Searching for B → τ+τ− Decays

Kristof De Bruyn

Rencontres de Physique des Particules 2017Marseille – April 26th, 2017

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 1 / 22

Searching for New Physics at the LHC

LHC Mission Objectives

X Search for the Higgs boson

× Search for evidence ofBeyond the Standard Model Physics

Two Complementary Strategies

Direct Searches (Energy Frontier):

Search for new on-shell resonancesI Bump-hunting

I + Missing energyI + x leptons + y jets + . . .

Indirect Searches (Precision Frontier):

Search for anomalies in SM quantities

I Higgs SectorI (Heavy) Flavour Sector

I CP Violation & CKM MatrixI Rare Decays

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 2 / 22

The LHCb Detector

Forward arm spectrometer to study b- and c-hadron decaysI Pseudo-rapidity coverage: 2 < η < 5

I Good impact parameterresolution to identifysecondary vertices:(15 + 29/pT) µm

I Invariant mass resolution:8 MeV/c2 (B → J/ψX )22 MeV/c2 (B → hh)

I Excellent particleidentification:95 % K ID efficiency(5 % π → K mis-ID)

I Versatile & efficienttrigger for b- andc-hadrons and forwardEW signals

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 3 / 22

Puzzling Tensions in b → s`+`− Transitions

B0 → K∗0µ+µ−

I Angular Observable P ′5

]4c/2 [GeV2q0 5 10 15

5'P

1−

0.5−

0

0.5

1(1

S)ψ/J

(2S)

ψ

LHCb data

Belle data

ATLAS data

CMS dataSM from DHMVSM from ASZB

LHCb, JHEP02 (2016) 104, arxiv:1512.04442

I Tension: 3.4σ

I Discussion on cc̄-contributions

B+ → K+µ+µ−

I Differential branching fractions

]4c/2 [GeV2q0 5 10 15 20

]2/G

eV4 c ×

-8 [

102 q

/dBd 0

1

2

3

4

5

LCSR Lattice Data

LHCb

−µ+µ+ K→+B

LHCb, JHEP06 (2014) 133, arxiv:1403.8044

I Similar effects inB0 → K 0µ+µ− and B+ → K∗+µ+µ−

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 4 / 22

Puzzling Tensions in b → s`+`− Transitions

B+ → K+`+`−

I RK = B(B+→K+µ+µ−)

B(B+→K+e+e−)

SM→ 1

]4c/2 [GeV2q0 5 10 15 20

KR

0

0.5

1

1.5

2

SM

LHCbLHCb

LHCb BaBar Belle

LHCb, PRL 113 (2014) 151601, arxiv:1406.6482

I Tension: 2.6σ

I Test of Lepton Universality

B0 → K∗0`+`− NEW!

I RK∗ = B(B0→K∗0µ+µ−)

B(B0→K∗0e+e−)

0 1 2 3 4 5 6

q2 [GeV2/c4]

0.0

0.2

0.4

0.6

0.8

1.0

1.2

RK∗0

LHCb Preliminary

LHCb

SM from CDHMV

SM from EOS

SM from flav.io

SM from JC

LHCb, LHCb-PAPER-2017-013

I Tensions: 2.2σ and 2.5σ

I Test of Lepton Universality

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 5 / 22

Effective Field Theory Framework

I Model-independent approach: Effective Hamiltonian

Heff =GF√

2

∑j=u,c

V ∗jdVjb︸ ︷︷ ︸CKM Elements

∑k

Ck︸︷︷︸Wilson Coefficient

Operator︷︸︸︷Ok +C ′k O′k

(1)

G. Buchalla et al., RMP 68 (1996) 1125, arxiv:9512380[hep-ph]

Ingredients:

I Coupling constants & CKM elementsI Wilson coefficients contain all perturbative short-distance effects⇒ Can be calculated within perturbation theory

⇒ Are the free parameters when fitting to the data

I Operators contain all non-perturbative long-distance effectsI Electromagnetic operator Oγ7 , important for b → sγ decaysI Semileptonic ops O9 (vector) and O10 (axial-vector), important for b → s`` decaysI Primed operators have mirrored (L ↔ R) chirality

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 6 / 22

Hints for New Physics?

I Best fit model has Wilson coefficientCNP

9 ≈ −1 (4 to 5σ)

I Suggests lepton universality violation

I What can explain this?

1 Statistical fluctuations

2 Not-yet-understood SM effects

3 New Physics

I Way forward?

I Increase the statistics

I Additional observables

I Additional decay channels

Branching Ratios

Angular Observables HPiLAll

-3 -2 -1 0 1 2 3

-3

-2

-1

0

1

2

3

C9NP

C10

NP

S.Descotes-Genon et al., JHEP 06 (2016) 092arxiv:1510.04239

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 7 / 22

Search for B0s → τ+τ− and B0→ τ+τ−

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 8 / 22

Standard Model Decay

I Flavour Changing Neutral Current

I Forbidden at Tree level

⇒ Loop suppressed

I Sensitive to new physics contributions

B0s

s

b

u, c, t

u, c, t

Wγ, Z0, H0

τ−

τ+

B0s

s

b

u, c, t ντ

τ−

τ+

W

W

I Purely leptonic final state makes it theoretically very clean

I Calculated up to NLO EW and NNLO QCD corrections

B(B0s → τ+τ−)

SM= (7.73± 0.49)× 10−7 (2)

B(B0→ τ+τ−)SM= (2.22± 0.19)× 10−8 (3)

Bobeth et al., PRL 112 (2014) 101801, arxiv:1311.0903

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 9 / 22

. . . and Beyond

New Physics Models

I New tree level processes (Z ′, W ′, . . . )

I Additional loop contributions(leptoquarks, 2HDM, . . . )

I Branching fraction can be as large as a %

R. Alonso, arxiv:1505.05164

Experimental Picture

I Previous best limit:

B(B0→ τ+τ−) < 4.1× 10−3 @ 90% C.L.

BaBar, PRL 96 (2006) 241802, arxiv:hep-ex/0511015

I No direct limit on B0s → τ+τ−

I Indirect constraint B(B0s → τ+τ−) < 3%

C. Bobeth and U. Haisch, APP B44 (2013) 127

arxiv:1109.1826

B0s

s

b

Z ′

τ−

τ+

B0s

s

b

X+

W−

tX0

τ−

τ+

B0s

s

b

u, c, t ντ

τ−

τ+

W

X

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 10 / 22

Reconstructing the τ

Available Options:

I Largest Decay Channels

B(τ− → π−π0ντ ) = 25.49± 0.09 %

B(τ− → e−ν̄eντ ) = 17.82± 0.04 %

B(τ− → µ−ν̄µντ ) = 17.39± 0.04 %

B(τ− → π−ντ ) = 10.82± 0.05 %

→ B(τ− → π−π+π−ντ ) = 9.31± 0.05 % ←

B(τ− → π−π0π0ντ ) = 9.26± 0.10 %

LHCb’s Choices

I Challenging to reconstruct electrons and π0 (and limited efficiency)

I Abundant pions

I Only covers 2 < η < 5 ⇒ always missing energy

⇒ Reduced signal efficiency

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 11 / 22

Experimental Signature

B0s → τ+(→ 3π)τ−(→ 3π)

PV

τ+

π+

π−

π+

ν̄τ

τ− π−

π+

π−ντ

B0s

B0 → D+(→ π+K−π+)D−s (→ K−K+π−)

PV

D+

π+

K−

π+

D−s

K−

K+

π−

B0d

Challenges

1 2 missing neutrinosI No narrow (mass) peak to fit,

no mass sidebands to exploitI Cannot differentiate B0

s from B0

2 6 pionsI Low efficiencyI Large combinatorial background

⇒ Need special tools!

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 12 / 22

Custom Tools: Isolation Variables

1 Neutral isolation variablesI Quantify neutral activity (π0, γ, . . . ) in a cone around the B candidate

2 Track isolation variablesI Aims to identify tracks that belong to the same b-hadron decay as the 6 pionsI Based on a BDT trained on simulated signal and b-hadron background

3 Vertex isolation variablesI Aims to identify tracks that make a good quality vertex with selected pions or τI Based on a BDT trained on simulated signal and b-hadron background

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 13 / 22

Analytic Reconstruction Method for B → τ+τ− A. Morda, CERN-THESIS-2015-264

Constraints

I B origin vertex (pp collision)

I τ decay vertices(advantage of 3π decay mode)

I Momentum conservation atB and τ decay vertices

I Masses for the B, τ and ντPV

τ+

π+

π−

π+

ν̄τ

τ− π−

π+

π−ντ

B0s

StrategyI Combine constraints in Lorentz invariant way ⇒ Leads to a fourth order polynomialI Roots provide analytic solutions for the τ momentaI 1 unknown degree of freedom remaining in calculation→ asymmetry (θ) of the triangle PV↔ SV(τ+)↔ SV(τ−)

LimitationsI Approximations required for θ

I Experimental resolution

⇒ Majority of cases no purely real solution can be found: limits applicability

I Use intermediate results to discriminate signal and backgroundKristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 14 / 22

Intermediate Resonances

I Predominantly proceeds through

τ− → a1(1260)−ντ → ρ(770)0π−ντ .

I Exploit this in analysis

Dec

ays

0

5

10

15

20

25

]2c[MeV/ −π+πm200 400 600 800 1000 1200

]2 c[M

eV/

− π+ πm

200

400

600

800

1000

1200

1 2 3

4 5 6

7 8 9

LHCb simulation Subsamples:

I Signal Region [SR]:(τ+ ∈ 5) & (τ− ∈ 5)

I Signal-Depleted Region:(τ+ ∈ 1, 3, 7, 9) ||(τ− ∈ 1, 3, 7, 9)

I Control Region [CR]:(τ± ∈ 4, 5, 8) & (τ∓ ∈ 4, 8)

Selection:

I Cut-based loose selection

I Two-stage neural network

I Optimise size of SR

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 15 / 22

Fit Strategy LHCb-PAPER-2017-003, arxiv:1703.02508

I Perform a 1-dimensional histogram fit to the output of a neural network

I Output is remapped such that signal is flat

I The Signal templates are taken from simulation

I The Background template is taken from data control region

Neural network output0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Frac

tion

of d

ecay

s

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

−τ+τ→0sB

LHCb simulation

Signal regionControl region

Neural network output0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Frac

tion

of c

andi

date

s

3−10

2−10

1−10

1

Control regionLHCb

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 16 / 22

Fit Model LHCb-PAPER-2017-003, arxiv:1703.02508

Events:

Signal: 16% B0s → τ+τ− Simulation versus 7% data

Sig.-Depleted: 13% B0s → τ+τ− Simulation versus 37% data

Control: 58% B0s → τ+τ− Simulation versus 47% data

I . . . so the data control region might also contain signal.

Model:

N SRdata = s × N̂ SR

sim︸︷︷︸Sig PDF

+fb ×(N CR

data − s · εCR

εSR× N̂ CR

sim

)︸ ︷︷ ︸

Bkg PDF

I s: signal yield (free parameter)

I fb: scaling factor for background template (free parameter)

I εi : efficiencies, taken from simulation

I ̂: indicates normalised distributions

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 17 / 22

Fit to Data LHCb-PAPER-2017-003, arxiv:1703.02508

Background-Only Model

Can

dida

tes

1

10

210

310

410

LHCb

DataTotalBackground

Neural network output0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Pull

5−0

5

Nominal Fit Model

Can

dida

tes

1

10

210

310

410

LHCb

DataTotal

Signal×1 −Background

Neural network output0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Pull

5−0

5

s = Nobsτ+τ− = −23+63

−53 (stat)+41−40 (syst)

I Compatible with the background-only hypothesis

I In good agreement with expectations from pseudo-experiments

N toy, SM

τ+τ−= 0+62−40 (stat)+40

−42 (syst)

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 18 / 22

From Yield to Branching Ratio LHCb-PAPER-2017-003, arxiv:1703.02508

B(B0s → τ+τ−) = αs × Nobs

τ+τ−

I Assume all signal comes fromB0

s → τ+τ−

i.e. ignore B0→ τ+τ− completely

I Determine αs using B0→ D−D+s

normalisation mode

]2c[MeV/ +sD−Dm

5000 5100 5200 5300 5400 5500 5600 5700

)2 cC

andi

date

s / (

5 M

eV/

0

200

400

600

800

1000

1200

1400

1600

1800

LHCb

Data+sD−D →0B+sD

−*D →0B

*+sD−D →0B

Comb. bkg.

αs =εD−D+

s × B(B0→ D−D+s )× B(D+ → π+K−π+)× B(D+

s → K+K−π+)

NobsD−D+

s× ετ+τ− × [B(τ− → π−π+π−ντ )]2 × fd

fs

I Fit to data, Efficiencies from simulation, External Input

αs = (4.07± 0.70)× 10−5 → NSMτ+τ− = 0.019

αd = (1.16± 0.19)× 10−5 → NSMτ+τ− = 0.002

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 19 / 22

Branching Fraction Limit LHCb-PAPER-2017-003, arxiv:1703.02508

B0s → τ+τ−

)−τ+τ→0sB(B

0 0.002 0.004 0.006 0.008 0.01

-val

uep

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

LHCb

B0→ τ+τ−

)−τ+τ→0B(B0 0.0005 0.001 0.0015 0.002 0.0025 0.003

-val

uep

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

LHCb

Branching Fraction Limit (CLs Method)

I First limit on B0s → τ+τ−

B(B0s → τ+τ−) < 5.2 (6.8)× 10−3 @ 90 (95) % C.L.

I Improved limit for B0→ τ+τ−

B(B0→ τ+τ−) < 1.6 (2.1)× 10−3 @ 90 (95) % C.L.

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 20 / 22

Future Improvements for B0s → τ+τ−?

I Only scaling with luminosity

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035

Year

10−6

10−5

10−4

10−3

10−2

95%

C.L.

Lim

iton

B(B

0 s→

τ+τ−

)

LS1 LS2 LS3 LS4Run 1 Run 2 Run 3 Run 4 Run 5

SM

Unofficial

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 21 / 22

Conclusion

I Tensions in b → s`+`− transitions motivate studies of decays involving τ leptons

I The search for B → τ+τ− decays at LHCb is challenging, but possible

I First limit on the B0s → τ+τ− branching ratio

B(B0s → τ+τ−) < 6.8× 10−3 @ 95 % C.L.

I Improved limit on the B0→ τ+τ− branching ratio

B(B0→ τ+τ−) < 2.1× 10−3 @ 95 % C.L.

Kristof De Bruyn (CPPM) Searching for B → τ+τ− Decays RPP’17 22 / 22