Transcript Selected Results and Prognostications on Vcb & Vub

Selected Results and Prognostications on Vcb & Vub :
A B Factory Perspective
Vivek Sharma
University of California San Diego
[email protected]
Ringberg Phenomenology Workshop on
Heavy Flavors : Rottach-Egern, Germany
The Two Approaches in Vxb
T opics today:
1. B  D
*
 & |V cb |
2. M om ents in B & |V cb |
3. B   [  ,  ]

& |Vub |
4. First steps tow ards Inclus ive B  X u  & | V ub |
5. Future D irections in clean |V xb | m easu rem e nt s
This talk is an “appetizer” not a review. See recent CKM workshop page for
complete results
http://ckm-workshop.web.cern.ch/ckm-workshop/ckmworkshops/Default2003.htm
Inclusive Semileptonic Decay Rate
Babar (also CLEO)
Basic foundation of all semileptonic studies. Modern measurements agree
(1) B  D
•
•
*(  ,0 )

 and |Vcb |
Experimentally favored (S/N) w.r.t BDl nu
Experimental Challenges:
1. Slow pion tracking ( helical path of
decreasing radius)
2. Charm branching fractions
3. Knowledge of higher mass Dnpi states:
• excited D**, Non-res. D(*) npi l nu ??
4. Form Factors (mostly for  measurement)
CLEO has been the most experienced player
on this topic  new results with ¼ total data
Silicon vertex tracker
A complete set of measurements from CLEO:
CLEO
| V cb | [46.9  1.4 sta t  2.0 syst  1.8 th ]  10
~6.5% Measurement
3
Branching Ratio of B  D* l nu
disagreement between
CLEO & Babar/Belle in value of
the branching ratio
Each experiment have much more data
left to analyze
Remove stat. fluctuation as source of
disagreement, focus on finding
systematic biases
“Repeat n>1 more time and very carefully”
|Vcb |World Average (CKM’03 preliminary)
Updated prelim average using
F(1) = 0.91  0.04
|Vcb|=[ 42.6
± 0.6(stat)
±1.0(syst)
± 2.1(theory) ] 10-3
3% Stat
Seeking Consistency
?
Moment
(II) Moments in BCleo
Decay:
Elegant Analyses
Measurements from CLEO
Use HQE/OPE to predict SL rate &
Moments of inclusive B decay spectra
Photon energy spectrum in B Xs g
Hadronic mass spectrum in BXc  
Lepton energy spectrum in BXc  
Lepton energy spectrum in BXc 
Cleo Moment Analyses:
Consistent picture
Hadronic mass spectrum in BXc 
Photon energy spectrum in B Xsg
Remarkable !
 = 0.39+0.03stat+0.06sys+0.12thGeV
l1=0.25+0.02stat+0.05sys+0.14th GeV2
1s Theoretical
Ellipse
PDG
|Vcb | [40.8  0.5 
sl
 0.4
exp
 , l1
theo
 0.9 
s
 3 ]  10
,M
B
3
3% measure
Comparing Exclusive & Inclusive Vcb Measurements
A veraging m any m easurem ents is t ricky w hen the m easurem ents
are done in different environm ents/presc riptions
I prefer to look for consistency w ithin ea ch experim ent
C LE O E xclusive : B  D
*

| V cb | [46.9  1 . 4 stat  2.0 syst  1.8 th ]  10
3
and
C LE O I nclus ive
|V cb | [40.8  0.5
PDG
 sl
 0.4
exp
 , l1
theo
 0. 9 
s
,M
3
B
]  10
3
S eem a bit different ?? V e ry int eresting to w atch w hat
B elle & B aB ar get from sim ilar set of m easurem en ts
Babar Hadronic Moment Analysis: ICHEP Preliminary

Strong dependence of moments
on p*min
For p*min=1.5 GeV/c and
=0.35 ± 0.13 GeV [1]
(reliance on b  sg spectrum)
l1= - 0.17 ± 0.06 ±0.07 GeV2
CLEO [1]
l1= - 0.226 ± 0.07  0.08 GeV2
But
these parameters do not describe
P* dependence of the moments!
l1(0.9 GeV/c) – l1(1.5 GeV/c)
= 0.22±0.04±0.05 GeV2
[1] CLEO PRL 87, 251808 (2001)
BABAR Preliminary
<MX2-MDspin2>

?
CLEO
OPE (Falk, Luke)
, l1 free param.
No non-resonant
states (MC)
OPE (Falk,Luke)
 = 0.35 GeV
p*min [GeV/c]
NB: Data points highly correlated
Can Hadronic B Decays Help Understand Nature of High Mass Dn states in
SL Decay?
B D** 
factorization ?
B D** l nu
D**
D**
Belle
D*+ 
D+ 
Non-reso
What can one
Learn from such
results ?
Desperately Seeking Vub !!
An important measurement in shaping - Real estate
Constraint from Vub
Desperately Seeking Vub ! : By Exclusive Reconstruction
• Recent measurements from Babar, Belle, Cleo  talk about this
• SL decays  missing   need to measure 4-momentum in absence of 
signature in detector  very demanding and very important for Vub !
– Exploit Hermiticity of detector (if you have few holes and have excellent
particle reconstruction capability
• CLEO is best for this purpose (95% hermetic) , then Belle, then Babar
MB=
New results comparing data q2 with models
Wins the most improved Vubmeasurement award
Measurement statistics
limited
Vub From Measurement of Exclusive Final States
Difficult to combine results from several
experiments due to diff ranges of modeling, FF
variations…dust needs to settle here (HFAG)
(4) |Vub| From Inclusive Measurements
CKM Workshop 2003
Note : superficially all measurement look too consistent !
Probably because of (large) common Theory systematic error ?
Desperately Seeking Vub : From Inclusive Measurements
interpretation
Luke
@
CKM

pv ?

pv ?

pv ?
The Perfect Detector for B Semileptonic Measurements Has
No Holes ! (B Decay “bomb” goes off in all directions)
This was an
animation
Need A
Spherical Detector !
B(4S)
1
B2
(4S) Detectors: Characteristic Features
Machine optics
Babar Not Hermetic Due to Intrusion of accelerator optics
near Interaction Region (dipoles)
Belle is Perhaps Better but not like CLEO
Missing Momentum Resolution : Hermiticity Issues
85 MeV  >> 160 MeV could be very costly in SL measurements
Need an alternate solution that pays in the long run
Beginning of an Experimental Paradigm Shift in Vxb
Measurements at B Factories
• Hermiticity, so vital for SL measurements is not the best feature of
Belle/ Babar detectors due to intrusion of machine into detector
– Necessary holes !
• But B-factory detectors recording ever increasing samples of BBbar pairs (> 100 Million BBbar recorded already 1000 Million)
– At the price of modest efficiency (4%), can fully
reconstruct one B decay into all hadronic final states
(Breco) and examine the other(recoiling) B decay
• This “Recoil side” studies perfectly suited for many Vxb studies
• Much “cleaner” and more powerful than neutrino reconstruction a
la CLEO
• I will show you (with example) that this is the most promising way
for the future Vub and other Semileptonic measurements
The Perfect (4S) Event: Example of Recoil Side Analysis
In this (rare) case all
particles were
sprayed within fiducial
volume of detector
Replace this with your
Favourite Vxb mode
Advantages in Recoil Side Measurements
• Full reconstruction of B1  “perfect” knowledge of B p
• Turn around and examine the recoiling B2 with this info
– Pmissing knowledge much better than in “neutrino
reconstruction)
– Most backgrounds in Vxb measurements “disappear”
• Udsc background (continnum)
• Leptons from other B (bclnu, b cs l nu)
• Combinatorial (1/2 event accounted for)
• The Y(4s) decay so much better understood,
– Various charge correlation (D not a Dbar) allows
background rejection
– Can fit entire event for event hypothesis in question
• Price to pay: Efficiency (but have ever growing data)
• Sometimes one can “have the cake and eat it too”
Example: Babar’s Inclusive bu l nu Measurement
Si

B
Cui
duality
Exptal measurement challenging because
• Rejection of large BXcl background
– (~60 times higher BR)
• Extrapolation to full phase space introduces
theoretical uncertainties
l
b

Theoretically relatively “simple” in principle but
– parton level calculation has to be
extended to account for hadronization
effects and Fermi motion (b quark
mass)
– calculation of decay rate relies on OPE
for which the convergence depends on
full acceptance and is impacted by
non-perturbative effects
l

u
Recoil Side Study : Technique
Xu
Breco
D*

Brecoil
Y(4S)
Reco side
l

Recoil side


D*
B
Y(4s)
missing
mass
squared
l
Xu
B
Y(4s)
B Candidate Mass
•MX reconstruction
•Kinematic constraints to
improve MX resolution
Fully Reconstructed B Sample
•Initial B Reco efficiency is 0.4%
•About 4000 B/fb-1
1300 B0
2700 B-
S/B~0.3
Analysis optimized to provide maximum
Number of reconstructed B without
consideration of “recoil side” physics
Require Lepton (p*>1.0GeV)
Use basic requirements of “recoil side “
Physics to clean up signal, e.g.
additional
•Lepton
•High energy photon
S/B~2.5
Steps in Measurement: B -> Xu l nu Signal Generation
Purely non-resonant
model based on the
De Fazio-Neubert
model
m X (G eV )
Hybrid model: Babar MC
Resonant (PDG +ISWG2)
+ non-resonant
m X (G eV )
Integral of the
hybrid model has to be
compatible with the
non-resonant one (duality
hypothesis)
m X (G eV )
Analysis Method: If it walks like a duck, talks like a duck …it
is a duck !...(well, most of the time)
•
Brecoil selection and reconstruction of the
X system B Xu l :
–
–
–
–
–
–
•
One and only one lepton with p*> 1 GeV/c
Correlation between lepton charge and Breco
flavor
(B0 mixing is corrected)
Cut on the missing mass: Mmiss2 < 0.5GeV2,
charge conservation: Qtot=0
Partially reconstructed neutrino to reject B0
D* l  events
kinematic fit (2-C): improve hadronic mass
resolution
Separate BXul in signal enriched and depleted:
–
used to perform the measurement
signal depleted : one or more K± or KS in the
event
•
•
m X (G eV )
signal enriched: veto events with K± and KS
•
–
reconstructed
generated
reconstructed
generated
used as control sample
Systematic error in measurement reduced by
measuring ratio
Ru/sl=B(B Xu l  )/B(B X l  )
m X (G eV )
Kinematic Fit To Entire (4S)  BBbar Event
Well known energy and
momentum of the incoming
Electron and positron
Pe+
Breco
Recoil
Pe-
(EPEPII , PPEPII) well known!
Energy and Momentum Conservation
Ebreco + EX + El + E - EPEPII = 0
Pbreco + PX + Pl + P - PPEPII = 0
Reco. B:
4 measured quantities
Lepton:
3 measured quantities
Neutrino:
3 unknown quantities
 4 Constraints
+ equal mass constraint
M(Breco)=M(X,l,)
 1 Constraint
5 Constraints – 3 unknown quantities
=
Over constrained system (x 2)
MX Correlation: Generated Vs Reconstructed
Linear Correlation
Unbiased Mass
Reconstruction
Mass Resolution ~ 300 MeV
Effect of Tight Recoil Side (Vub) Selection
S/B after recoil side selection
lepton requirement
quality cuts
kinematic constraint
Kaon rejection
m X (G eV )
S/B
S/B
m X (G eV )
Recoil analysis
S/B when only lepton required
S/B ~ 0.05
S/B ~ 1.7
m X (G eV )
Unbiased MX reconstruction and s(MX)~300 MeV.
m X (G eV )
Extraction of B(bul)
• Fit on the signal enhanced sample
• Three components to fit the MX distribution: bul , bcl , &
Hadronic background
• Signal efficiency (eselu eMxu), Breco efficiency ratio (etu/etsl) and lepton
efficiency ratio (elu/elsl) from MC
Then multiply
by B(BXl)
measured by BaBar
m X (G eV )
MX (bu l nu) Selection & Background Rejection
Mx cut optimized by minizing the
total error:
Statistical
+Branching ratio uncertainty
+Other experimental systematics
+Systematics from theory (mb & a)
 Optimal point is 1.63 GeV
 Cut lowered to a safer 1.55 GeV cut
(negligible change in the total error)
total error
statistical error
BR syst. error
detector syst. error
theory syst. error
Resulting MX Spectrum
Fit to the MX distribution
Background Subtracted spectrum
(MC)
Breakdown By Category & Stability In Event Selection
Ru/sl
All events, MX < 1.55 GeV
0.0197  0.0025
All events, MX < 1.4 GeV
0.0177  0.0025
All events, MX < 1.7 GeV
0.0211  0.0029
B0 decays, MX < 1.55 GeV
0.0246  0.0043
B+ decays, MX < 1.55 GeV
0.0168  0.0030
Electron sample, MX < 1.55 GeV
0.0226  0.0035
Muon sample, MX < 1.55 GeV
0.0166  0.0036
Theoretical Uncertainty Due to MX Cut
• b quark is not at rest in the B meson
(Fermi motion)
• Fermi motion depends on nonperturbative parameters (mb and a)
• Uncertainties on mb and a affect the
shape of MX spectrum
• MX spectrum reweighted
– (De Fazio et al JHEP 9906,017) taking into
m X (G eV )
account uncertainties on l1 and 
(from CLEO moments analysis f ( x )  N (1  x ) a e (1 a ) x ; x  f (  )
PRL87:251808,2001)
m b  mB  

2




 1
 a    3

l1



Theoretical Uncertainty on bu l nu rate
• Systematic error due to efficiency
of the MX cut
(MX <1.55 GeV) changes since
the MX spectrum changes
• Systematic error due to selection
efficiency (since the efficiency
depends mostly on MX itself)
eMxu
Two effects:
m X (G eV )
The combination of these two
effects (of the same order of
magnitude, ~9% each) gives:
s (theory) = 17.5%
m X (G eV )
Total Systematic Uncertainty in Rate Measurement
Statistical error (data+MC) 13.7%
Detector simulation errors
+
Fit systematics 9.8%
bcl and D decays modeling
+
bul decays modeling 6.0%
Fermi motion 17.5%
|Vub| Result : Preliminary
Measure the charmless semileptonic branching ratio
Ru/sl
And extract Vub
Interesting check of theory uncertainty:
result very stable if apply a cut on the
invariant mass of the lepton-neutrino system
(q2)
(Bauer et at. hep-ph/0111387)
q2 (GeV2)
This Result on |Vub|
Inclusive |Vub| measurements
Precision in this measurement alone is better than the LEP average
Future Prognostications (conservative)
Redoing the same analysis (no improvement) in 500fb-1 data , the errors should scale
as:
stat err.
exp. syst
theo. syst
total
NOW
6.8%
6%
10.5%
13.5%
500fb-1
< 2.7%
< 3%
10.5%??
11.2%
Measurement will be dominated by theoretical uncertainty if nothing improves
But… errors on mb and a should go down in future
Expect the total error can go well below 10% ?
Expected systematic error due to shape function? [decrease it with info from
Radiative Penguin measurements?]
Future Direction with More Data: q2 vs MX analysis
A combination of cuts on q2 and MX
reduces theoretical error (Bauer et
al. hep-ph/0111387)
With 80fb-1 this 2D technique is not
suited (additional 40% efficiency due to
q2 cut)
With 500 fb-1 can lead to better
precision
With a combination of
MX <1.7GeV
q2> 8.0 GeV2
Theory error < 9 % ?
Future: other possible checks & approaches (?)
• Try combination of variables.
Ciuchini et al. ph/0204140
This approach (uses bul and bsg) in not dependent on shape function. Since
resonances in bsg have to be removed, efficiency will go down by >50%. In
500fb-1 stheo(Vub) ~ 5% ??
• Can we measure mb directly on our data-sample?  E W  | pW |
Kowalewski et al. (ex/0205038) claim one can, using
With the current data-sample (80fb-1) s(mb)~120MeV
 in 500fb-1 s(mb)~50MeV  stheo(Vub) ~ 6% ??
 Too aggressive expectation ?
Reconstructing Exclusive B0 - l+  , B+ 0 l+ 
On the Recoil Side
B0l(*)
~500fb-1
B+0l (*)
18
Fitted MX
16
Fitted MX
14
12
10
8
6
4
2
m X (G eV )
(*)cuts not yet optimized
m X (G eV )
Sensitivity With 500 fb-1 ?
B0l
B+0l
B+0l
B+l
q2 Spectrum: Distinguishing Between Models
q2 spectrum for B0l-
models
reconstructed
500 fb-1
q2(GeV2)
In 500fb-1 enough statistics to discriminate among models ?
Conclusions
• Many improvements in
– Exclusive & inclusive measurements of Vxb
– Interplay between theory and experiment crucial
• Already shows nice synergy (precise results)
• Gathering more focus and attention at BaBar & Belle (now that
Sin2beta is not the primary focus)
• Battle for precision Vub developing
BaBar analyses using the Recoil
This reconstruction has been used by:
• Measurement of the hadronic moments in SL
decays
• B t 
• Measurement of the SL BR (b0, b+)
And it will be used by other analysis:
• B s g
• B D* t 
• Ratio of production N(B0)/N(B+)
• B 0()l, B 0()l, B  l, ...
• And many others
Data-MC Comparisons
Vub inclusive (CKM workshop)
Recoil of B D* l  in Belle
Sample Purity : Flexible
S/sqrt(S+B)
purity
10
20
30
30
40
yield
50
60
(*103)
10
20
30
30
40
50
60
Yield
3)
yield (*10
(4S) Detectors: Belle
Belle
Two ’s from J/y  
Two ’s from KS  
We reconstruct B mesons
from detector hit signals
B Factory Detectors: Babar & Belle
• Excellent tracking and calorimetry E > 60 MeV
• Excellent charged Kaon identification
• Excellent KS  + - identification
• Due to encroachment of accelerator optics near interaction point
– Holes in the front and back
• Hermiticity limited (CLEO much better)
• B mesons decay at rest  remnants thrown in ALL directions
– Not collimated as at LEP/SLD
Exclusive Decay Rate Measurements
Errors : Stat, syst, FormFactor
Theory error = 50% of entire spread
-
Experiments have to agree to make
Similar variation in their analysis so that their
results may be compared.
Need a better concept of “theory” error !
-
B1
B2
SM and Unitary Triangle
Constraint from Vub
SemiExclusive Reconstruction
Aim is to collect as many as possible fully reconstructed Bs in order to study
the property of the recoil.
• Reconstruct B  D(*) n mK pKs q 0 but the intermediate resonances are
not requested
• This is the so-called SemiExclusive Reconstruction. For instance in B 
D* if you don’t request the  invariant mass in the a1 window you do
SemiExclusive reconstruction
Vub From Measurement of B   /  l 
B   
B  
Wins the most improved Vub measurement award
SemiExclusive Reconstruction II
Two steps:
• Reconstruction of the D meson in hadrons
• Reconstruction of the B meson in hadrons
the signal box is defined using two variables:
Resolution from
beam energy
Sensitive to
E measurement
Uncorrelated variables
(just the beam energy but small uncertainty)
Recoil Side Physics : Targeting Exclusive bu l nu modes
Bl,
Bl,
Bl,
Ba1l,
...
• 170 BXul events on data after all cuts for MX<1.55GeV
• exclusive channels can be studied with ~same technique
• large potential to perform exclusive Vub analyses
The Story So Far & Plan of This Talk
• Semileptonic B decays have been studied since discovery of (4S)
– This audience knows it all !
– See recent CKM workshop page for complete results
• http://ckm-workshop.web.cern.ch/ckm-workshop/ckmworkshops/Default2003.htm
• In this talk
– Give a short review of current status
– What B factory detectors can (not) do
– Pick a few interesting new results and project them in
future (generate some discussion)
• Future is 2007 or ~500 fb-1 at Babar/Belle each
• Warning : I am an interested observer in this activity, not an active
participant + my optic is primarily (4S) based