Tau Physics at ATLAS: recent results and prospects

Quentin Buat - Simon Fraser University

1Séminaires du LAL - February 2nd, 2016

2

Run1 legacy: success of the Standard Model

• Discovery of the Higgs boson by ATLAS and CMS, driven by the bosonic final states

• Couplings and CP measurements confirm agreement with the SM expectations3

pp

total

80 µb−1

JetsR=0.4

|y |<3.0

0.1< pT < 2 TeV

DijetsR=0.4

|y |<3.0y ∗<3.0

0.3<mjj < 5 TeV

W

fiducial

35 pb−1nj ≥ 0

nj ≥ 1

nj ≥ 2

nj ≥ 3

nj ≥ 4

nj ≥ 5

nj ≥ 6

nj ≥ 7

Z

fiducial

35 pb−1nj ≥ 0

nj ≥ 1

nj ≥ 2

nj ≥ 3

nj ≥ 4

nj ≥ 5

nj ≥ 6

nj ≥ 7

t̄t

fiducial

e, µ+X

nj ≥ 4

nj ≥ 5

nj ≥ 6

nj ≥ 7

nj ≥ 8

tt−chan

total

WW

total

γγ

fiducial

Wt

total

2.0 fb−1

H

fiducial

H→γγ

VBFH→WW

ggFH→WW

H→ZZ→4ℓ

total(γγ,ZZ )

H→ττ

WZ

total

13.0 fb−1

ZZ

total

Wγ

fiducial

WW+WZ

semilept.fiducial

Zγ

fiducial

t̄tW

total

t̄tZ

total

95% CL

upper

limit

t̄tγ

fiducial

ZjjEWK

fiducial

Wγγ

fiducialnjet=0

W±W±jjEWK

fiducial

ts−chan

total

95% CL

upper

limit

0.7 fb−1

σ[p

b]10−3

10−2

10−1

1

101

102

103

104

105

106

1011

LHC pp√s = 7 TeV

Theory

Observed 4.5 − 4.9 fb−1

LHC pp√s = 8 TeV

Theory

Observed 20.3 fb−1

Standard Model Production Cross Section Measurements Status: March 2015

ATLAS Preliminary

Run 1√s = 7, 8 TeV

Run1 legacy: beyond the Standard Model?

4

Model ℓ, γ Jets EmissT

!L dt[fb−1] Limit Reference

Ext

rad

ime

nsi

on

sG

au

ge

bo

son

sC

ID

ML

QH

eavy

qu

ark

sE

xcite

dfe

rmio

ns

Oth

er

ADD GKK + g/q − ≥ 1 j Yes 20.3 n = 2 1502.015185.25 TeVMD

ADD non-resonant ℓℓ 2e, µ − − 20.3 n = 3 HLZ 1407.24104.7 TeVMS

ADD QBH→ ℓq 1 e, µ 1 j − 20.3 n = 6 1311.20065.2 TeVMth

ADD QBH − 2 j − 20.3 n = 6 1407.13765.82 TeVMth

ADD BH high Ntrk 2 µ (SS) − − 20.3 n = 6, MD = 3 TeV, non-rot BH 1308.40754.7 TeVMth

ADD BH high!pT ≥ 1 e, µ ≥ 2 j − 20.3 n = 6, MD = 3 TeV, non-rot BH 1405.42545.8 TeVMth

ADD BH high multijet − ≥ 2 j − 20.3 n = 6, MD = 3 TeV, non-rot BH 1503.089885.8 TeVMth

RS1 GKK → ℓℓ 2 e, µ − − 20.3 k/MPl = 0.1 1405.41232.68 TeVGKK mass

RS1 GKK → γγ 2 γ − − 20.3 k/MPl = 0.1 1504.055112.66 TeVGKK mass

Bulk RS GKK → ZZ → qqℓℓ 2 e, µ 2 j / 1 J − 20.3 k/MPl = 1.0 1409.6190740 GeVGKK mass

Bulk RS GKK →WW → qqℓν 1 e, µ 2 j / 1 J Yes 20.3 k/MPl = 1.0 1503.04677760 GeVW′ mass

Bulk RS GKK → HH → bb̄bb̄ − 4 b − 19.5 k/MPl = 1.0 1506.00285500-720 GeVGKK mass

Bulk RS gKK → tt 1 e, µ ≥ 1 b, ≥ 1J/2j Yes 20.3 BR = 0.925 1505.070182.2 TeVgKK mass

2UED / RPP 2 e, µ (SS) ≥ 1 b, ≥ 1 j Yes 20.3 1504.04605960 GeVKK mass

SSM Z ′ → ℓℓ 2 e, µ − − 20.3 1405.41232.9 TeVZ′ mass

SSM Z ′ → ττ 2 τ − − 19.5 1502.071772.02 TeVZ′ mass

SSM W ′ → ℓν 1 e, µ − Yes 20.3 1407.74943.24 TeVW′ mass

EGM W ′ →WZ → ℓν ℓ′ℓ′ 3 e, µ − Yes 20.3 1406.44561.52 TeVW′ mass

EGM W ′ →WZ → qqℓℓ 2 e, µ 2 j / 1 J − 20.3 1409.61901.59 TeVW′ mass

EGM W ′ →WZ → qqqq − 2 J − 20.3 1506.009621.3-1.5 TeVW′ mass

HVT W ′ →WH → ℓνbb 1 e, µ 2 b Yes 20.3 gV = 1 1503.080891.47 TeVW′ mass

LRSM W ′R→ tb 1 e, µ 2 b, 0-1 j Yes 20.3 1410.41031.92 TeVW′ mass

LRSM W ′R→ tb 0 e, µ ≥ 1 b, 1 J − 20.3 1408.08861.76 TeVW′ mass

CI qqqq − 2 j − 17.3 ηLL = −1 1504.0035712.0 TeVΛ

CI qqℓℓ 2 e, µ − − 20.3 ηLL = −1 1407.241021.6 TeVΛ

CI uutt 2 e, µ (SS) ≥ 1 b, ≥ 1 j Yes 20.3 |CLL | = 1 1504.046054.3 TeVΛ

EFT D5 operator (Dirac) 0 e, µ ≥ 1 j Yes 20.3 at 90% CL for m(χ) < 100 GeV 1502.01518974 GeVM∗EFT D9 operator (Dirac) 0 e, µ 1 J, ≤ 1 j Yes 20.3 at 90% CL for m(χ) < 100 GeV 1309.40172.4 TeVM∗

Scalar LQ 1st gen 2 e ≥ 2 j − 20.3 β = 1 Preliminary1.05 TeVLQ mass

Scalar LQ 2nd gen 2 µ ≥ 2 j − 20.3 β = 1 Preliminary1.0 TeVLQ mass

Scalar LQ 3rd gen 1 e, µ ≥1 b, ≥3 j Yes 20.3 β = 0 Preliminary640 GeVLQ mass

VLQ TT → Ht + X 1 e, µ ≥ 2 b, ≥ 3 j Yes 20.3 T in (T,B) doublet 1505.04306855 GeVT mass

VLQ YY →Wb + X 1 e, µ ≥ 1 b, ≥ 3 j Yes 20.3 Y in (B,Y) doublet 1505.04306770 GeVY mass

VLQ BB → Hb + X 1 e, µ ≥ 2 b, ≥ 3 j Yes 20.3 isospin singlet 1505.04306735 GeVB mass

VLQ BB → Zb + X 2/≥3 e, µ ≥2/≥1 b − 20.3 B in (B,Y) doublet 1409.5500755 GeVB mass

T5/3 →Wt 1 e, µ ≥ 1 b, ≥ 5 j Yes 20.3 1503.05425840 GeVT5/3 mass

Excited quark q∗ → qγ 1 γ 1 j − 20.3 only u∗ and d∗, Λ = m(q∗) 1309.32303.5 TeVq∗ mass

Excited quark q∗ → qg − 2 j − 20.3 only u∗ and d∗, Λ = m(q∗) 1407.13764.09 TeVq∗ mass

Excited quark b∗ →Wt 1 or 2 e, µ 1 b, 2 j or 1 j Yes 4.7 left-handed coupling 1301.1583870 GeVb∗ mass

Excited lepton ℓ∗ → ℓγ 2 e, µ, 1 γ − − 13.0 Λ = 2.2 TeV 1308.13642.2 TeVℓ∗ mass

Excited lepton ν∗ → ℓW , νZ 3 e,µ, τ − − 20.3 Λ = 1.6 TeV 1411.29211.6 TeVν∗ mass

LSTC aT →W γ 1 e, µ, 1 γ − Yes 20.3 1407.8150960 GeVaT mass

LRSM Majorana ν 2 e, µ 2 j − 20.3 m(WR ) = 2.4 TeV, no mixing 1506.060202.0 TeVN0 mass

Higgs triplet H±± → ℓℓ 2 e, µ (SS) − − 20.3 DY production, BR(H±±L → ℓℓ)=1 1412.0237551 GeVH±± mass

Higgs triplet H±± → ℓτ 3 e,µ, τ − − 20.3 DY production, BR(H±±L→ ℓτ)=1 1411.2921400 GeVH±± mass

Monotop (non-res prod) 1 e, µ 1 b Yes 20.3 anon−res = 0.2 1410.5404657 GeVspin-1 invisible particle mass

Multi-charged particles − − − 20.3 DY production, |q| = 5e 1504.04188785 GeVmulti-charged particle mass

Magnetic monopoles − − − 7.0 DY production, |g | = 1gD , spin 1/2 Preliminary1.34 TeVmonopole mass

Mass scale [TeV]10−1 1 10√s = 7 TeV

√s = 8 TeV

ATLAS Exotics Searches* - 95% CL ExclusionStatus: July 2015

ATLAS Preliminary"L dt = (4.7 - 20.3) fb−1

√s = 7, 8 TeV

*Only a selection of the available mass limits on new states or phenomena is shown.

No new physics, lot

of constraints on models at the TeV

scale

Run2: 8 TeV → 13 TeV

5

100 10001

10

100

gg Σqq qg

WJS2013

ratios of LHC parton luminosities: 13 TeV / 8 TeV

lu

min

osi

ty r

atio

MX (GeV)

MSTW2008NLO

_

Very large gain for High-mass states

Increase of ~2 for ‘common’ processes, drives the trigger rates

Upgrading for two years

6

• New inner detector

• Closest layer to the beam pipe.

• Trigger system upgrade

• Increase L1 output rate from 75 to 100 kHz

• New Central Trigger Processor

• Merge L2 and HLT farms and additional ressources to increase output rate

• Various other upgrade/consolidation

Insertable B Layer (IBL)

Back in business!

7

2012: <μ> = 21

• Recorder 4.2fb-1 of pp collisions, peak luminosity of 5.1033cm-2s-1

• Very smooth operations, smaller dataset than expected

2011: <μ> = 9

DAQ, Trigger

• DQ efficiency: 87%

• main loss due to IBL off in 2 runs for safety reasons

• Luminosity: 3.2±0.2 fb-1

• Measured with vdM scans

• Very smooth trigger operations

8]-1s-2 cm30instantaneous Luminosity [104200 4250 4300 4350 4400 4450 4500

Trig

ger R

ate

[Hz]

500100015002000250030003500400045005000

High Level Trigger TotalData Scouting (partial EB)Main Physics (full EB)Detector Calibration (partial EB)Detector Monitoring (partial EB)

= 13 TeVs

pp Data Oct 2015

ATLAS Trigger Operation

Performances

9

Effic

ienc

y

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1ATLAS Preliminary

-1 = 13 TeV, 3.2 fbs<2.47η-2.47<

Data: full, MC: open

Loose Medium Tight

[GeV]TE

10 20 30 40 50 60 70 80

Dat

a / M

C

0.8

0.9

1

e

210×6 310 310×2 310×3

⟩ re

coil

T/p

lead

jet

T p⟨

0.98

0.99

1

1.01

1.02

1.03

1.04

1.05

1.06

[GeV]recoilT

p210×6 310 310×2 310×3

MC

/ D

ata

0.960.98

11.021.04

DataPythia8Sherpa 2.1Herwig++

ATLAS Preliminary-1 = 13 TeV, 3.3 fbs

Multijet Events R = 0.4, EM+JES (in-situ)tanti-k

< 1.2lead jetη

η

2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

Dat

a/M

C

0.951

1.05

η2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

Effic

ienc

y

0.9

0.95

1

η2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

0.4

0.6 PreliminaryATLAS

-1 = 13 TeV, 3.3 fbs

muonsTight muonsMedium

muonsLoose

Data MC

Even

ts

0

1000

2000

3000

4000

5000ATLAS Preliminary

-1 L dt = 3.3 fb∫ = 13 TeVs

T&Phadτµ → ττ →Z

Data 2015 ττ→ZW+jets and multijets OtherStat. Unc.

) [GeV]µ,τVisible Mass (

40 60 80 100 120 140 160

Dat

a/ex

p.

0.51

1.52

μ

jets

[GeV]T EΣ

0 100 200 300 400 500 600 700

RM

S [G

eV]

mis

sy

,Em

iss

xE

0

5

10

15

20

25

T EΣData : fit 0.52

ATLAS PreliminaryMay 2015 comm., 13 TeV

MET

τ

Performances

10

Effic

ienc

y

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1ATLAS Preliminary

-1 = 13 TeV, 3.2 fbs<2.47η-2.47<

Data: full, MC: open

Loose Medium Tight

[GeV]TE

10 20 30 40 50 60 70 80

Dat

a / M

C

0.8

0.9

1

210×6 310 310×2 310×3

⟩ re

coil

T/p

lead

jet

T p⟨

0.98

0.99

1

1.01

1.02

1.03

1.04

1.05

1.06

[GeV]recoilT

p210×6 310 310×2 310×3

MC

/ D

ata

0.960.98

11.021.04

DataPythia8Sherpa 2.1Herwig++

ATLAS Preliminary-1 = 13 TeV, 3.3 fbs

Multijet Events R = 0.4, EM+JES (in-situ)tanti-k

< 1.2lead jetη

e

η

2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

Dat

a/M

C

0.951

1.05

η2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

Effic

ienc

y

0.9

0.95

1

η2.5− 2− 1.5− 1− 0.5− 0 0.5 1 1.5 2 2.5

0.4

0.6 PreliminaryATLAS

-1 = 13 TeV, 3.3 fbs

muonsTight muonsMedium

muonsLoose

Data MCμ

jets

[GeV]T EΣ

0 100 200 300 400 500 600 700

RM

S [G

eV]

mis

sy

,Em

iss

xE

0

5

10

15

20

25

T EΣData : fit 0.52

ATLAS PreliminaryMay 2015 comm., 13 TeV

MET

Even

ts

0

1000

2000

3000

4000

5000ATLAS Preliminary

-1 L dt = 3.3 fb∫ = 13 TeVs

T&Phadτµ → ττ →Z

Data 2015 ττ→ZW+jets and multijets OtherStat. Unc.

) [GeV]µ,τVisible Mass (

40 60 80 100 120 140 160

Dat

a/ex

p.

0.51

1.52

τ

First look at data shows good performance and good data/MC agreement in key measurements

Tau Physics at the LHC

• The LHC is not a place to make precision measurements of the tau lepton

• As the heaviest lepton in the SM, the tau is a very useful probe of the EWSB mechanism:

• Direct coupling the Higgs boson

• Establishing the tau polarization in W/Z decays

• Lepton Flavour Violation

• Many exotica and SUSY signatures involve taus, their discovery potential is usually weaker than signatures with light leptons (e/mu)

11

In this talk, instead of a broad review of the O(30) tau-related ATLAS results, I want to discuss the ATLAS strategy to detect hadronically-decaying taus and

highlight a few recent results

Tau lepton at ATLAS

• m𝜏 = 1.777 GeV, c𝜏 = 87 μm , decays: 𝜏 →lν (35%), 𝜏 → hadron(s) + ν (65%)

• Leptonic tau decays are nearly indistinguishable from prompt light lepton

• Hadron collider vocabulary: tau = hadronic decay ≈ collection of π± and π0

12

• Signature: calorimeter deposit associated with 1 or 3 tracks

• Reconstruction seeded by an anti-kT jet algorithm with a radius of 0.4: • Core within 0.2, isolation ring [0.2, 0.4] • Tracks associated to the core or the isolation ring

Tau Energy Scale• Seeded with an anti-kT jet (dR=0.4) at the LC scale.

• MC-driven corrections to agree with the simulated true visible ET

• Deconvolution method and in-situ Z→ττ measurement to provide a final ~1% correction and assert the systematic uncertainties:

• ~2-3% for 1prong taus and 3-4% for 3prongs taus

13 [GeV]τTP

20 30 40 50 60 70 80 100 200

Frac

tiona

l unc

erta

inty

00.010.02

0.030.040.050.060.07

0.080.09

0.1

Single particle resp. Material modeling

Underlying event Non-closure

Pile-Up Total uncertainty

|<0.3η|1 prong decays

2012 Data + Simulation PreliminaryATLAS = 8 TeV s

[GeV]vism20 40 60 80 100 120

Even

ts /

3 G

eV

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Even

ts /

3 G

eV

0

1000

2000

3000

4000

5000

6000

7000Data 2012

τ hadτ µ→Z

W+jetsSame Sign

, Topll→*γZ/

Stat. Unc.

1−track

ATLAS-1 Ldt = 20.3 fb∫

= 8 TeVs

[GeV]vism20 40 60 80 100 120D

ata/

Sim

.

0.60.8

11.21.4

Eur. Phys. J. C75 (2015) 303

Tau Identification• Multivariate discriminant (BDT) based on a combinaison of calorimetric and track-based

variables.

• Two discriminants trained separately for 1prong and 3prongs candidates

• Very different calorimetric response, usage of displaced vertex possible in 3prongs

• Achieve an efficiency of 40% for a rejection of 60 (medium working point)

14

Signal Efficiency0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Inve

rse

Back

grou

nd E

ffici

ency

1

10

210

310

410

|< 2.5η> 40 GeV, |Tp

ATLAS =8TeVsData 2012,

Tau identification

1-track

3-track

centf0 0.2 0.4 0.6 0.8 1

Arbi

trary

Uni

ts

0

0.05

0.1

0.15

0.2

0.25 (Simulation)ντ →, W ττ →Z, Z’

Multi-Jet (Data 2012)

= 8 TeVs ATLAS

1-track

|< 2.5η> 15 GeV, |Tp

Efficiency includes track reconstruction

Eur. Phys. J. C75 (2015) 303

Tau Identification

• Very robust response agains pile-up obtained by explicitly correcting the input variables using nvtx

• Agreement with data checked in Z→ττ→μτh events

• Scale factor deviates from unity by a few %

15

Barrel Endcap Barrel Endcap Barrel Endcap

Cor

rect

ion

Fact

or

0.9

1

1.1

1.2

1.3

1.4

Loose Medium Tight

ATLAS = 8 TeVs, -1 L dt = 20.3 fb∫

Data 20121-track

channelhadτµτ→Z

channelhadτeτ→Z

-dependent)ηCombination (

Combination (inclusive)

Number of primary vertices5 10 15 20 25

Sign

al E

ffici

ency

0

0.2

0.4

0.6

0.8

1

1.2loosemediumtight |< 2.5η> 15 GeV, |

Tp

1-track

=8 TeVsSimulation,ATLAS

Eur. Phys. J. C75 (2015) 303

Triggering with taus?

• We try to avoid it but it is required for key signatures.

• Fully Hadronic final state: BR(X→τhτh) = 42%

• Hadronic taus are the best trigger handle available

• Semi leptonic final state: BR(X→τhτl) = 46%

• At low instantaneous luminosity, most searches can rely on the single lepton triggers

• With the nominal Run2 setup, combined lepton + tau triggers will be required

16

ATLAS (Tau) Trigger System: Run1

17

Level 1Calorimeter + Muon

System(coarse granularity)

Level 2Seeded by L1 Region Of

Interest. Fast tracking

Event FilterFull granularity, full

tracking in the Region of Interest

75 kHz 3 kHz 600 Hz

τ

T)/Offline pτ

T - Offline pτ

T(L2 p

-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1

Even

ts /

0.05

0

500

1000

1500

2000

2500

Even

ts /

0.05

0

500

1000

1500

2000

2500

3000

3500Data 2012

ττ→ZW+jets

OtherSame SignStat. Unc.

-1 L dt = 20.3 fb∫ = 8 TeVs PreliminaryATLAS

τ

T)/Offline pτ

T - Offline pτ

T(L2 p

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1Dat

a/M

C

0.60.8

11.21.4

τ

T)/Offline pτ

T - Offline pτ

T(EF p

-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1Ev

ents

/ 0.

025

0

2000

4000

6000

8000

10000

12000

Even

ts /

0.02

5

0

2000

4000

6000

8000

10000

12000

14000

16000

18000 Data 2012ττ→Z

W+jets

OtherSame SignStat. Unc.

-1 L dt = 20.3 fb∫ = 8 TeVs PreliminaryATLAS

τ

T)/Offline pτ

T - Offline pτ

T(EF p

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1Dat

a/M

C0.60.8

11.21.4

Tau Energy Resolution

[GeV]TpOffline tau

0 20 40 60 80 100

Effic

ienc

y

0

0.2

0.4

0.6

0.8

1

L1

L1 + L2

L1 + L2 + EF

-1 Ldt = 20.3 fb ,∫ = 8 TeVs

ATLAS

hadτµ

τ→Data 2012, Z

20 GeV tau trigger

Tau Trigger Run1 Performances

18

Efficiency wrt offline identified taus

{

L1: very slow turn-on

HLT ID ≠ Offline ID

L2 ID ≠ Offline ID

ATLAS Trigger System: Run2 upgrade

19

Level 1Calorimeter + Muon

System(coarse granularity)

Level 2Fast tracking

Event FilterFull Granularity, Full

tracking in the RoI

High Level Trigger (L2 and EF are merged)

100 kHz 1000 Hz

• L1Topo:128 items available for combination

• L1Calo: Relative isolation cut

• L2/EF merger:• Topo-clustering available right at the L1 output.• Full tracking limited to the RoI

- usually not fast enough to be run right at the L1 output (need a L2-like preselection)

• FTK:• Full scan of the tracking system right at the L1

output. Available in 2016…

Tau Trigger Run2 Performances

20

• Better performances than Run1

• Relative Isolation at the L1

• Cell-level precision available at the L1 output rate (EF/L2 merged)

• HLT/offline ID menu harmonized

Efficiency wrt offline identified taus

Tau Trigger Run2: Efficiency Measurement

21

• Tag&Probe analysis using Z→ττ→μτh events: very good agreement with simulation

• Data/simulation scale-factors measured close to unity.

Tau Particle Flow

• Identify and classify the different hadronic decay modes:

• Crucial information for decay-dependent measurements

• Improve the energy resolution

22

1. Reconstruct and identify individual neutral pions within the tau energy deposit:

•Estimate EM energy from charged pions using Etrk-EHAD. Combine EM shower shape variables into a BDT do discriminate π0 from pile-up and h± remnants cluster.

2. Using the first layer of the ECAL, attempt to disentangle cases where two neutral pions share the same cluster.

3. Combine with the charged information from the tracking system to classify the tau and measure its kinematic properties

Tau Particle Flow: Classification

• Efficient separation of decays with and without neutral pions

• A lot of 1p2n to 1p1n migration

23

89.7 16.0 4.3 1.2 0.3

9.4 74.8 56.3 0.9 2.5

0.4 6.0 35.4 0.1 0.4

0.2 0.6 0.3 92.5 40.2

0.2 2.5 3.6 5.3 56.6

Generated decay mode

±h 0π ±h 0π2≥ ±h ±h3 0π1≥ ±h3

Rec

onst

ruct

ed d

ecay

mod

e

±h

0π ±h

0π2≥ ±h

±h3

0π1≥ ±h3

ATLAS Simulation Tau Particle Flow

ττ→*γ/ZDiagonal fraction: 74.7%

Reconstructed decay mode

±h 0π ±h 0π2≥ ±h ±h3 0π1≥ ±h3

Even

ts /

Dec

ay m

ode

0

2000

4000

6000ATLAS )-1Data (8 TeV, 5.0 fb

)0π1≥ ±h (3ττ→*γ/Z)±h (3ττ→*γ/Z

)0π2≥ ±h (ττ→*γ/Z)0π ±h (ττ→*γ/Z

)±h (ττ→*γ/ZBackgroundStat. Uncertainty

arXiv:1512.05955

Tau Particle Flow: Energy Resolution

24

[GeV]genTE

20 40 60 80 100

reso

lutio

nT

ER

elat

ive

0

0.1

0.2

0.3

0.4

ττ→*γ/ZATLAS Simulation Tau Particle Flow

Constituent-basedBaselineCore resolutionTail resolution

• Decay classification allows a more reliable use of the tracking information to measure pT

• Huge gain in the core of the distribution

• Poor resolution in mis-classified decays, need to combine with calo-only (run1 approach) method

arXiv:1512.05955

Tau detection at ATLAS

• Run1 tau reconstruction was very successful and used for key ATLAS analysis

• Tau Energy Scale is measured within a few %

• Tau Identification algorithms have a good efficiency of (60% for the medium WP) with a powerful jet suppression allowing for tau final states to be detected

• Run2 tau reconstruction:

• Successful commissioning (both offline and trigger)

• Target to validate and use the new Tau Particle Flow strategy in 2016

25

Higgs coupling to fermions: H→ττ

• Large branching ratio

• Presence of leptons in the Final States (less background than H→bb)

• Only leptonic decay reachable by the LHC (H→μμ should be visible in HL-LHC)

• All the production modes reachable by the LHC

26

Does the 125 GeV boson decay to 𝜏-lepton pairs ?

27

BDT output-1 -0.5 0 0.5 1

Even

ts /

0.4

1

10

210

310

410Data

=1.4)µ(125) (H=1)µ(125) (H

ττ →ZOthers

τFake Uncert.

VBFhadτhadτ

ATLAS Internal

-1, 20.3 fb = 8 TeVs

BDT output-1 -0.5 0 0.5 1D

ata

/ Mod

el

0.51

1.5BDT output-1 -0.5 0 0.5 1

Even

ts /

0.17

10

210

310

410 Data=1.4)µ(125) (H=1)µ(125) (H

ττ →ZOthers

τFake Uncert.

VBFhadτe + hadτµ

ATLAS Internal

-1, 20.3 fb = 8 TeVs

BDT output-1 -0.5 0 0.5 1D

ata

/ Mod

el

0.51

1.5BDT output-1 -0.5 0 0.5 1

Even

ts /

0.11

1

10

210

310

410Data

=1.4)µ(125) (H=1)µ(125) (H

ττ →ZOthersFake leptonUncert.

VBFµµ + µe + ee

ATLAS Internal

-1, 20.3 fb = 8 TeVs

BDT output-1 -0.5 0 0.5 1D

ata

/ Mod

el

0.51

1.5

Had-Had

)2τ, 1τ(R∆2 4

Even

ts /

0.2

0

50

100

150

200

250

300

350

Data(125)H50 x

ττ →Z+single-toptt

OthersτFake

Uncert.

VBFhadτe + hadτµ ATLAS Internal

Pre-fit-1, 20.3 fb = 8 TeVs

Lep-HadLep-Lep

Z→𝜏𝜏: 𝜏-embedded Z→μμ data

Top+EWK: MC simulations

jet-faking-𝜏h: Data Driven

• Event categorized to target VBF and ggH production modes separately, all tau-decays considered

• MVA analysis to exploit most of the kinematic properties: • Boosted Decision Trees trained to separate

the background from H(125)→𝜏𝜏 • Particularly powerful in the VBF category

τ-embedded data• Z(→ll)+jets measurements: large data/MC modelling

• Require a sophisticated reweighing to model Z(→ττ)+jets

• Using τ-embedded Z(→μμ)+jets data:

• very good modelling

28

[GeV]missTE

0 20 40 60 80 100 120 140 160

Eve

nts

/ 5

GeV

0

100

200

300

400

500 Data

embeddingττ→Z

<40 GeV)ττ

MC (mττ→*

γ/Z

Others

Uncert.

-120.3 fb = 8 TeV, s

Boostedττ ATLAS

ττ→ZControl Region

[GeV]missTE

0 20 40 60 80 100 120 140 160

Data

/Model

0.60.8

11.21.4

2015 JINST 10 P09018

|jj y

∆/d

|σ

) d- l+ l→* γ

Z/σ

(1/

-410

-310

-210

-110 = 7 TeV)sData 2011 (ALPGENSHERPA

+ SHERPAATHLACKB

ATLAS )µ 2 jets (l=e,≥)+ -l+ l→*(γZ/-1 L dt = 4.6 fb∫

jets, R = 0.4tanti-k| < 4.4jet > 30 GeV, |yjet

Tp

0 1 2 3 4 5 6

NLO

/ D

ata

0.60.8

11.21.4 + SHERPAATHLACKB

0 1 2 3 4 5 6M

C /

Dat

a

0.60.8

11.21.4 ALPGEN

| (leading jet, 2nd leading jet)jj y∆|0 1 2 3 4 5 6

MC

/ D

ata

0.60.8

11.21.4 SHERPA

MVA• Run1 H→ττ heavily relies on MVA techniques

• Improves the sensitivity from 2.5σ (cut-based) to 3.5σ

• BDT trained to separate H(125) from the background, but also MissingMassCalculator, offline (and online!) BDT for tau ID, …

• Higgs ML challenge

• New techniques could be investigated:

• Deep Neural Networks are becoming more and more popular (winner of the Higgs ML challenge).

• Potential applications in ATLAS: energy calibration, particle ID, jet substructure, sig/bkg separation…

• Matrix Element techniques: expensive computing needs are becoming reachable with GPU clusters

29

Run 1 Analysis: results

30

µ = 1.42+0.27� 0.26 (stat.)

+0.33� 0.25 (syst.) ± 0.1 (theory syst.)

- Observed (expected) significance: 4.5 (3.5)𝝈 - Measured signal strength consistent with the SM in all channels/categories

JHEP 04 (2015) 117

ATLAS+CMS Higgs combination

• Combining with CMS, claim discovery (>5σ) of the tau coupling to the 125 GeV Higgs

• H→ττ brings strong constraints on kF

• Similar constraint to H→WW to the VBF production mode.

31

ATLAS-CONF-2015-044

H→ττ prospectives: Run2 and beyond

• Extension of the SM:

• high mass search in the ττ final states

• Lepton Flavour Violation

• Test the SM with new and more precise Higgs property measurements

• Observation of the VH production mode

• Inclusive and differential cross-section

• Hff CP quantum numbers

• HVV and ggH couplings in VBF

• Higgs self-coupling

32

H→ττ prospectives: Run2 and beyond

• Extension of the SM:

• high mass search in the ττ final states

• Lepton Flavour Violation

• Test the SM with new and more precise Higgs property measurements

• Observation of the VH production mode

• Inclusive and differential cross-section

• Hff CP quantum numbers

• HVV and ggH couplings in VBF

• Higgs self-coupling

33

Search for High Mass Neutral Higgs

• First ATLAS run2 results with taus

• Many extensions of the SM predict an extended Higgs sector

• MSSM favours down-type fermion couplings at high tanβ

• ggH and associated-b productions are considered in this search

34

ATLAS-CONF-2015-061

g

g

h/H/A

g

g b̄

b

h/H/A

ggH

bbH

Search for High Mass Neutral Higgs

35

• Search performed in the hadronic channels

• Fake background estimated using data-driven techniques

• Z+jets, top and diboson from simulation

ATLAS-CONF-2015-061

Search for High Mass Neutral Higgs

36

[GeV]Am200 300 400 500 600 700 800 900 1000 1100 1200

βta

n10

20

30

40

50

60

70

80

= 3

00 G

eVH

m

= 8

00 G

eVH

m

= 1

000

GeV

Hm

ObservedExpectedσ1 σ2

ATLAS Run-I (Obs.)ATLAS Run-I, SM Higgsboson couplings (Obs.)

/////////

-1=13 TeV, 3.2 fbsPreliminary, ATLAS hMSSM scenario

, 95% CL limitsττ →H/A • In MSSM mh

mod+, the strongest constraint is tanβ>10 for mA = 200 GeV

• For mA > 700 GeV, the run1 result is superseded.

ATLAS-CONF-2015-061

Lepton Flavour Violation: H→τμ

• LFV couplings can occur in many SM extensions

• BR(H→eμ) < O(10-8): strong constraints from μ→eγ searches

• BR(H→τμ) < O(10%): loosely constrained from τ→μγ searches

• CMS H→τμ search has a slight excess of 2.4σ. They exclude BR(H→τμ) < 1.57% at 95% CL

37

Lepton Flavour Violation: H→τμ

• Analysis performed in the τhμ final state

• Background estimation similar to the H→ττ analysis, dedicated tune of the mass estimator

• Slight excess corresponding to 1.3σ

• BR(H→τμ) < 1.85% at 95% CL. Weaker than CMS limit but no combination (yet) with the eμ channel.

38

JHEP11(2015)211

H→ττ prospectives: Run2 and beyond

• Extension of the SM:

• high mass search in the ττ final states

• Lepton Flavour Violation

• Test the SM with new and more precise Higgs property measurements

• Observation of the VH production mode

• Inclusive and differential cross-section

• Hff CP quantum numbers

• HVV and ggH couplings in VBF

• Higgs self-coupling

39

Higgs properties: VH→ττ

• Search carried in hadronic channels VH→ττ, still completely stat-dominated

• WH→τhτl is the most sensitive channel (also the least affected by Fakes)

• Run2 dataset should provide observation/exclusion of VH→ττ at the SM expected rate

40 = 125 GeVHm at SMσ/σ95% CL limit on

0 2 4 6 8 10 12 14 16 18 20

Combination

)hτlτ→(H) νl→(W

)hτhτ→(H) νl→(W

)hτlτ→(H) ll→(Z

)hτhτ→(H) ll→(Z

ObservedExpectedσ1±σ2±

ATLAS-1 = 8 TeV, 20.3 fbs√

[GeV]2TM0 20 40 60 80 100 120 140 160

Even

ts /

20 G

eV

0

2

4

6

8

10

12

14

16

18 Data (125 GeV)WH

Fake Factor BGWZOthersZZsyst. + stat.

ATLAS

-1 = 8 TeV, 20.3 fbs√

)hτlτ→(H) νl→(W

arXiv:1511.08352

Higgs properties: incl. and diff. cross-sections

• Get the 5σ (inclusively and per production mode)!

• Strong VBF signal in H→ττ

• Good place to measure fiducial corrections that still have large O(6-10%) systematic uncertainties

• Probe High pT regime of the Higgs boson

• H+jets: Break degeneracy with a high-pt jet that resolves the top loop but not the NP-loop

41

arxiv:1405.4295

Higgs properties: HVV vertex• Probe HVV tensor structure by exploiting large

VBF signal

• Run1 analysis in preparation:

• Cut on the BDT score to select high VBF Higgs purity events

42

τ±

τ∓

HV

V

q

q

q

q

BDT output-1 -0.5 0 0.5 1

Even

ts /

0.17

10

210

310

410 Data=1.4)µ(125) (H=1)µ(125) (H

ττ →ZOthers

τFake Uncert.

VBFhadτe + hadτµ

ATLAS Internal

-1, 20.3 fb = 8 TeVs

BDT output-1 -0.5 0 0.5 1D

ata

/ Mod

el

0.51

1.5

Proof of principle, analysis to be repeated

with Run2 data

Higgs properties: Hff vertex• Test CP of the Yukawa Hff coupling

• Access through spin correlation and angular distribution of tau decay products

• New Tau Particle flow will bring:

• Mode-specific angle calculation. For example exploit the τ→ρ decay mode

43

τ±

τ∓

HV

V

q

q

q

q

DiHiggs: HH→bbττ

• Sensitivity to the SM not expected before HL-LHC dataset

• DiHiggs search carried in bbWW, bbγγ, bbττ and bbbb final states

• Still room for a lot of improvements with HL-LHC upgrade

• 1MHz L0, L1 tracking

• High eta coverage

44

[GeV]ττm

0 100 200 300 400 500E

vents

/ 2

5 G

eV

0

20

40

60

80

100

120

140

160

180

200

220

Data

Top quark

+jetsττ→Z

Others

τFake

Systematics

Non-reso. hh(10 pb)

ATLAS-1 = 8 TeV, 20.3 fbs

hadτ + ehad

τµ

Analysis ��bb ��WW ⇤ bb⌧⌧ bbbb Combined

Upper limit on the cross section [pb]Expected 1.0 6.7 1.3 0.62 0.47Observed 2.2 11 1.6 0.62 0.69

Upper limit on the cross section relative to the SM predictionExpected 100 680 130 63 48Observed 220 1150 160 63 70

Analysis ��bb ��WW ⇤ bb⌧⌧ bbbb Combined

Upper limit on the cross section [pb]Expected 1.0 6.7 1.3 0.62 0.47Observed 2.2 11 1.6 0.62 0.69

Upper limit on the cross section relative to the SM predictionExpected 100 680 130 63 48Observed 220 1150 160 63 70

Phys. Rev. D 92, 092004 (2015)

Things that go bump in the light*

• Deviation from the SM expectations observed around 750 GeV in the diphoton spectrum

• Using a narrow width resonance, the local p0 is estimated at 3.6σ (2.0σ global)

• Width ~ 45 GeV

• Can we see it in the ditau final state?

45*In Particular

[GeV]γγm200 400 600 800 1000 1200 1400 1600

Even

ts /

40 G

eV1−10

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

Conclusion

• LHC Run1 program was very successful for tau physics

• Observation of Higgs coupling to the tau lepton by ATLAS and CMS

• Successful energy upgrade of the LHC in 2015

• Commissioning the tau trigger, validating the offline/online algorithms

• Preliminary results supersede the Run1 limits on the MSSM Higgs at high tanβ

• The 2016 dataset will allow for more precise measurements in H→ττ

• Stay tuned!

46

Additional Material

47

Tau Energy Scale Uncertainties

• Tau Energy Scale uncertainties are calculated using the deconvolution method

• Uncertainties on particle energies taken from E/p measurements in low mu runs, test beam studies and simulation are propagated to the constituents of hadronic tau decays

48