dream machines: b physics with e + e - colliders

Dream Machines:Dream Machines:B Physics with eB Physics with e++ee-- Colliders Colliders

2007 Fermilab Academic Lecture Series: Flavor Physics2007 Fermilab Academic Lecture Series: Flavor PhysicsJeffrey Berryhill (FNAL) Jeffrey Berryhill (FNAL)

2

OutlineOutline

An Introduction to Y(4S) Experiments and Measurement Techniques

•Dream Machines: Accelerators, Detectors, and the Y(4S) environs

•Exclusive B branching fractions and direct CP violationKinematic constraintsMultivariate background discriminationCase Study: B0 ➔ K+ -

•Time dependent CP violation of exclusive B decaysB decay length differenceB flavor taggingCase Study: B0 ➔ Ks

•B branching fractions to inclusive or neutrino-ful final states Recoil side reconstructionCase Study: B- ➔

3

B meson production with eB meson production with e++ee-- collisions collisions

Y(4S) lightest bb bound state decaying to B mesons Mass = 10.58 GeV (B meson mass = 5.279 GeV) Peak production xsec = 1.1 nb Decays exclusively to B+B- (50%) and B0B0 (50%) B pair produced nearly at rest in Y(4S) frame ( = 0.06)

bb threshold

BB threshold

4

Other production at 10.58 GeVOther production at 10.58 GeV

Final State Xsec (nb)

Events/fb-1

Y(4S) → BB 1.1 1.1 M

uu, dd, and ss 2.0 2.0 M

cc 1.4 1.4 M

1.0 1.0 M

~2.2 million b quarksper fb-1 =

0.55M B+B- + 0.55M B0B0

Y(4S) signal/bkg = 1:4

Also (very low multiplicity): QED ( ee() , () ) Two-photon collisions

A B factory at Y(4S) is also acharm and tau factory!

At 10.54 GeV, “off-resonance” events collected to study or measure background

5

e+3.1 GeV

e-9 GeV

Y(4S)

Y(4S)B0

B0

Asymmetric-energy B FactoriesAsymmetric-energy B Factories

B Meson = 0.56

6

PEP-II at SLAC

KEKB at KEK

Belle

BaBar

9GeV (e) 3.1GeV (e+)peak luminosity:

1.121034cm2s1

Asymmetric-energy B FactoriesAsymmetric-energy B Factories

8GeV (e) 3.5GeV (e+) peak luminosity:

1.651034cm2s1

B Meson = 0.56No crossing angle,dipole separates beams

B Meson = 0.42Beams cross at an angle“crab crossing”

7

PEP-II at the Stanford Linear PEP-II at the Stanford Linear Accelerator CenterAccelerator Center

BaBar detectorBaBar detector

LinacLinac

PEP-II Storage RingsPEP-II Storage Rings

SF Bay ViewSF Bay View

8

PEP-II performancePEP-II performance

Operated continuously since 1999

PEP-II top luminosity: 11.2 x 1033cm-2s-1 (more than 3x design goal 3.0x1033)

Beams carry >1 Amp of current each, injected continuously

Beams collide in a small area of about

100 m X 6 m

In 1 day : created 1.5 million

b quarks !780 million b quarks produced!

Also hundreds of millions each of u, d, s, c, and

Int. Lum. = 391 fb-1

Off resonance = 37 fb-1

9

Integrated Luminosity

PEP-IIfor BaBar

KEKBfor Belle

KEKB + PEP-II

World Y(4S) sample > 2 Billion B mesons

10

The BaBar detectorThe BaBar detector

Cherenkov Detector (DIRC)

144 quartz barsK, separation

Electromagnetic Calorimeter6580 CsI crystals

e+ ID, and reco

Drift Chamber40 layers

Tracking + dE/dx

Instrumented Flux Return19 layers of RPCs (now LSTs) and KL ID

Silicon Vertex Tracker

5 layers of double sided silicon strips

e+ [3.1 GeV]

e- [9 GeV]

e eff. = 90%→ e fake rate = 0.1%

eff. = 90%→ fake rate = 1-2%

K eff. = 80%→ K fake rate = 0.1% eff. = 80%K → fake rate = 1%

(t) ≈ 1 ps

11

Particles producedParticles produced

(Quasi) stable detectable “building blocks”Mesons Leptons Baryons+ 0 ( pair) e p K+ KS ( pair) KL n

Charmed mesons

D+ (K- KS+)D0 (K-+, K- + 0, K-KS)D*0 (D0 0, D0 )D*+ (D0 +, D+ 0)Ds

+ ( +, K+ KS)Ds

+* (Ds+ )

J/e+e-

Light mesons

’

K* (K)(K+ K-, KS KL)

Others: (2S), c, c, K2*(1430), C, f0(980), a0,1,2

12

BB++→→ K* K*++ Candidate Candidate

High energy photon absorbedby ECAL

Ks →

and +

Muon from other B decay

Large fraction of stable particles identified by speciesand kinematics measured with high precision

13

OutlineOutline





14

Anatomy of a Branching Fraction Anatomy of a Branching Fraction MeasurementMeasurement

•Signal candidate reconstruction: Collect sample of events with exclusivefinal state X (particle ID and kinematic constraints)

•Background rejection: Cuts which exploit event shape differencesto reduce continuum and other backgrounds

•Signal yield extraction: Estimator of NX for selected events usingmulti-dimensional likelihood fit over kinematic and background rejection variables

•Efficiency estimation: Fraction X of produced events selected (checked against numerous control samples)

•B counting: Total number of B pairs NBB produced in collisions

•Putting it all together:

BF(B0 → X) = NX / X NBB

15

Kinematic ConstraintsKinematic Constraints

•For rare decay studies or high precision, eliminating random combinationsof particles as possible B candidates is essential •For a completely reconstructed B meson candidate with daughters (pi, mi), two uncorrelated kinematic constraints:

Beam-constrained mass:

Candidate should have mass mB2 = (5.279 GeV)2 = E2 – p2

AND, Y(4S) decays exclusively to a B pair, so E = E(beam) in Y(4S) frame

pB, E(beam) in Y(4S) frame

pB and E(beam) precisely known, so Mbc should agree with mB to within3 MeV! Independent of mass hypotheses for daughters

16


Energy difference:

Candidate energy should coincide with beam energy in Y(4S) frame

p, E(beam) in Y(4S) frame

Resolution typically 10-30 MeV

Depends explicitly on mass hypotheses for daughters

Incorrect particle ID or missing particles will shift E from 0(suppresses backgrounds from other B decays: false ID or “cross-feed”)

17

B0 →KsJ/(ee)


Feed-down fromJ/ K*+

Also: Intermediate particle masses (e.g. composites listed previously)and decay helicity angles can provide additional rejection

18

Background SuppressionBackground Suppression

Exploit fact that continuum events are more jet-like than BB events

•Event shape variables: spherical B vs. jet-like continuum(Fox-Wolfram moments, Legendre moments, sphericity, thrust angle)

•B flavor tagging variables: kaons and leptons in the rest of the event•B candidate vertex separation fromrest of the event

Sometimes combined into a single Fisher discriminant or neural net variable

19

Signal ExtractionSignal Extraction

Unbinned maximum likelihood fit of (mES, E, et al.): maximizes LH function

i components: signal, peaking backgrounds, combinatorial background, with normalization Ni and shape parameters i

j events: each event with a vector of variables xj

Pi typically multidimensional: Pi = Pi(mES)*Pi(E)*….. (usually uncorrelated)

Signal yield, background yield and background shape parameters are typically free parameters in fit (background systematics absorbed intostatistical error on signal yield)

Signal yield or other parameters of key interest are BLIND until all analysis techniques finalized

20

B countingB counting

•Select all on-resonance multi-track events satisfying very loose event shape cuts ( Non(MH) )

•Use same selection on off-resonance data ( Noff(MH) )

•Select ee → events as an estimator of luminosity in on and offresonance samples ( Non() and Noff() )

In practice, is very close to 1

B counting systematic uncertainty is typically 1%

21

Case Study: BCase Study: B00 →→K+K+ BF and A BF and ACPCP

BaBar preliminary BF, hep-ex/0608003 227M BBBaBar preliminary ACP, hep-ex/0607106 347M BB

Measurement of rare b →s penguin decay and its direct CP asymmetry

Simultaneous study of K, , KK (pion hypothesis for all tracks)

5D likelihood fit of

F = event shape Fisher discriminant, C = Cherenkov angle in DIRC

background at few % level

C PDFs and asymmetry in efficiencies from large (400k) D*+→ D control sample

22


Signal fit PDFs vs. data

(1660 events)

qq Bkgfit PDFs vs. data

(65000)

“sPlot” method of projecting out of the data each PDF component(a fancy background subtraction using LH function)

23


Signal fit PDFs vs. data

(1660 events)

qq Bkgfit PDFs vs. data

(65000)

“sPlot” method of projecting out of the data each PDF component(a fancy background subtraction using LH function)

24


4.5 evidence for direct CP violation!

K+-

K-+

No contamination

25


CDF is now an evencompetitor

Hard to reconcile withnull result in K+0

(4.9 differencein world average)

26

OutlineOutline





27

BB00 → J/→ J/ K KSS00 and sin 2 and sin 2

Decay dominated by a single “tree-level”Feynman diagram: b → ccs

J/ identified cleanly by decay to a lepton pair; KS identified cleanly by decay to pion pair.Both particles are CP eigenstates → both B0 and B0 decay to them

Time-dependent CP violation has amplitude sin 2 and frequency m

Works for several other b →ccs decays as well; results can be combined

S =

BaBar preliminary: hep-ex/0607107, 347 M BB

28

Determine time between decays from vertices

Δz=(βγc)Δt

(t) ≈ 1 ps

Time-Dependent CP Violation: Experimental Time-Dependent CP Violation: Experimental techniquetechnique

Determine flavor and vertex position of other B decay

l-

K-

Asymmetric energiesproduce boosted Υ(4S), decaying into coherent BB pair

e- e+

B0

B0

Fully reconstruct decay to CP eigenstate f

J/ψ

KS

µ+

µ- π+

π-

Naively: compute CP violating asymmetry A(t) = N(f ; t) – N(f ; t) N(f ; t) + N(f ; t)

In reality, extract A from unbinned LH fit over signal & control samples

Signal extraction

29

t Measurementt MeasurementTime diff. distributions of CP B decays with a B (B) tag

is average mistag rate (3-40%) is mistag difference between B and B tag (~1%)Observed f’s are convolved with resolution functions

BtagB0

BtagB00BBtag0BBtag

fobs = f± (t ) X R( t ) f±( t ), = = 0

30

t Measurementt MeasurementTime diff. distributions of CP B decays with a B (B) tag

31

Flavor TaggingFlavor Tagging

Six

Q = 30.4 %!

NQ

1σ(sin(2β))

32

B Candidate SamplesB Candidate Samples

Extract control sample of hadronic B decaysthat are self-tagging and negligible CPV

B → D(*)- h+, h+ = +,+, a1+

100,000 events!

For each flavor tagging category, measure , and resolution functions.

Extract signal of CP B decays

11,000 events

33

Asymmetry ExtractionAsymmetry ExtractionSimultaneous fit to t distributions of CP events and hadronic eventsover six tagging categories and six decay modes

34

CKM Global Fit (Sep.2006)

35

CKM Global Fit (Sep.2006)

36

Time-Dependent CPV in b →sss Decays

Smaller than bccs in all of 9 modes

Smaller than bccs in all of 9 modes

Naïve average of all b s modes

sin2eff = 0.52 ± 0.052.6 deviation betweenpenguin and tree (b s) (b c)

Naïve average of all b s modes

sin2eff = 0.52 ± 0.052.6 deviation betweenpenguin and tree (b s) (b c)

Hazumi, ICHEP 2006

37

OutlineOutline





38

Recoil Side ReconstructionRecoil Side ReconstructionLesson from sin 2To gain precise control over one of the B mesons,reconstruct the other B meson in high-purity, high-efficiency modes

Efficiency penalty is at least 1%, so signal B process cannot be too rare ( BF >10-6 ) and backgrounds must be large for this to be optimal

Use cases: Inclusive B decay rates and distributions: large rates, but fewer kinematic constraints on the signal decays, due to neutrinos ( b → cl or ul ) ornumerous final states ( b → s )

Exclusive B decays to one or more neutrinos: lack of kinematic constraintson the signal side means recoil side is the only recourse for backgroundrejection ( B → , B → , B → K(*) , B → invisible)

Recoil B(D(*)+n

Signal B(anything!)

Y(4S)

39

Case Study: B+ → Case Study: B+ →

Belle PRL 97 (2006) 251802, 449M BB

Simple decay through weak annihilation

Sensitive to B decay constant fB or to charged Higgs boson

SM BF ~ 1 X 10-4

Up to 3 neutrinos in the final state requires recoil reconstruction sample

40


Total Event ReconstructionOn recoil side, require charged B decays to high purity, high rate hadronic decays using previously described signal extraction methods

180 modes, 680k events, Recoil side efficiency = 0.13%

41


Total Event ReconstructionOn signal side, require simplest decay modes and no other tracks

5 modes, 300 events, recoil side efficiency = 15%

qq background eliminated!Remaining background from B decays to neutrals (0, n, KL)

Extra neutral energy (ECL)peaked at 0 for signal

Background broad in (ECL)(shape from MC)

42


Signal extracted from 1D likelihood fit to ECL

24+\- 7 signal events observed!

Significance = 3.5

Consistent excess in all decay modes

43

BF sensitive to H+ in Type II 2HDM at largetan


For MSSM, there are significant SUSY radiative corrections which complicatesinterpretation

Sensitivity comparable torecent MSSM Higgs searchat Tevatron

44

SummarySummary

Highlights of B physics capability at Y(4S)

•Awesome accelerator performance results in Giga-B samples

•Unique kinematic constraints and control samples allow for precise/rare studies of exclusive B decays and direct CPV

•Highly efficient flavor tagging makes for world’s best time-dependent CPV

•Recoil side reconstruction allows for detection of B decays to virtually anything

•Can be extended in the future to a Super-B factory (100X) or Bs physics at the Y(5S)

dream machines: b physics with e + e - colliders

Documents

p pp0h h p p

p p0w p pp0h gg

p muon

light mesonsr p p

p fake rate

p e fake rate

b pair

ks p pair