting closer to observing the Standard Model Higgs boson decaying to Zγ• Old study on Run 2 prospects suggested that:

- 300 fb-1 of 14 TeV data for expected limit of 2.53 x SM at 0.67σ- 3000 fb-1 of 14 TeV data for expected limit of 0.74 x SM at 2.12σ- Values are pessimistic!

• New analysis techniques can enhance the sensitivity, reducing the amount of data required- Z mass slices- Dedicated VBF category- ptγ/mllγ

• The diphoton excess at 750 GeV disappeared with 2016 data- It remains important to search for new high mass particles- Discovery of any new resonance would be incredibly exciting!

https://cds.cern.ch/record/1611186?ln=en

Thank you for your attention

38

Backup

39

Z Mass Slices: Event Migration

40

• Events could be incorrectly assigned to a Z slice based on narrow width• Kinematic analysis uses 5% systematic uncertainty on Mllg resolution.• Reconstructed Z mass smeared by 5%

- Z boson 4-vector reevaluated- Higgs boson recreated by combining Z candidate with original, unchanged photon

• Expected limit evaluated, shifts by 5%- Not a systematic uncertainty, but figure of merit for worst-case-scenario

[GeV]Hm120 125 130 135 140 145 150

)γZ→

(HSM

σ)/γZ→

(Hσ

95%

CL

limit

on

02468

101214161820 Z Slice & Kinematic Observed

Z Slice & Kinematic Expectedσ 1±Z Slice & Kinematic σ 2±Z Slice & Kinematic

5% Smearing Observed5% Smearing Expected

σ 1±5% Smearing σ 2±5% Smearing


Higgs production cross-section vs mass

41

[GeV] HM100 200 300 400 500 1000

H+X

) [pb

]

→(p

p σ

-210

-110

1

10= 7 TeVs

LHC

HIG

GS

XS W

G 2

010

H (NNLO+NNLL QCD + NLO EW)

→pp

qqH (NNLO QCD + NLO EW)

→pp

WH (NNLO QCD + NLO EW)

→pp

ZH (NNLO QCD +NLO EW)

→pp

ttH (NLO QCD)

→pp

[GeV] HM80 100 200 300 400 1000

H+X

) [pb

]

→(p

p σ

-210

-110

1

10

210= 8 TeVs

LHC

HIG

GS

XS W

G 2

012

H (NNLO+NNLL QCD + NLO EW)

→pp

qqH (NNLO QCD + NLO EW)

→pp

WH (NNLO QCD + NLO EW)

→pp

ZH (NNLO QCD +NLO EW)

→pp

ttH (NLO QCD)

→pp

SM Higgs: Run 1 Background Composition

42

100 150 200 250 300

Even

ts /

2 G

eV

0

10

20

30

40

50

60

70

80γll→Sherpa Z

ll + jets→Sherpa ZWZttbarData

[GeV]γllm100 150 200 250 300

Dat

a / B

gd

01234567

100 150 200 250 300

Even

ts /

2 G

eV

0

20

40

60

80

100 γll→Sherpa Zll + jets→Sherpa Z

WZttbarData

[GeV]γllm100 150 200 250 300

Dat

a / B

gd

01234567

100 150 200 250 300

Even

ts /

2 G

eV

0

100

200

300

400

500 γll→Sherpa Zll + jets→Sherpa Z

WZttbarData

[GeV]γllm100 150 200 250 300

Dat

a / B

gd

00.20.40.60.811.21.41.61.822.22.4

100 150 200 250 300

Even

ts /

2 G

eV

0

100

200

300

400

500

600γll→Sherpa Z

ll + jets→Sherpa ZWZttbarData

[GeV]γllm100 150 200 250 300

Dat

a / B

gd

00.20.40.60.8

11.21.4

7 TeV eeγ 7 TeV μμγ

8 TeV eeγ 8 TeV μμγ

Z Slices: Proof of Concept 1

43

Injected Signal [no units]7.5 8 8.5 9 9.5 10 10.5 11 11.5

SC

L

00.020.040.060.08

0.10.120.140.160.180.2

Inclusive

Kinematic Categories

Z Slice Categories

Z Slice & Kinematic Categories

• Evaluate compatibility of the background-only case with a signal plus background hypothesis• Level of signal in hypothesis increased in steps of 0.1 times the SM expectation• Corresponding CLs value at each step evaluated• Identify point where CLs falls below 5%

Z Slices: Proof of Concept 2

44

Injected Signal [no units]5.5 6 6.5 7 7.5 8 8.5 9 9.5

BC

L

0.86

0.88

0.9

0.92

0.94

0.96

0.98

1Inclusive

Kinematic Categories

Z Slice Categories

Z Slice & Kinematic Categories

• Evaluate compatibility of the signal plus background case with a background-only hypothesis • Level of signal in hypothesis increased in steps of 0.1 times the SM expectation• Corresponding CLb value at each step evaluated• Identify point where CLb rises above 95%

Z Slices: Impact on expected limits

45

[GeV]Hm120 125 130 135 140 145 150

)γZ→

(HSM

σ)/γZ→

(Hσ

95%

CL

limit

on

02468

101214161820

Kinematic Expected

Z Slice Expected

Z Slice & Kinematic Expected


F-Test

46

• F-test used to identify if a more complex parameterisation will significantly improve the quality of the fit

• Χ02 (Χ12) is the Χ2 value of a binned fit with (less) more complex parameterisation• pk is the number of degrees of freedom of each fit• n is the number of bins in the fitted distribution• If probability of finding values of F more extreme than the value measured on data < 5%

• Reject less complex parametrisation, use moire complex parametrisation• Binning for F-test guarantees sufficient number of events in each bin

Who I am…

47

Super-Kamiokande

CMSATLAS

Who I am…

47

Super-Kamiokande

CMSATLAS

The Baby LHC

Documents

Search for Z-Photon Resonances with ATLAS...1 bb LHC HIGGS XS WG 2011 ττ cc gg γγ Zγ WW ZZ 9 The H→Zγ Decay • Z boson is unstable and immediately decays in various ways -