jet physics at the lhc...2015/10/02 · higgs simpliﬁed cross sectionshiggs + jets: [les houches...

Jet Physics at the LHC

Frank Tackmann

Deutsches Elektronen-Synchrotron

DESY Theory WorkshopOctober 02, 2015

Frank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 0 / 27

Introduction

Introduction.


Introduction

Why Jets?

QCD doesn’t let us observe quarks and gluons directly, only jets of hadrons

ud

u

u

u

u

u

u

u

u

uu

d

u

u

u

Jets tell us the QCD final state of the hard interaction process

36 years ago:

e−

e+

q

q

qe−

e+g

q

Today: Essentially the same (just a bit more complicated ...)


Introduction

Executive Summary of Higgs Production.

∼ 2/3 of Higgs bosons are produced at low pT

0 jets

g

g

H

t

0-1 jets

g

g

H

t

g

∼ 1/3 of Higgs bosons have sizeable pT0 (normal) jets

H

Vq

q

V

1 jet

g

g

H

t

2 (quark) jetsq

H

q

q

qV

Kinematics and number of jets distinguishes different Higgs processesDiscriminates against different backgrounds (e.g. in H→WW, ττ, bb)


Introduction

Why Else Jets?

Jets are Ubiquitous


Introduction

Why Else Jets?

Jets are essential also in EW final states

Recall BR(W→ qq′) = 67%

BR(Z → qq) = 70%

Diboson excess is in dijet mass spectrumof two filtered W/Z-tagged jets[ATLAS 1506.00962] 1.5 2 2.5 3 3.5

Eve

nts

/ 1

00

Ge

V

1−10

1

10

210

310

410Data

Background model

1.5 TeV EGM W', c = 1

2.0 TeV EGM W', c = 1

2.5 TeV EGM W', c = 1

Significance (stat)

Significance (stat + syst)

ATLAS-1 = 8 TeV, 20.3 fbs

WZ Selection

[TeV]jjm1.5 2 2.5 3 3.5

Sig

nific

ance

2−

1−

0123

Search strategies increasingly rely onexploiting jet substructure

Boosted decays: top-jets, W/Z-jets, Higgs-jets b

b

H

Distinguish quark jets (from BSM cascades)from gluon jets (QCD backgrounds)

q q ℓ− ℓ+

g q χ2 ℓ+ χ1

Jet mass, shape, charge, tracks, ...


Introduction

Why Else Jets?

Jets are essential also in EW final states

Recall BR(W→ qq′) = 67%

BR(Z → qq) = 70%

Diboson excess is in dijet mass spectrumof two filtered W/Z-tagged jets[ATLAS 1506.00962] 1.5 2 2.5 3 3.5

Eve

nts

/ 1

00

Ge

V

1−10

1

10

210

310

410Data

Background model

1.5 TeV EGM W', c = 1

2.0 TeV EGM W', c = 1

2.5 TeV EGM W', c = 1

Significance (stat)

Significance (stat + syst)

ATLAS-1 = 8 TeV, 20.3 fbs

WZ Selection

[TeV]jjm1.5 2 2.5 3 3.5

Sig

nific

ance

2−

1−

0123

Search strategies increasingly rely onexploiting jet substructure

Boosted decays: top-jets, W/Z-jets, Higgs-jets b

b

H

Distinguish quark jets (from BSM cascades)from gluon jets (QCD backgrounds)

q q ℓ− ℓ+

g q χ2 ℓ+ χ1

Jet mass, shape, charge, tracks, ...Frank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 4 / 27

Introduction

Outline.

I can only cover a few selected topics:Higgs and jetsJet hierarchiesNew jet algorithm: XConeQuark-gluon separation

QCD corrections in jet processes are typically sizeable(N)NLO calculations are often necessary→ see Giulia Zanderighi’s talk

Jets involve multiple physical scalesFixed-order perturbation theory is often not enoughRequire understanding all-order structure, resummation, hadronic effects,...


Higgs + Jets

Higgs + Jets.


Higgs + Jets

Differential Higgs+Jets Measurements.

dσ/dpjetT

0 20 40 60 80 100 120 140

[pb

/GeV

]j1 T

p /

dσd

2−10

1−10

1

XH + STWZ

XH + JetVHeto

Hbb + Htt + VH = VBF + XH

H→pp ATLAS

data, tot. unc. syst. unc.

-1 = 8 TeV, 20.3 fbs

0≥ jetsN = 0.4, R tkanti-

[GeV]j1

Tp

0 20 40 60 80 100 120 140

ST

WZ

Rat

io to

0

1

2

3

σ(N jets)

0 1 2 3 4 5 6

[pb

]σ

1

10

210

XH8 + Y+PNLOPSNXH8 + Y+PMG5_aMC@NLO

XH + HERPA 2.1.1SXHSTWZ +

XHBLPTW +

Hbb + Htt + VH = VBF + XH

H→pp ATLASdata, tot. unc. syst. unc.

-1 = 8 TeV, 20.3 fbs

> 30 GeVjet

Tp = 0.4, R tkanti-

jetsN

1≥ 2≥ 3≥ = 0 = 1 = 2N

NLO

PS

Rat

io to

0

2

4

Combined high-resolution H → γγ and H → 4` [ATLAS arXiv:1504.05833]

Still statistics limited, but theory uncertainties will become relevant with∼20 times more statistics (∼ end of Run2)Frank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 6 / 27

Higgs + Jets

Higgs Simplified Cross Sections.[Les Houches & LHC Higgs WG2]

Production cross sections split into kinematic bins→ most involve jets

H→ττ· · ·· · ·· · ·

H→

γγ

VBF loose (MVA)

high pTt

VBF tight (MVA)

low pTt

ttH leptonic

VH leptonic

H→ZZ· · ·· · ·· · ·

= 0-jet

H→

WW

= 1-jet

≥ 2-jet VBF cuts

H→

bb

MVA high pT (V )

MVA low pT (V )

· · ·

· · ·

· · ·

σ(ttH)

σ(V

H) low pV

T

high pVT

very high pVT

ratios of Γγγ ΓZZ ΓWW Γbb Γττ (ΓZγ Γµµ)

= 0-jet

≥ 1-jet

σ(bbH) σ(tH)

Rest

σ(V

BF)

≥ 2-jet VBF cuts

high-q2 BSM

≃ 2-jet

& 3-jet

= 0-jet

≥ 1-jet

σ(g

gF)

≥ 2-jet VBF cuts· · ·· · ·

· · ·· · ·

· · ·µi, κi

gk

EFTcoeffs

specificBSM


Higgs + Jets

Jet Binning.

g

g

H

t

g

g

H

t

g

g

g

H

t

pTpcutT

dσ/dpT

σ0(pcutT ) σ≥1(p

cutT )

Spectrum describes transition between 0-jet and ≥ 1-jet regionsNeed consistent treatment of theory uncertainties across entire spectrum

I quite nontrivial because it requires nontrivial correlations(simple factor-2-scale-variation-recipes are not good enough)

I Understood much better by now, but still open issues to figure out

Complete description from combining resummation+fixed orderFrank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 8 / 27

Higgs + Jets

Rapidity-Weighted Jet Bins.[Gangal, Stahlhofen, FT]

Generalize pjetT by defining

Tfj = pTj f(yj)

T jetf = max

j∈J(R)Tfj

Can choose different rapidityweighting functions such that T jet

f

I can be resummedI is insensitive to forward rapidities

Count jets according to Tfj0 jets: σ0(T jet

f <T cut)

≥ 1 jets: σ≥1(T jetf >T cut)

⇒ Provides different slicing through jetphase space

00

1

1 2 3 4

0.2

0.4

0.6

0.8

−1−2−3−4

TBj

TCj

pT j

yj

f(y

j)

jet

jet

00

0

1

1

2

20

40

60

80

100

120

0.5 1.5 2.5yj

pTj[G

eV]

pTj ≤ 30 GeV

TBj ≤ 30 GeV

TCj ≤ 15 GeV


Higgs + Jets

Resolving Jet Bins.(only using H → γγ here)

Pretend uncertainties are much smallerand we want to resolve the “excess”

Different slices of jet phase space addsnontrivial information

One might conclude that “excess” sits athigher pjet

T at more forward rapidities

⇒ Ultimately measure double-differential inpjetT − T

jetf

è

è è

è

è

0

10

20

30

40

50

Njets

σN

[fb]

mH =125.4GeV

pcutT =30GeV

ATLAS H→γγ

ggH(STWZ, BLPTW)+XH

XH=VBF+V H+ttH

≥ 0 = 0 ≥ 1 = 1 ≥ 2

è

è

è

è

0

1

10

10 20 30 40 500.01

0.1

T jetC [GeV]

dσ/dT

jet

C[fb/GeV]

mH =125.4GeV ATLAS H→γγ

ggH(NLL′+NLO)+XH

XH=VBF+V H+ttH

è

è

èè

è

0

1

20 40 60 80 100 120 1400.01

0.1

pjetT [GeV]

dσ/dpjet

T[fb/GeV]

mH =125.4GeV ATLAS H→γγ

ggH(NNLL′+NNLO)+XH

ggH(NLL′+NLO)+XH

XH=VBF+V H+ttH


Jet Hierarchies

Jet Hierarchies.


Jet Hierarchies

In More Dimensions and with More Jets...

pjet1T

pjet2T

pcutT

pcutT

0-jet1-jet

≥2-jet(excludedbypjet2

T

<pjet1

T

)

... it quickly gets more complicated

Multiscale problem

· · · �∼ pjet2T

�∼ pjet1T

�∼ QDifferent multiple hierarchiespossibleResummation often limited to LL(parton shower)

I pcutT �pjet1

T ∼mH limit isknown at NLL′ [Liu, Petriello]

Need to better understand distributionand correlations

between different kinematic variablesacross different jet multiplicities

⇒ Increasingly important as we probe higher and higher scales(and in particular if we find new heavy particles)


Jet Hierarchies

Jet Hierarchies.[Bauer, FT, Walsh, Zuberi ’11]

[Procura, Waalewijn, Zeune ’14][Larkoski, Moult, Neill ’15]

[Pietrulewicz, FT, Waalewijn (prep.)]

SCET+: Extension of SCET to treat additional(emerging) scale hierarchies beyond standard SCET

Adds two types of intermediate modesI collinear-soft: soft mode arising from a collinear sectorI soft-collinear: collinear mode arising from a soft sectorI Relative scaling and interactions depend on specific observable

⇒ Resummation of hierarchiesin hard jet kinematics

3 2

13

2

1

32

1

3 2

1

t ∼ u ∼ s

t� u ∼ s

t� u� s

t ∼ u� s

⇒ Resummation of double-differential spectra→ talk by Lisa Zeune


Jet Hierarchies

Jet Rates and Nonglobal Logarithms.Thrust with cone jets [Chien, Hornig, Lee 1509.04287]

2(Λ/Q)R2 ∼ τ � R2 � 1

With SCET+ can resum ln τ and also lnR

For τ ∼ R2 resum all global lnR in total jet rateI But not appearing non-global logs of Λ/Q

“Resum” nonglobal logs by explicitly resolving andsumming infinite set of emissions

Resolved soft subjets→ “dressed gluons”[Larkoski, Moult, Neill 1501.04596, 1508.07568]

I Infinite sum reproduces BMS equation[Dasgupta, Salam; Banfi, Marchesini, Smye]

I Instead: Expand in number of dressed gluons→ Systematic reorganization of pert. series

Renormalize “Color-density matrix”[Caron-Huot 1501.03754]

Renormalize matrix of bare emissions[Becher, Neubert, Rothen Shao 1508.06645]

��

��

��

��

��

��

��

�

��(�)/��

��-��

��

�-��

�-��

�-��

�-��

�� -��

(αs/π)Nc ln(mL/mR)


Jet Hierarchies

Phenomenology of Small-R Jets.[Dasgupta, Dreyer, Salam, Soyez, Cacciari]

Jet algorithms induce logarithms ∼ αs lnR

Use a generating functional formalism to numerically resum leading(αs lnR)n terms (LLR)

I Exploits angular ordering, evolution variable t ∼ (αs/2π) ln(R20/R

2)

Studied several jet observables, including jet trimming and filteringI Trimming is particularly sensitive to small-R effects


Jet Hierarchies

Phenomenology of Small-R Jets.

[plots stolen from F. Dreyer’s BOOST 2015 talk]

Inclusive jet spectrumLLR has noticeable effect forR . 0.4


Jet Hierarchies

Theory for Jet Substructure.

Jet substructure methods

Typically involve complex algorithms and cuts, very sensitive toresummation and nonperturbative effects

I Vast majority of studies and applications heavily rely (necessarily) on partonshower Monte Carlos

In past few years focus increased on gaining a better analytic andsystematic understanding in two important and complementary ways[many groups/authors ...]

I Better analytic understanding, tools, calculationsI Instead of optimizing solely for performance, also taking theory controlability

into account in designing new algorithms and methods


Jet Hierarchies

Analytic Boosted Boson Discrimination.[Larkoski, Moult, Neill]

First factorization and analytic resummation of a 2-prong variable D2

(similar to 2-subjettiness but based on energy correlation functions)

For boosted boson signal and QCD background (latter is the hard part)

Use SCET+ to resum and combine different subjet limits

e(�)2 ⇠

⇣e(↵)2

⌘�/↵

e(↵)3 ⌧

⇣e(↵)2

⌘3

jet axis

R

Collinear Subjets

Jna

Jnb

S+nanbnt

Sntnt

e(↵)2 ⇠ e

(�)2

e(↵)3 ⌧

⇣e(↵)2

⌘3

jet axis

RJn

Soft Subjet

Snnnsj

Snsj nsj

Jnsj

e(↵)2 ⇠ e

(�)2

jet axis

R

e(↵)3 ⇠

⇣e(↵)2

⌘2

Jn

Soft Haze

Snn


Jet Hierarchies

Analytic Boosted Boson Discrimination.[Larkoski, Moult, Neill]

��

��

��

��

��(��)

��

��(��)

��

�� +�- → ��

� ��

�� ∈ [�� ] �� =�

��

��

�� -�

��-�

��-�

��-�

� ��

��(��)� � ��

�� ++

�� +�- → ��

�� ∈ [�� ] �� =�

Region relevant for discrimination is very sensitive to hadronizationI Described by nonperturbative soft function with field-theoretic definitionI Good description with single-parameter shape function

Should be possible to extend to other substructure observables and ppcollisions


XCone

XCone.

Substructure without substructure


XCone

XCone: An eXclusive Cone-Jet Algorithm.[Stewart, FT, Thaler, Vermilion, Wilkason]

Basic constructionStart from standard N-jettiness

TN({nk}) =∑

i

min {ρjet(pi, n1), . . . , ρjet(pi, nN), ρbeam(pi)}

I Given N jet axes {nk}, partitions event into N jet regions and beam regionI Shape and size of jet regions depend on ρjet and ρbeam measures

(New XCone measure yields conical jets and maintains theory control)

Minimize over all jet axes

TN = minn1,n2,...,nN

TN({nk})I Finding exact global minimum is computationally expensiveI Finding approximate (local) minimum is sufficient and inexpensive

Key featuresExclusive: Returns preselected number of jets

I Yields best interpretation of the event in terms of the signal one is looking forI Final TN value provides additional quality measure for free

Smoothly transitions between well-separated and boosted regimesFrank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 19 / 27

XCone

Comparison to anti-kT .

Boosted tt→ hadrons event (from BOOST 2010 sample)

Well-separated jetsXCone jets practically the sameas leading anti-kT jets

Adjacent/overlapping jetsanti-kT merges signal jets, picksup ISR, FSR jetsXCone still finds jets, dynamicallysplit by nearest-neighbor


XCone

Case Study: Boosted Higgs Reconstruction.[Thaler, Wilkason]

Boosted H → bb with increasing pT

XCone provides stable performanceacross kinematic regimes

I Effectively provides automatic transitionfrom 2-jettiness to 2-subjettiness

I Further improvements possible, e.g. withexplicit ISR veto

(GeV)T,min

p200 300 400 500 600 700 800 900 1000

Sig

nal

Eff

icie

ncy

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

= 2β

= 1β

Tak

R = 0.5

[100,150] GeV∈m

Higgs Efficiency for N = 2


XCone

Case Study: Boosted Higgs Reconstruction.[Thaler, Wilkason]

(GeV)jjm0 50 100 150 200 250

Rel

ativ

e O

ccurr

ence

0

0.05

0.1

0.15

0.2

0.25

> 200 GeV)T

Higgs N = 2 (p

= 2β

= 1β

Tak

R = 0.5

> 200 GeV)T

Higgs N = 2 (p

(GeV)jjm0 50 100 150 200 250

Rel

ativ

e O

ccurr

ence

0

0.05

0.1

0.15

0.2

0.25

> 500 GeV)T

Higgs N = 2 (p

= 2β

= 1β

Tak

R = 0.5

> 500 GeV)T

Higgs N = 2 (p

(GeV)jjm0 50 100 150 200 250

Rel

ativ

e O

ccurr

ence

0

0.05

0.1

0.15

0.2

0.25

> 800 GeV)T

Higgs N = 2 (p

= 2β

= 1β

Tak

R = 0.5

> 800 GeV)T

Higgs N = 2 (p

Boosted H → bb with increasing pTXCone provides stable performanceacross kinematic regimes

I Effectively provides automatic transitionfrom 2-jettiness to 2-subjettiness

I Further improvements possible, e.g. withexplicit ISR veto

(GeV)T,min

p200 300 400 500 600 700 800 900 1000

Sig

nal

Eff

icie

ncy

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

= 2β

= 1β

Tak

R = 0.5

[100,150] GeV∈m

Higgs Efficiency for N = 2


XCone

Case Study: Boosted Top Reconstruction.

Classic example of jet substructure: pp→ tt→WWbb→ qqqqbb

(GeV)T, min

p200 300 400 500 600 700

Sig

nal

Eff

icie

ncy

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

= 2β = 1β

(Bst)Tak (Res)Tak

R = 0.5 [150,200] GeV∈m

> 50 GeVW+ m

3×Top Efficiency for N = 2

(GeV)T, min

p200 300 400 500 600 700

Bac

kgro

und M

ista

g

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

= 2β = 1β

(Bst)Tak (Res)Tak

R = 0.5 [150,200] GeV∈m

> 50 GeVW+ m

3×QCD Mistag for N = 2

(GeV)T,min

p200 300 400 500 600 700

Im

pro

vem

ent

BS

/

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

= 2β = 1β

(Bst)Tak (Res)Tak

R = 0.5 [150,200] GeV∈m

> 50 GeVW+ m

3×Signal Significance for N = 2

Use XCone with N = 2× 3 (R2 � 1, R3 = 0.5)

Compare toI “Resolved”: 2 kT jets (R� 1) with 3 anti-kT subjetsI “Boosted”: 2 anti-kT jets (R = 1.0) with 3 kT subjets

⇒ Better performance across all pTFrank Tackmann (DESY) Jet Physics at the LHC 2015-10-02 22 / 27

Quark-Gluon Discrimination

Quark-Gluon Discrimination.

a.k.a. Hunting the White Whale of Jet Substructure



Quark-Gluon Discrimination.

QCD background jets from ISR and FSR are dominantly “gluon-jets”

Signal jets are dominantly “quark-jets”I Color singlets couple to quarks only

⇒ Would be powerful discriminator if it could be exploitedI See b-taggingI Long history, but still (almost embarassingly) not well under controlI Initiated a comprehensive study at Les Houches this year



What Are We Even Talking About?

Jesse Thaler Ñ Report of the Les Houches Quark/Gluon Subgroup 3

What is a Quark Jet?!From lunch/dinner discussions

A quark parton!

A Born-level quark parton!

The initiating quark parton in a Þnal state shower!

An eikonal line with baryon number 1/3"and carrying triplet color charge!

A quark operator appearing in a hard matrix element"in the context of a factorization theorem!

A parton-level jet object that has been quark-tagged using a soft-safe ßavored jet algorithm (automatically collinear safe if you sum constituent ßavors)!

A phase space region (as deÞned by an unambiguous hadronic Þducial cross section measurement) that yields an enriched sample of quarks (as interpreted by some suitable, though fundamentally ambiguous, criterion)

Ill-DeÞned

Well-DeÞned

[stolen from Jesse Thaler, Les Houches]



What Are We Even Talking About?

Jesse Thaler Ñ Report of the Les Houches Quark/Gluon Subgroup 3

What is a Quark Jet?!From lunch/dinner discussions

A quark parton!

A Born-level quark parton!

The initiating quark parton in a Þnal state shower!

An eikonal line with baryon number 1/3"and carrying triplet color charge!

A quark operator appearing in a hard matrix element"in the context of a factorization theorem!

A parton-level jet object that has been quark-tagged using a soft-safe ßavored jet algorithm (automatically collinear safe if you sum constituent ßavors)!

A phase space region (as deÞned by an unambiguous hadronic Þducial cross section measurement) that yields an enriched sample of quarks (as interpreted by some suitable, though fundamentally ambiguous, criterion)

Ill-DeÞned

Well-DeÞned What we mean

What people

sometimes

think we mean

[stolen from Jesse Thaler, Les Houches]



Les Houches Quark/Gluon Performance Study.[Les Houches q/g subgroup]

Compare e+e−→ uu (“quark-tagged”) to e+e−→ gg (“gluon-tagged”)

Generalized jet angularities: λκβ =∑i∈jet z

κi θ

βi

Separation power:1

2

[S(λ)−B(λ)]2

S(λ) +B(λ)

“Les Houches Angularity” λ(κ = 1, β = 0.5), R = 0.6, Q = 200 GeV

quarksPythia-8205Sherpa-2.1.1Vincia-1201Herwig-2 7 1

0

1

2

3

4

5

λ(κ=1, β=0.5), R=0.6

1/N

dN

/dλ

0 0.2 0.4 0.6 0.8 1

0.6

0.8

1

1.2

1.4

λ

MC

/Dat

a

gluons

Pythia-8205Sherpa-2.1.1Vincia-1201Herwig-2 7 1

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

λ(κ=1, β=0.5), R=0.6

1/N

dN

/dλ

0 0.2 0.4 0.6 0.8 1

0.6

0.8

1

1.2

1.4

λ

MC

/Dat

a

separation power

Pythia-8205Sherpa-2.1.1Vincia-1201Herwig-2 7 1

0

0.5

1

1.5

2

λ(κ=1, β=0.5), R=0.6

1/N

dN

/dλ

0 0.2 0.4 0.6 0.8 1

0.6

0.8

1

1.2

1.4

λ

MC

/Dat

a



Quark/Gluon Performance: Some Lessons Learned.total separation power

Preliminary

gluon rejection at 50% quark eff.

Preliminary

Improving quark/gluon robustness seems synonymous with controllingfinal state shower uncertainties (LEP measured quark not gluon event shapes)

Hadronization can be important, even for IRC safe angularities⇒ Comparison to analytic predictions, uncertainties, stay tuned ...



Summary.

Jets are our window onto the hard interaction

Jets inevitably involve nontrivial QCD corrections and several physicalscales

I Fixed-order calculations are often necessary but seldom sufficient

As experimental analyses are getting more sophisticated they make moredetailed use of jets

I More differential, more exclusive binning, more detailed substructure, ...I Important to keep theory under control

Calculations and methods are constantly being refined and developedI But there are also many open issues

Many things I have not covered (apologies ...)→ See e.g. many interesting talks at BOOST 2015


jet physics at the lhc...2015/10/02 · higgs simpliﬁed cross sectionshiggs + jets: [les houches...

Documents