[PPT] - Searches for Lepton flavour violating H / Z l decays with the PowerPoint Presentation

SLIDE 1

Searches for Lepton flavour violating H/Z → τl decays with the ATLAS detector at 8 TeV

On behalf of ATLAS collaboration Hartger Weits (NIKHEF) April 11, 2016

1 / 28

SLIDE 2

Lepton Flavour Violation is an established fact

☞ 2001 at Sudbury Neutrino Observatory ☞ nobel prize 2015: for the discovery of neutrino oscillations, which shows that neutrinos have mass ☞ neutrino mixing can be incorporated by introducing PMNS matrix

  νe νµ ντ   = VPMNS   ν1 ν2 ν3   ☞ This makes LFV Z & H decays possible:

D1:

νi νj W l1 l2

D2:

W W νi l1 l2 l l

☞ However, prediction νSM of BF(Z → τl) ∼ 10−54 [1]

2 / 28

SLIDE 3

Collider experiments well suited for production of leptons

most sensitive Z → τl searches stem from LEP

◮ Br(Z → τµ) < 1.2×10−5,

Br(Z → τe) < 9.8×10−6 [2, 3]

◮ they had a cleaner environment, we have more statistics

DELPHI Interactive Analysis

Run: 26154 Evt: 2958 Beam: 45.6 GeV Proc: 1-Oct-1991 DAS : 25-Aug-1991 21:46:38 Scan: 4-Dec-1992 TD TE TS TK TV ST PA Act Deact 1 ( 37) ( 0) 35 ( 35) ( 1) ( 0) ( 0) 2 ( 4) ( 3) ( 0) ( 0) ( 0) ( 0) ( 0) ( 0) X Y Z

H → τl new measurement

◮ CMS found 2.4σ excess :

Br(H → τµ) = 0.84+0.39

−0.37 % [4]. ◮ no excess in electron channel: Br(H → τe) < 0.7 % (preliminary results [5])

3 / 28

SLIDE 4

Search for H/Z → eτ/µτ decays in the τhadchannel

Missing Mass Calculator [6] MMMC

τl

: invariant mass of the Z or H quadratic equation p2

z,ν +αpz,ν +β = 0

most likely solution L = P(∆R)×P( E T)

R Δ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Events 1 10

2

10

3

10

4

10 τ τ → Z decays τ 1-prong R Δ 0.1 0.2 0.3 0.4 0.5 Events 1 10

2

10

3

10 τ τ → Z decays τ 3-prong

4 / 28

SLIDE 5

Data–driven methods & Monte Carlo corrections

Data–driven Z → ττ : ☞ from Z → µµ QCD multi-jets: ☞ OS/SS symmetry Control Regions t/t¯ t W +jets Z/VV → ll

[GeV]

miss T

E , e T

m

20 40 60 80 100 120 140

[GeV]

miss T

E ,

had

τ T

m

20 40 60 80 100 120 140

Fraction of Events

0.5 1 1.5 2 2.5 3

3

10 ×

+ jets τ τ → Z Simulation Preliminary, ATLAS

[GeV]

miss T

E , e T

m

20 40 60 80 100 120 140

[GeV]

miss T

E ,

had

τ T

m

20 40 60 80 100 120 140

Fraction of Events

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

3

10 ×

+ jets W Simulation Preliminary, ATLAS

[GeV]

miss T

E , e T

m

20 40 60 80 100 120 140

[GeV]

miss T

E ,

had

τ T

m

20 40 60 80 100 120 140

Fraction of Events

0.5 1 1.5 2 2.5

3

10 ×

τ e → 125 H Simulation Preliminary, ATLAS

[GeV]

miss T

E , e T

m

20 40 60 80 100 120 140

[GeV]

miss T

E ,

had

τ T

m

20 40 60 80 100 120 140

Fraction of Events

0.2 0.4 0.6 0.8 1 1.2 1.4

3

10 × SR2 SR1 WCR

1

L dt = 20.3 fb

∫

= 8 TeV s Preliminary, ATLAS

5 / 28

SLIDE 6

Z → µτhad[7]

Events / 5 GeV

0.5 1 1.5 2 2.5 3 3.5

3

10 ×

Data

3

BR=10 µ τ → Z + jets (OS-SS) τ τ → Z + jets (OS-SS) W Same Sign Others

Syst. Unc.

100 150

ATLAS Preliminary

events

had

τ µ SR1

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

τ µ MMC

m

Data / BKG

0.6 0.8 1 1.2 1.4 100 150

Events / 5 GeV

1 2 3 4 5

3

10 ×

Data

3

BR=10 µ τ → Z + jets (OS-SS) τ τ → Z Same Sign + jets (OS-SS) W Others

Syst. Unc.

100 150

ATLAS Preliminary

events

had

τ µ SR2

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

τ µ MMC

m

Data / BKG

0.6 0.8 1 1.2 1.4 100 150

◮ Br(Z → τµ) = −1.6+1.3 −1.4 ×10−5, best fit value ◮ Br(Z → τµ) < 1.69(2.58)×10−5, observed (expected) 95 % C.L

6 / 28

SLIDE 7

H → µτhad[8]

◮ Br(H → τµ) = 0.77±0.66 %, best fit value ◮ Br(H → τµ) < 1.85(1.24) %, observed (expected) 95 % C.L.

7 / 28

SLIDE 8

H → eτhad[7]

◮ Br(H → τe) = −0.471.08 −1.18 %, best fit value ◮ Br(H → τe) < 1.81(2.07) %, observed (expected) 95 % C.L.

8 / 28

SLIDE 9

Search for H → eτ/µτ decays in the τlepchannel

ℎ

ℓ ℓ’

ℓ’

The final discriminant used in this channel is the collinear mass mcoll defined as: mcoll =

2pℓ1

T

pℓ2

T +E miss T

(cosh∆η −cos∆φ).

(1) This quantity is the invariant mass of two massless particles, τ and l1, computed with the approximation that the decay products of the τ lepton, l2 and ν, are collinear to the τ, and that the E miss

T

riginates from the ν.

9 / 28

SLIDE 10

H → eτlep/µτlep [7]

Dilepton events are divided into two mutually exclusive samples:

◮ µe sample: pµ T ≥ pe T

: H → µτ → µeνν would be here

◮ eµ sample: pe T > pµ T

Events/10 GeV 1 10

2

10

3

10

noJets

SR µ Data e

Symm. background
Tot. background

Post-fit uncertainty

ATLAS Preliminary

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

coll

m

50 100 150 200 250 300 350 400 450

Data/Bkg

0.5 1 1.5 2

Events/10 GeV 1 10

2

10

3

10

noJets

e SR µ Data

Symm. background
Tot. background

Post-fit uncertainty (BR=1%) τ µ → H

ATLAS Preliminary

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

coll

m

50 100 150 200 250 300 350 400 450

Data/Bkg

0.5 1 1.5 2

◮ Br(H → µτ) < 1.79(1.73) %, Br(H → eτ) < 1.36(1.48) %

10 / 28

SLIDE 11

Combined Results

), % τ e → H 95% CL upper limit on Br( 2 4 6 8 10 12 14

ATLAS Preliminary

1

L dt = 20.3 fb

∫

= 8 TeV s , Comb τ e , Comb

lep

τ e

withJets

, SR

lep

τ e

noJets

, SR

lep

τ e , Comb

had

τ e , SR2

had

τ e , SR1

had

τ e

σ 1 ± Expected σ 2 ± Expected Observed Excluded

), % τ µ → H 95% CL upper limit on Br( 2 4 6 8 10 12 14

ATLAS Preliminary

1

L dt = 20.3 fb

∫

= 8 TeV s , Comb τ µ , Comb

lep

τ µ

withJets

, SR

lep

τ µ

noJets

, SR

lep

τ µ , Comb

had

τ µ , SR2

had

τ µ , SR1

had

τ µ

σ 1 ± Expected σ 2 ± Expected Observed Excluded

◮ Combined result: Br(H → µτ) < 1.43(1.01) %, Br(H → eτ) < 1.21(1.48) %

11 / 28

SLIDE 12

Complementary low energy decay: τ → 3µ [9]

ℓi ¯ ℓk ℓk ℓj Z

¯ [MeV]

µ 3

m

1500 1600 1700 1800 1900 2000 2100

Events / 30 MeV

5 10 15 20 25

ATLAS

1

=8 TeV, 20.3 fb s

selection) Data (tight+x>x selection)

1

Data (tight+x>x Fit to the SB data Fit uncertainty selection) Signal (tight+x>x Sidebands (SB) Signal region

◮ trained BDT, predict event count from sidebands invariant mass m3µ ◮ Br(τ → 3µ) < 3.76×10−7(3.94×10−7) observed (expected) at 90% C.L.

12 / 28

SLIDE 13

Conclusion

◮ LHC offers a new opportunity to look for charged lepton flavour violating decays ◮ interesting from the standpoint of new physics models w.r.t. neutrino oscillations

→ unambiguous sign of new physics

◮ several searches1 have been performed at ATLAS with different techniques

H/Z → lτhad : template fit using MMMC

τl

H → lτlep : completely data–driven technique on symmetry argument τ → 3µ : counting experiment after BDT selection Z → eµ : bump hunting

◮ no significant excess found ◮ determining more Higgs properties at ATLAS ◮ Z → τµ will be competitive with LEP after Run 2 and/or τlep ◮ τ → 3µ: expected to be competitive with Belle result with Run2 data and trigger

improvement

1Z → eµ is an older analysis,see backup, most sensitive limit

13 / 28

SLIDE 14

References I

J I Illana, M Jack, and Tord Riemann. Predictions for Z → µτ and Related Reactions,hep-ph/0001273. desy-99-165. ug-ft-112. lc-th-2000-007. (hep-ph/0001273. DESY-99-165. UG-FT-112. LC-TH-2000-007):34 p, Jan 2000.

P. Abreu et al.

Search for lepton flavor number violating Z0 decays.

Z. Phys., C73:243–251, 1997.
R. Akers et al.

A Search for lepton flavor violating Z0 decays.

Z. Phys., C67:555–564, 1995.

Vardan Khachatryan et al. Search for Lepton-Flavour-Violating Decays of the Higgs Boson.

Phys. Lett., B749:337–362, 2015.

14 / 28

SLIDE 15

References II

CMS Collaboration. Search for lepton-flavour-violating decays of the Higgs boson to etau and emu at sqrt(s)=8 TeV. 2015.

A. Elagin, P. Murat, A. Pranko, and A. Safonov.

A New Mass Reconstruction Technique for Resonances Decaying to di-tau. Nucl.Instrum.Meth., A654:481–489, 2011. Robert Clarke et al. Search for lepton flavour violating decays of the Higgs and Z bosons with the ATLAS detector. Technical Report ATL-COM-PHYS-2015-1362, CERN, Geneva, Nov 2015.

15 / 28

SLIDE 16

References III

Georges Aad et al. Search for lepton-flavour-violating H → µτ decays of the Higgs boson with the ATLAS detector. JHEP, 11:211, 2015. Georges Aad et al. Probing lepton flavour violation via neutrinoless τ − → 3µ decays with the ATLAS detector. 2016. Robert H. Bernstein and Peter S. Cooper. Charged Lepton Flavor Violation: An Experimenter’s Guide.

Phys. Rept., 532:27–64, 2013.

Shikma Bressler, Avital Dery, and Aielet Efrati. Asymmetric lepton-flavor violating Higgs boson decays.

Phys. Rev., D90(1):015025, 2014.

16 / 28

SLIDE 17

References IV

Gianluca Blankenburg, John Ellis, and Gino Isidori. Flavour-Changing Decays of a 125 GeV Higgs-like Particle.

Phys. Lett., B712:386–390, 2012.
P. Abreu et al.

A Search for lepton flavor violation in Z0 decays.

Phys. Lett., B298:247–256, 1993.

Georges Aad et al. Search for the lepton flavor violating decay Ze in pp collisions at √sTeV with the ATLAS detector.

Phys. Rev., D90(7):072010, 2014.

17 / 28

SLIDE 18

Backup

18 / 28

SLIDE 19

Lepton Flavour Violation

Isidor Isaac Rabis famous question about the muons existence, Who

rdered that?, was prescient and deep. His question, in modern terms, asked

why are there flavours and generations? Why are there muons and taus in addition to the electron? The same question applies to the quark and neutrino sectors. We believe there are three generations in each sector, and that the number in each sector must be the same. We see quarks changing generations, as codified in the Cabibbo-Kobayashi-Maskawa matrix, and neutrinos changing from muon to electron to tau neutrinos according to the Pontecorvo-Maki-Nakagawa-Saka matrix. Lepton Flavour Violation is an established fact, but only in the neutral neutrinos. What about their charged partners? Is there Charged Lepton Flavour Violation? [10]

19 / 28

SLIDE 20

Control regions

Events / 5 GeV

50 100 150 200 250 300

Data 2012

3

BR=10 µ τ → Z + jets (OS-SS) W Same Sign Others (OS-SS)

Syst. Unc.

100 150 200

ATLAS Preliminary

events

had

τ µ WCR

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

τ µ MMC

m

Data / BKG

0.6 0.8 1 1.2 1.4 100 150 200

Events / 5 GeV

20 40 60 80 100 120 140

Data 2012

3

BR=10 µ τ → Z Top (OS-SS) Same Sign Others (OS-SS)

Syst. Unc.

100 150 200

ATLAS Preliminary

events

had

τ µ TCR

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

τ µ MMC

m

Data / BKG

0.6 0.8 1 1.2 1.4 100 150 200

Events / 1 GeV

100 200 300 400 500 600

Data 2012

3

BR=10 µ τ → Z (OS-SS) τ → j µ µ → Z Others (OS-SS)

Syst. Unc.

85 90 95 100

ATLAS Preliminary

events

had

τ µ ZmmCR

1

L dt = 20.3 fb

∫

= 8 TeV s

[GeV]

µ

+

µ

m

Data / BKG

0.6 0.8 1 1.2 1.4 85 90 95 100

◮ a jet faking a τhad is not well modelled by MC simulation

20 / 28

SLIDE 21

SM backgrounds Z → ττ t/t¯ t W +jets Z/VV → ll H → ττ QCD multi-jets Event classes real lepton and τhad jet misidentified as a τhad lepton misidentified as a τhad Processes where a jet fakes a τhad are not well modelled by Monte Carlo simulation, we use the following assumptions:

◮ The shape of the MMMC τl

distribution in the signal regions is the same for OS and SS events for the QCD multi-jet background.

◮ The scale factor k = N(data)/N(MC) is the same for the processes in the signal

and corresponding control regions for the electroweak backgrounds.

21 / 28

SLIDE 22

The symmetry method [11] is completely data–driven, has very few background systematics and mostly limited by statistics. It is based on the following two premises:

1. Standard Model processes are to a good approximation symmetric under the

exchange e ↔ µ [11].

2. Br(H → µτ) = Br(H → eτ) 2

Dilepton events are divided into two mutually exclusive samples:

◮ µe sample: pµ T ≥ pe T

: H → µτ → µeνν would be here

◮ eµ sample: pe T > pµ T

: H → eτ → eµνν would be here small asymmetries that need to be accounted for:

◮ misidentified and non-prompt leptons ◮ trigger and reconstruction efficiency

2The bound on µ → eγ decays suggests that the presence of a H → µτ signal would exclude the

presence of a H → eτ signal, and vice versa, at an experimentally observable level at the LHC [12].

22 / 28

SLIDE 23

Z → eµ [14]

[GeV]

µ e

m 70 75 80 85 90 95 100 105 110 Events / 2 GeV 100 200 300 400 500 600 700

MC stat. error µ µ ee/ → Z τ τ → Z Multijet W Diboson Top Data µ e → Z

5

10 × B = 1.0

1

= 8 TeV, 20.3 fb s

ATLAS

70 75 80 85 90 95 100 105 110 Events / GeV 50 100 150 200 250 300

Data Fit

7

10 × B = 7.5

1

= 8 TeV, 20.3 fb s

/DOF = 0.75

2

χ ATLAS [GeV]

µ e

m 70 75 80 85 90 95 100 105 110

Data - Fit

20
10

10 20

Br(Z → eµ) < 7.5×10−7, observed 95 % C.L., significantly more restrictive than that from the LEP experiments [13].

23 / 28

SLIDE 24

Reconstructing τ

◮ hadronic: 1 ν, leptonic: 2 ν ◮ missing energy → sideband analysis not possible ◮ performing a template fit

24 / 28

SLIDE 25

FitModel

Nbkg

OS = rQCD ·Ndata SS

+NZ→ττ

add-on +NW +jets add-on +Nt/t¯ t add-on +NVV →ll add-on +NH→ττ add-on

+NZ→ll(l→τfake)

add-on

+NZ→ll(j→τfake)

add-on

, where the ratio rQCD = NQCD

OS /NQCD SS

accounts for the rate difference in QCD multi-jets when requiring OS or SS events, which is caused by their different flavour composition.

◮ NW +jets add-on = kOS W +jets ·NW +jets OS

−rQCD ·kSS

W +jets ·NW +jets SS

. Because the W +jetsbackground consists of a jet misidentified as a τhad, a rate correction is

applied. The quark and gluon composition differ for OS and SS events causing

some charge asymmetry NOS > NSS. Therefore two separate corrections are

btained from a control region, namely kOS

W +jets and kSS W +jets.

25 / 28

SLIDE 26

QCD multi-jet: data driven

◮ virtually impossible to model by Monte Carlo simulation ◮ make use of symmetry between same and oppositely charged τhadl events

Events 200 400 600 800 1000 1200 1400 1600 channels

had

τ + µ

1

L=20.3fb

∫

= 8 TeV, s

Opposite Sign(OS) Same Sign(SS) visible [GeV]

τ τ

m

20 40 60 80 100 120 140

(OS/SS)

QCD

r

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5

Events 1000 2000 3000 4000 5000 channels

had

τ e+

1

L=20.3fb

∫

= 8 TeV, s

Opposite Sign(OS) Same Sign(SS) visible [GeV]

τ τ

m

20 40 60 80 100 120 140

(OS/SS)

QCD

r

0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5

26 / 28

SLIDE 27

Signal: Z → τhadl

Figure: Artist’s Impression

☞ pT(µ) ∼ mZ/2 ☞ µτhad back to back ☞ ντhad ∼ collinear

) [GeV] µ ( T p 10 20 30 40 50 60 70 80 90 100 100 200 300 400 500 Entries 5659 Mean 38.64 RMS 12.42 Underflow Overflow 46

µ τ → Z

) had τ , µ R( ∆ 1 2 3 4 5 6 200 400 600 800 1000 1200 Entries 5659 Mean 3.16 RMS 0.5508 Underflow Overflow 1

µ τ → Z

) τ ν , had τ R( ∆ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 200 400 600 800 1000 1200 Entries 5659 Mean 0.1125 RMS 0.1048 Underflow Overflow 34

µ τ → Z

27 / 28

SLIDE 28

Reconstructing Z

R Δ 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Events 1 10

2

10

3

10

4

10 τ τ → Z decays τ 1-prong R Δ 0.1 0.2 0.3 0.4 0.5 Events 1 10

2

10

3

10 τ τ → Z decays τ 3-prong

◮ 2 solutions from quadratic equation ◮ model angle between hadron - neutrino: P(∆R) ◮ model missing energy resolution: P(

E T)

◮ take most likely according L = P(∆R)×P(

E T)

28 / 28