Status of B hcK

1
Status of B? hcK

• Stefania Ricciardi (Royal Holloway)
• for the hc group
• BABAR Collaboration Meeting, SLAC, December 2002

SACLAY Gautier Hamel de Monchenault Witold
Kozanecki
MANCHESTER Frank Jackson
BRISTOL Nick Barlow Phil Clark Marc Kelly
ROYAL HOLLOWAY Chris Marker S.R.
IRVINE Dieter Best
2
Outline

• Recent achievements
• Sin2b
• Branching Fractions
• Current Studies
• More decay modes
• Improved selection
• Background studies
• Efficiency variation across D.P.
• Publication Plans

3
Sin2b with B0 ? hc K0s
Summer 2002 Measurement sin2b
0.59 0.32 (PRL vol89, n.20) based on
132 events selected out of full sample
(run1run2 88million BBbar events) Including
J/y region hc(J/y) ? KsKp hc(J/y) ? KKp0
Purity 63 Background is combinatorial and
peaking
KsKp
hc
J/y
Supporting internal Etac-CP documentation
BAD381,BAD442
4
Branching Fractions hep-ex/0203040 (Moriond
QCD,2002)
Supporting Etac-BF docs BAD222, BAD336

• Analysis based on Run1 data only ( 22.7 million
  U(4S)?BB)
• Searching for
• 
  BF (PDG 2002)
• 1. hc ? KsKp- c.c. (1.8 ? 0.6)
• 2. hc ? KK-p0 (0.9 ? 0.3)
• 3. hc? 2(KK-) (2.1 ? 1.2)

Combining 1 and 2 we obtained BF( B ? hc K)
( 1.50 ? 0.19 ? 0.15 ? 0.46) x 10-3 BF( B0 ? hc
K0) ( 1.06 ? 0.28 ? 0.11 ? 0.33) x 10-3
Decay 3. was Not included in final result because
of small signal significance and poorly-known
BF(hc? 4K)
5
Additional hc decay modes
Promising channels under study
hc ? pp
BFPDG (0.12 0.04)
hc ?hpp
BFPDG (4.9 1.8)
- Analysis is still blind See
talks in Charmonium II session by Dieter Best and
Phil Clark respectively
6
hc ? ff
B Estimated efficiency 17.8 Expected signal
19 Expected background (from
sidebands fit) 0.22 0.15
BFPDG (0.71 0.28)
Another 4 charged-K final state Cut on KK-
invariant mass ? 14 MeV/c2 from
Mf effective for background rejection Loose PID
requirements (4 NN-NotAPi)
B0 Estimated
efficiency 18.6 Expected signal
7 Expected background 0.04 0.04
DE
Run1Run2 Full data set
B
MES
7
hc ? ff Yields
After unblinding B Signal
16.5 4.1
(19 expected) Background 0.6
0.6 (0.2
expected)
After unblinding B0 Signal
3.6 2.1
(7 expected) Background 0.7
0.7 (0.04
expected)
B
B0
MES(GeV)
MES(GeV)
8
Improving the SelectionNew skim

• Main changes
• Same PID selector at skim and analysis level
• Better understanding of efficiency
• Removed ? cut and replaced with p cut

n ? cosine of helicity angle under B-gthcK
hypothesis
p
Old analysis cuts
Old skim ?lt1.1
Correlation between n (p,MX) Cutting part of
J/y signal
New skim analysis cut
Mx (Gev/c2)
9
Improved selection re-optimization of old
modes
Old procedure (Run1)
New procedure

• Un-smeared SP3 MC
• for signalbackground
• low statistics in background
• due to lack of generic MC
• Use PDG BF(hc?X) and
• CLEO BF(B ? hcK) for signal yield
• Fisher only used photons from
• B candidate
• Cutcount analysis
• optimized all cuts including
• m_ES, m_etac

• SP4 fully-corrected MC for signal
• (PID/pi0 killing neutral energy
  shift/smearing)
• DE sidebands of data for background
• Use our Run1 measurement
• BF(B ? hcK) x BF(hc?X)
• Fisher retrained using all photons
• ML fit to m_ES,mX
• Optimized all other cuts
• (including n to reduce higher
• combinatorial background of new
  skim)

10
Example Fisher optimization (Nick Barlow, Marc
Kelly)
KsKp K
Signal

• Important to reduce continuum
• Trained independently for each decay
• mode using off-resonance data for bkg
• and MC for signal
• Based on 18 quantities including
• 9 CLEO energy-cones and various
• event-shape variables

Bkg
KKp0
KsKp
4KK
Old gt0.0 New gt -0.08
Old gt -1.4 New gt -0.48
Old gt0.5 New gt 0.16
11
New selection summary
Main cuts are relaxed. Will benefit of higher
efficiency smaller systematics
12
Expected Yields
For Run1 Run 2 data sample
Charged
Neutral
Expected higher significance in all channels
13
Sources of background

• Full MC studies showed
• Pure combinatorial
• typically 20-30 of signal according to mode
• Peaking background (10-20 of signal)
• Mainly B decays to exact matches
• For example for KsKp Ks final state, scaled to
  80.8/fb
• Partial reconstruction negligible
• Inclusive hc
  negligible
• These MC studies suffered from unknown BF, low
  statistics
• ? CANNOT rely on MC for absolute yield estimation
  of background

14
Determination of signal and background

• 2D unbinned maximum likelihood fit to mX, mES
  used for sin2b
• B?hCK
• Gaussian in mES
• BW x Gaussian in mX
• B?J/yK
• Gaussian in mES
• Gaussian in mX
• Peaking Bkg
• Gaussian in mES
• flat in mX
• Combinatorial
• Argus in mES
• flat in mX

Old skim/selection Full data sample
MKsKp
Major assumption to be verified
15
PravdaMC studies
High efficiency wrt SP4 kaons and pions do no
decay Additional killing per track required set
at 5 in these studies Good agreement for most
important distributions (for KsKp and 4K
channels) we can use it to check background
shapes
KsKp K
DE
Blue Signal SP4 Red Signal Pravda
mES
mKsKp
16
PravdaMC Generic B production
We interfaced Pravda to our reconstruction
code and we could generate 500 /fb of generic B0
and B in about 1 month CPU time 50 h/fb
(mainly due to reconstruction) on Linux public
queues _at_ SLAC Total 50K hours
PravdaMC B Generic (500/fb)

• We can then
• Verify assumptions
• on background estimation
• Isolate individual contributions
• of different background
• components
• Study fit systematics due
• to different bkg composition

17
Pravda MCstudy of bkg shape
2.75 lt MX lt 2.85 GeV/c2
2.85 lt MX lt 2.95 GeV/c2
MES plots in MX slices KsKp K reconstructed from
B generic J/y and hc signal removed at generator
level
2.95 lt MX lt 3.05 GeV/c2
3.05 lt MX lt 3.15 GeV/c2
peaking ev. (from fit in MX slices)
3.15 lt MX lt 3.25 GeV/c2
MES
18
Pravda MC mES PlotsKsKp K bkg components
Exact matches
Non resonant
hc and J/y inclusive
D0 inclusive
K inclusive
Exact matches
Exact matches
D0KsK / D0Ksp
KKK
Others
Tracks from both B
19
Pravda MC4K
2.75 lt MX lt 2.85 GeV/c2
Not many B survives selection cuts Main
contribution from non-resonant B decays to
4K and ccbar inclusive More complex shape in
MX-MES Small fraction of peaking expected (1-3
of signal) but we cannot trust MC BFs
2.85 lt MX lt 2.95 GeV/c2
2.95 lt MX lt 3.05 GeV/c2
3.05 lt MX lt 3.15 GeV/c2
3.15 lt MX lt 3.25 GeV/c2
20
Dalitz Plot studies
Dalitz plot from ?? ? ?c ?Ks0K?
(B.Mayer, GHM)
M(Kp-)
Large resonant component hc ? K(1430)K ? KsKp
hc ? K K- ? Ksp K hc ? K0 Ks ? Kp Ks
M(Ksp?)
MC generates non-res. 3-body decays ? Efficiency
variation across phase-space to be corrected
21
Alternative Dalitz Plot representationSee talk
by F. Jackson _at_Dalitz Workshop (Dec 7th)
Phase space boundary shape in m2(Kp) vs m2(Ksp)
is not convenient
Transformation of phase space element
d(m12)2d(m23)2 ? p q dm12 dcos?H(12)
p mom of 2 in (12) rest frame
q mom of 3 in mother rest frame
Method suggested By W.M. Dunwoodie
22

Efficiency Parameterisation

• Angular distribution enables use of orthogonal
  Legendre polynomials P(c) where c cos?H
• Coefficients pi obtainable through L.P.
  orthogonality
• pi 1/Ntot?jPi(cj) mean value of Pi
• For 2-D efficiency, repeat this
• parametrisation in several bins of m
• Fewer boundary problems

Reco distribution dN(c)/dc Ntot ??ipiPi(c) True
distribution dNgen(c)/dc N0P0 Efficiency ?(c)
dN(c)/dc / dNgen(c)/dc
23
Example using ?C ? KS K?-
Squared Dalitz plot m( K?- ) vs. cos?H(K?-)
cos?H(K?)
m(K?)
Legendre polynomials to 2nd order (checked to
4th order. 3rd order should be included, 4th not
necessary)
p2 p1 efficiency
mass slice
cos?H(K?)
24

Efficiency Correction

• Corrected Yield
• NY ?1/?(m,c)
• Test correction on multiple MC samples, observe
  spread of results
• D (NY Ngen)/?(NY)
• ?(NY) ?(1- ?) ?1/?2(m,c) (binomial)
• expect ltDgt 0 and st.dev.(D) 1

88 K B ??C K events
ltDgt, RMS(D) consistent with 0,1
25
A To-Do list to BF publication

• New modes
• finalize cut optimization and unblind (ff done)
• Peaking background checks completed for
• KsKp and 4K
• Pravda does not give sensible results for KKp0
• (broad distributions, small efficiencies)
• Efficiency corrections and Background
  subtraction
• Understand how to deal with it
• ML extraction of yields on new skim,new cuts
• (RooFit)tools already in place
• All modes study systematic errors
• done for old modes on Run1 Data/SP3 MC
• Main contributions PID efficiency, tracking
  efficiency, background model
• To be repeated for all modes on
  Run1Run2/SP4

26
Conclusions
Moving ahead for BF refereed publication
including 3 additional hc decay modes and
refined analysis Aiming to early 2003 Very much
wanted include into publication etac-width
measurement with exclusive decays On old skim
G(hc) 32.9 ? 5.0 (stat) MeV In good
agreement with BaBar gg measurement and
significantly higher than PDG (16.0 /- 3.4
MeV)! Nice short-term project awaiting for a
volunteer 4 PHD thesis close to the End (Chris,
Frank, Nick,Melanie) Other channels and analysis
available to smart students Willing to join this
group (hc ? KKp, rr, baryons)