Study of rare and suppressed processes in B meson decays with the ATLAS experiment

processes in B meson decays with the ATLAS experiment

Paolo Iengo (CERN)

Stefanos Leontsinis (NTUA – BNL) On behalf of the

ATLAS Collaboration

Outline

¤  The ATLAS experiment

¤  Detector

¤  Trigger for B Physics

¤  Search for B

àµ

µ

rare decay

¤  Study of the B

àK*(àK

π

) µ

µ

decay

¤  Conclusions

The ATLAS detector

¤  Inner Detector (|η|<2.4)

¤  Silicon pixel, strip & transition radiation

¤  2T solenoidal field

¤  Good mass resolution (60-100 MeV)

¤  S/B discrimination

¤  Limited PID: K/π separation only for p_T<1GeV

¤  Good impact parameter resolution (<100 µm) even with high pile-up

¤  Precise B measurements

¤  Muon Spectrometer (|η|<2.7)

¤  Trigger Chambers: RPC, TGC

¤  Tracking Chambers: MDT, CSC

¤  0.5-2T toroidal field

¤  σ(p

)/p

~ 0.05% p

+ 1.5%

¤  ID dominates for p_T < 100GeV

Trigger for B Physics

¤  ATLAS trigger

¤  Output rate ~400 Hz

¤  Three selection levels

¤  B-Trigger

¤  High p_T trigger priority for general purposes experiments (Higgs, SUSY etc.)

¤  Keep low-p_T trigger thresholds at high luminosity

¤  Total bandwidth is the main limitation

¤  Di-muon trgger

¤  >5 fb

recorded in 2011 (7 TeV)

¤  <µ> = 9.1

¤  bb-production mostly at large |η|

¤  ~150G B0 pairs and 60M B_sàJ/ψ φ

¤  >20 fb

recorded in 2012 (8 TeV)

¤  <µ> = 20

Excellent data quality and data taking efficiency

Analyses presented here based on 2011 data

Search for B 0 s àµ + µ -

¤  Motivation

¤  Process sensitive to FCNC

¤  Helicity suppressed in SM

¤  Powerful probe for New Physics

¤  NP can increase (but also decrease) Br(B_sàµ⁺µ^-)

¤  Recent results consistent with SM expectations

¤  Still room for destructive

interference between NP and SM leading to Br suppression

¤  Data from LHC Run II could give

a final answer

B s àµ µ : Analysis Overview

¤  Unbiassed

¤  Blind analysis

¤  signal region (±300MeV around B⁰_s mass) blinded

¤  Use MC to model data

¤  Background data in sidebands split in two

¤  50% to optimize selection cuts

¤  50% to measure the bkg yield after cuts optimization

¤  Signal/Background discrimination

¤  Multivariated analysis

¤  BDT based on 13 variables

¤  Signal extraction

¤  Event count in signal region

¤  Background estimation from sidebands

¤  Limit extraction with CLs method

ATLAS-CONF-2013-076

B 0 s àµ + µ - : Analysis Overview

¤  1/(4.45±0.38)10^-3

¤  from PDG+LHCb(Phys.Rev.D85(2012)032008)

¤  Dominant systematic uncertainty

€

BR B (

→ µµ ) = N

α

ε

tot

α

ε

⋅

1 N

⋅ f

f

⋅ BR B ( ± → J / ψ K±)

' ( ) )

* + , ,

} 

Additional systematic uncertainties:

}  Data-MC absolute K efficiency: 5%

}  Vertex reconstruction efficiency: 2%

}  K⁺/K^- asymmetry: 1%

¤  Efficiency and acceptance derived from simulations calibrated on data

¤  Systematics taken from data-MC discrepancies

¤  Relative measurement

¤  B^±àJ/ψ K^±

¤  Partial cancellation of systematics

¤  Single event sensistivity (SES) = B_sàµµ which would give 1 signal event in the data sample

B s àµ µ Background Composition

¤  Real muons (continuum):

¤  bbàµµX dominant background

¤  Modeled using data in sidebands

¤  “Fake” muons (decays in flight, punch- throughs):

¤  Bàhh (KK, Kπ, ππ)

¤  “quasi irreducible” due to close topology

¤  BRx(fake rate) ≈ 10^-9, close to SM B_sàµµ

¤  Contribution estimated with MC:

almost negligible

¤  Single muon + “fake” (e.g. BàµKν)

¤  Negligible contribution, outside our search windows

Bàhh reconstructed as µµ

B 0 s àµµ: Bkg Discrimination and BDT selection

¤  Continuum discrimination:

¤  13 best performing discriminating variables chosen by MVA

¤  Pointing angle α_2D among most discriminating ones

¤  BDT shown as most powerful event classifier (trained on MC)

¤  Good Data/MC agreement

¤  Selection optimized in (Δm,q) space

¤  Δm = signal mass window width

¤  q = BDT output (event classifier)

¤  Working point computed as max of P(Δm,q) = ε

/(1/√N

)

¤  ε_sig from MC

¤  N_bkg from sideband data

¤  Working point: q>0.118; |Δm|<121 MeV

Reference channel (B àJ/ψK ) Yield

¤  BDT trained for B

also used on B

¤  Minimize selection systematics

¤  N(B

àJ/ψ K

) extraction

¤  Unbinned maximum likelihood fit

¤  Per-event mass resolution dm in the fit

¤  Main systematics from varying continuum background fit model

N(B

àJ/ψ K

) = 15214±1.1%(stst)±2.4%(syst)

Limit on Br(B 0 s àµ + µ - )

¤  SES = (2.07 ± 26%(stat) ± 12.5%(syst))x10

¤  N

expected in signal window = 6.75

¤  N

in signal window = 6

¤  Upper limit extracted with modified frequentist approach (CLs method)

¤  Br

< 1.5 x 10

@ 95% CL

New results on 2012 data

(20 fb

) coming soon,

including Br(B

àµ

µ

)

B d àK (àK π ) µ µ

¤  Motivation

¤  Exclusive final state for b à s l

l

transition

¤  Only loop-mediated in SM

¤  Br = 1.1x10

¤  Observables sensitive to NP

¤  Lepton forward-backward asymmetry: A_FB

¤  Fraction of longitudinal polarized K^*0: F_L

¤  Marginally affected by hadronic uncertainties

¤  Measurement: differential angular distribution of the 4-particle final state

as a function of the di-muon mass (q

)

B 0 d àK *0 (àK + π - )µ + µ - Analysis Strategy

¤  Kinematic observables:

¤  Differential decay rate

¤  3 angles: θ_L, θ_K, φ

¤  Dimuon invariant mass q² (K^*0 on shell)

¤  Angular distributions are measured

integrating out two variables due to limited statistics

¤  Use φ symmetry à integration on φ

¤  Alternatively integration on cosθ_L and cosθ_K

¤  <A

> and <F

> extraction

¤  Extended unbinned maximum-likelihood fits

¤  1-D fits in bins of q² to mass, θ_L and θ_K

ATLAS-CONF-2013-038

B d àK (àK π )µ µ Signal Yield

¤  Selection cuts

¤  q² < 2 GeV² (limited statistics due to trigger acceptance)

¤  3σ veto on J/ψ and ψ(2S)

¤  8.68 < q² < 10.09 J/ψ àµµ

¤  12.86 < q² < 14.18 ψ(2S)àµµ

¤  Rejection of radiative c-decays and remaining J/ψ and ψ(2S) in tails

¤  |(m_Bd^rec-m_Bd^PDG|) – (m_µµ^rec-m_cc^PDG)| < Δm

¤  B

mass likelihood fit

¤  Cut based selection optimized on MC

¤  Gaussian (with per-event errors) for signal

¤  Exponential for background

¤  K^*0(àK⁺π^-)accepted in mass range [846,946] MeV

¤  Result

¤  N

= 466±34

¤  N

= 1132±43

B 0 d àK *0 (àK + π - )µ + µ - Likelihood Fit

¤  Extended unbinned max-likelihood fit:

¤  Performed sequentially

¤  Fit mass distribution with fixed signal yield

¤  Simultaneous unbinned mass-angular fit with PDF mass parameters and signal fraction from previous step

¤  Separately in each six q

bins

B d àK (àK π )µ µ Results

¤  Measurements in agreement with SM and other experiments

¤  Accuracy comparable to the best previous measurement in the region q

up to 18 GeV

¤  Statistics uncertainty dominates

(àwill improve with full 2011+2012 dataset)

Conclusions

¤  ATLAS has high quality b-physics program

¤  The large amount of Heavy Flavor data collected is potentially sensitive to New Physics

¤  processes that are naturally suppressed in the Standard Model

¤  Analysis of rare and suppressed decay channels

¤  B

à µ

µ

Br(B

à µ

µ

)

< 1.5 x 10

@ 95% CL

¤  B

à K

µ

µ

à K

π

µ

µ

A

and F

in agreement with SM expectations and previous measurements accuracy comparable to the best previous measurements

¤  Extension of the analyses to the full 2011+2012 (5+20 fb

) data sample

to be completed soon