Pinning down the JPC values of the X3872

1
Pinning down the JPC values of the X(3872)

• K. Miyabayashi
• for
• S.K. Choi S.L. Olsen

Feb, 2005 Belle Analysis Software Meeting
2
JPC possibilities for J 2
3
P-violn DD-allowed JPCs unlikely(reduce type
size of these entries by x1/2)
4
Use e7 e37 data(include B?KS pp-J/y)
Signal (47 ev)
Sidebands (114/10 11.4 ev)
5
Follow advice from Sherlock Holmes
Eliminate all other factors, the one that
remains must be the truth.
The sign of four
6
Areas of investigation

• Angular correlations
• Search for radiative decays
• Fits to the M(pp) distribution

7
Angular Correlations
8
Strategy for each JPC, find a distrib ?0if we
see any events there, we can rule it out
y is 1--
example1-- sin2qKm
Use y to check accept.
y c2/dof 8.9/9
qKm
K
compute angles in J/y restframe
D.V. Bugg hep-ph/0410168v2
9
cosqKl for X(3872) events
c2/dof 60.3/9
fit with sin2qKl bkgd
see 8 evts/bin
expect 23evts/bin
background scaled from sidebands
X(3872) is not 1-- !
10
1- and 2--
use J/y helicity angle qJ/y
For the y?pp-J/y, this should be flat
K
X
qJ/y
J/y
cosqJ/y
11
1- and 2--
2-- sin2qJ/y cos2qJ/y
1- sin2qJ/y
c2/dof36/9
c2/dof27/9
cosqJ/y
cosqJ/y
can rule out 1- (Cl10 -4) 2-- is unlikely
(CL0.16)
12
0- Rosner (PRD 70 094023)
0- sin2q sin2y
c2/dof31/9
q
cosq
c2/dof61/9
y
safe to rule out 0-
cosy
13
0 Rosner (PRD 70 094023) again
In the limit where X(3872), pp, J/y rest frames
coincide dG/dcosqlp ? sin2qlp
qlp
c2/dof 59/9
rule out 0
cosqlp
14
1
compute angles in X(3872) restframe
1 sin2ql sin2c
c2/dof 10.4/9
Rosner (PRD 70 094023)
ql
K
cosql
c
c2/dof 8.1/9
1 looks okay!
cosc
15
1 again
compute c in pp rest frame
compute q in J/y rest frame
c2/dof 8.9/9
c2/dof 8.9/9
cosql
cosc
16
Reduce type size for JPC values that fail angle
tests
17
Observation of X(3872)?g J/y
18
Select B?K J/y g(include both K Ks)

• Tight J/y cuts
• K idgt0.5 / Belle-standard good Ks
• Eggt40 MeV
• p0 veto (c2gt4.0)
• K veto (M(Kg)gt1.0 GeV)
• R2lt0.4 cosqB lt 0.8
• Mbc 5.28lt0.0055 GeV (2s)
• DElt0.034 GeV (2s)
• E7 ? E37

19
M(g J/y)
B?Kcc1 cc1?g J/y
X signal Region 32 MeV
X(3872)?
M(gJ/y)
20
Expand cc1 region
Fit to determine s M(gJ/y) 10.7 MeV
M(gJ/y) mcc1 lt 25 MeV ( 2.4s)
Use these fits to get means sigmas for Mbc and
DE
Nev(cc1) 653 26
21
M(g J/y) in X(3872) region
32 MeV
32 MeV
32 MeV
Use this fit to get Efact for ARGUS
sideband
X(3872)
Nev(X3872) 13.4 4.4
No signif. peaking bkgnd scaled fit 1.0 2.1
evts
Signif ? 2lnL(1.0) -2lnL(13.4) 5.1 s
22
Br(X?g J/y) determination
Br(X?g J/y) Br(X?pp-J/y)

Nev(X?g J/y) Nev(cc1?g J/y) Nev(X?pp-
J/y) Nev(y? pp- J/y)
e(cc1?g J/y) e(y?g J/y) e(y? pp-
J/y) e(X?pp- J/y)
Br(B?Kcc1)Br(cc1?g J/y) Br(B?Ky) Br(y?pp-J/y)
?
?
from PDG 1.02 0.22
data 0.30 0.11
1.02 0.10
0.30 0.11 0.06
(Previous 90 CL upper limit was lt0.4)
23
Evidence for C1is becoming overwhelming

• B?g J/y only allowed for C1
• same for B?wJ/y (reported earlier)
• M(pp) for X?pp-J/y looks like a r

X?ppJ/y (pp-J/y in an S-wave) c2/dof63/39
X?rJ/y (r-J/y in an S-wave) c2/dof35/39
24
C -1 is ruled outreduce typesize of C-1
entries
25
Remaining states

• 1 r J/y in an S-wave
• 2- a P-wave

26
Fits to the M(pp)Distribution
J/y
X?rJ/y in P-wave has a q3 centrifugal barrier
q
X
r
q
27
M(pp) can distinguish r-J/y S- P-waves
P-wave c2/dof 80/39
S-wave c2/dof 35/39
(CL0.014)
(CL 67)
q roll-off
q3 roll-off
Shape of M(pp) distribution near the kinematic
limit favors S-wave
28
r-J/y in a P-wave is unlikelyreduce type-size of
all J- entries
29
The one which remains 1 (passes all the
tests)
c2/dof 10.4/9
c2/dof 35/39
cosql
M(pp-)
c2/dof 8.1/9

• consistent with
• observations of
• X? g J/y
• X?w J/y

cosc
30
Could the X(3872) be the cc1 ?

• Mass is way off 3872 vs 3929 ?? 3990 MeV
• Br(X?g J/y) 0.3 is much too small

theory range
Rough expectation for pure charmonium
Barnes, Godfrey hep-ph/0311162
G(23P1 ? g J/y) 11 keV
G(23P1 ? ppJ/y) G(y ? po J/y) 0 (0.3 keV)
30
isospin violating
we measure 0.3 two orders-of-magnitude
smaller ? cc1 component of the X(3872) must be
small
31
Summary

• We learn a lot from angular distribs with 50
  evts
• Rule out 0 -, 0 , 1 -- and 1- 2 is
  unlikely (CL0.16)
• B?gJ/y (plus wJ/y M(pp)) rules out all C-1
• M(pp) favors S-wave r-J/y (P-wave CL0.014)
• Rules out 2-, 1- 0-
• Only 1 passes all tests
• angular distributions M(pp) fitted well
• cc1 assignment unlikely (Br(gJ/y) is too small)
• DD molecule models favor 1
• Tornqvist hep-ph/0308277
• Swanson PLB588, 189(2004)

32
plan

• Generate MC for each JPC
• compare MC distributions with data
• Search for X?gy ?(theory for cc1 6x gJ/y )
• PRD on X(3872) properties?
• angular studies
• other decay modes
• X? gJ/y ? include e39 data?
• X?DD, DD ?? ,? separate paper?
• X?3p J/y ? include e39 data?
• M(pp) fits

33
Back-up slides
34
Q why not do 2-d fits for angular distributions?
A We are not measuring parameters, we
doing hypothesis testing, so, we need a
binned c2. With lt 50 events, 2-dim binning
with reasonable bin sizes will have
1event/bin.
35
Q Is the M(gJ/y) look-back plot consistent with
a 10.5 evt X?gJ/y signal?
A YES
13.4 evt Signal
(yield fixed at value from Mbc-DE fit s scaled
from cc1?gJ/y peak)