LNF, INFN
May 26th-27th 2005
Rare decays program @KLOE
Matteo Martini
INFN Laboratori Nazionali di Frascati
On behalf of the KLOE Collaboration
DAFNE: the Frascati f - factory
e+e- collider @ s = Mf = 1019.4 MeV
2 interaction regions (KLOE –
DEAR/FINUDA)
Separate e+, e- rings to minimize beam-beam
interactions
Crossing angle: 12.5 mrad ( px(f)12.5 MeV )
M. Martini, K rare decays
1
Integrated luminosity (pb-1)
KLOE: data taking
2000: 25 pb-1
2002
80×106 f decays
2001
2001: 176 pb-1
550×106
2000
f decays
2002: 296 pb-1
920×106 f decays
First
published
results
Analysis
in
progress
Days of running
New KLOE running in progress:
• Lpeak= 1.4 × 1032 cm-2s-1
• 2004: integrated Lum.: 700 pb-1
Goal: collect 2 fb-1 by Dec. 2005
M. Martini, K rare decays
2
The KLOE detector
Superconducting coil
(B = 0.52 T)
Al-Be beam pipe
(spherical, 10 cm ,
0.5 mm thick)
Instrumented
permanent magnet
quadrupoles
(32 PMT’s)
sp/p = 0.4 % (tracks with q > 45°)
sxhit = 150 mm (xy), 2 mm (z)
sxvertex ~1 mm
s(Mpp) ~1 MeV
sE/E = 5.7% /E(GeV)
st = 54 ps /E(GeV)  50 ps
svtx(gg) ~ 2 cm (p0 from KL  p+p-p0)
Drift chamber
Electromagnetic
calorimeter
 Gas mixture: 90% He
+ 10% C4H10
 4 m   3.75 m, CF
frame
 12582 stereo–stereo
sense wires
 almost squared cells
 lead/scintillating fibers
(1 mm ), 15 X0
 4880 PMT’s
 98% solid angle
coverage
M. Martini, K rare decays
3
Kaon production and properties
The f meson decays at rest providing monochromatic and pure kaon beams
KSKL (K+K-) produced in pure JPC = 1-- state:
KS (K+)
KL (K-)
f
i
K ,p
(
2
1
L
Contamination ~10-10
K S , -p - K L , -p K S ,p
)
Tagging: observation of KS,L (K+,-) signals presence of KL,S (K-,+)
precise measurement of absolute BR’s and interference measurement of KS KL
system
lS = 6 mm: KS decays near IP
l = 0.9 m: 60% acceptance for
lL = 3.4 m: Appreciable acceptance
for KL (~0.5lL)
NSL ~106 /pb-1 ; p* = 110 MeV/c
kaon tracking
N+- ~ 1.5106 /pb-1 ; p* = 127 MeV/c
M. Martini, K rare decays
4
Tagging of KS and KL “beams”
KL “crash”
= 0.22 (TOF)
KS  p+pKS  p-e+n
KL  2p0
KL tagged by KS  p+p- vertex at IP
Efficiency ~ 70%
KL angular resolution: ~ 1°
KL momentum resolution: ~ 1 MeV
KS tagged by KL interaction in EmC
Efficiency ~ 30%
KS angular resolution: ~ 1° (0.3 in f)
KS momentum resolution: ~ 1 MeV
4 x105 tags/pb-1
3x105 tags/pb-1
M. Martini, K rare decays
5
Rare Kaon physics at KLOE
Outlook:
KS  3p0
- final results
- prospects @2 fb-1
hep-ex/0505012
submitted to PLB
KS  p+p-p0
- status of the analysis
- prospects @2 fb-1
KS  gg
- preliminary study
M. Martini, K rare decays
analysis in
progress
Paving the
road
6
Search for KS p 0p 0p 0
Observation of KS  3p0 signals CP violation in mixing and/or in
decay:
SM prediction: GS000 = GL000|+000|2, giving:
BR(KS  3p0) = 1.9 x 10-9
Best limit from direct search SND ’99 :
BR(KS  3p0) < 1.4 x 10-5
Interference meas. NA48 ’05:
BR(KS  3p0) < 7.4 x 10-7
Uncertainty on KS  3p0 amplitude limits precision of CPT test from
unitarity (Bell-Steinberger):
( 1 + i tan fSW ) ( e - i md ) =  A* ( K S  f ) A ( K L  f )
CP
CPT
f
CPLEAR ’99 :
Imd = (2.4  5.0) x 10-5 3p0 uncert. dominates
after NA48 meas. : Imd = (-0.2  2.0) x 10-5 error now dominated by h+M. Martini, K rare decays
7
Search for KS p 0p 0p 0
DATA=450pb-1 (2001+2002); MC =0.9fb-1 (all available statistics)
Preselected signal sample (KLCRASH and 6 photons): 39538 events
Normalization Sample (KLCRASH
and 4 photons): 23.5x106 events
-- KS3p0 (MC)
-- MC BKG
 DATA
A kinematic fit is applied on the
Ks side requiring the conservation
of 4-momentum (NDOF=11).
 c2FIT
c2FIT/NDF < 3 is
not enough
(2/3 of bkg rejected)!
Other discriminating variable have to
be used:
(z2, z3)
M. Martini, K rare decays
8
Search for KS p 0p 0p 0
Rejection of bkg:
KS p0p0 + 2 accidental/split g’s
Define signal box in (z2 , z3) plane:
 z3  pairing of 6g clusters with best p0 mass estimates
 z2  best pairing of 4g’s out of 6: p0 masses, E(KS), P(KS), c.m.
angle between p0’s
z2
Data
MC KS  3p0
In the (z2 , z3) plane we define a signal
and five control boxes. The agreement
between DATA and MC, after each
analysis step, is better than 10% in each
region.
Signal generated with BR=10-5 (SND)
M. Martini, K rare decays
z3
9
Search for KS p 0p 0p 0
z2
Data
MC KS  3p0
Signal generated with
BR=10-5 (SND)
z3
Other analysis cuts:
-Track veto to reject events with tracks coming from IP
- Final cut on residual KS energy: E(KS)-SEp
M. Martini, K rare decays
10
Search for KS p 0p 0p 0
Nsel(data) = 2 events selected as signal, with efficiency 3p = 24.5%
Nsel(bkg) = 3.130.82stat0.37sys bkg events expected from MC
Can state: N3p < 3.45 @ 90% CL
Measuring 3p=24.5% from MC generated signal and normalizing signal counts to
KS  p0p0 in the same data set we obtain @90% c.l.:
(
)
BR K S  3p 0 =
N 3p
N 2p
 3p
(
)
BR K S  2p 0  1.2  10-7
 2p
Which translates into a limit on |h000|
@90% c.l.:
h000
(
=
A( K
A K S  3p 0
L
 3p 0
NA 48
)=
)
KLOE
0
 L BR ( K S  3p )
= 0.018
 S BR ( K L  3p 0 )
M. Martini, K rare decays
11
Search for KS p 0p 0p 0
• Increased statistics: x 6.5 improvement
– Luminosity
– Add tagging by KL vertex in DC
x5
 x 1.3
• Increased background rejection
– Largest bkg source after all cuts is the splitting of e.m.
clusters
• Merging procedure removes bkg but leaves signal untouched
• Candidates in data go from 2 to 0, in MC from 3.13 to 2
– Optimization of kinematic fit in progress
– Overall better reduction of the known background expected
If we will be able to suppress the background to a ~negligible level
UL will improve by 6.5 x 1.5 ~ 10
M. Martini, K rare decays
12
Search for KS p +p -p 0
Present status for the
BR(KSp+p-p0):
• Decay amplitude is composed of CPC (3x10-7) and CPV (1.2x10-9) parts
• Direct measurement of the BR is possible using the entire KLOE data set.
• Measurement can be used to verify cPT predictions. These predictions are poorly
tested.
Currently, we have performed the search using a sample of 740 pb-1 of data:
- 373 pb-1 from 2001-2002 data taking
- 367 pb-1 from 2004 data taking
Assuming BR(KSp+p-p0) = 3 x 10-7  230 signal events produced
M. Martini, K rare decays
13
Search for KS p +p -p 0
Selection: KL-crash tag with 2 low momentum tracks from IP
g
KL
Preselection algorithm:
- Require vertex at origin with zero net charge
p-
- Require 2 prompt neutral clusters
g
p+
- Each pair of clusters is a p0 candidate. For each:
- close 3-body kinematics using m(p0), m(KS)
- set t0 using pair of clusters
- use p(KS) and p(f) to search KL-crash cluster in 20° cone
- choose p0 pair that best agrees with reconstructed KL momentum
M. Martini, K rare decays
14
Search for KS p +p -p 0
c2 after preselection
for MC signal and
background
Application of kinematic fit to reject bkg.
Using c2 < 30 (NDOF = 8):
- Cut efficiency = 48.5%
- 98.8% of bkg rejected
- MC(SIG) = 3.3%  3.9 events expected
MC gives 93 bkg events after kinematic fit
g
The ’ bkg are due
to charged kaon
events
p-
g
g
g
g
g
M. Martini, K rare decays
K+
p+
15
Search for KS p +p -p 0
We studied more in detail the various background classes
and developed a set of cuts to reduce them:
K± events
K±pp0 has monochromatic p± momentum at 206 MeV;
cut on p*
Dalitz
dedicated MC production of Dalitz events;
require TCA cuts for significant reduction of this background
After these cuts we still have 3 ’ bkg events
’
±
dedicated MC production of K± (Ke3 + Km3 + ’), K  all
Cut on the energy of prompt clusters not associated to any
tracks or to p0 (Efree)
M. Martini, K rare decays
16
Search for KS p +p -p 0
Compare DATA and MC:
DATA
Standard
Background
MC
DTOF
p*
c2
EFREE
25130 events
in the data with
no cuts applied
M. Martini, K rare decays
17
Search for KS p +p -p 0
At the end of analysis  signal efficiency 1.5%
Preliminary results with 740 pb-1:
- candidates: 6 events
- background: ~ 3.5 events
- observed events consistent with expectation within the statistical
error (100%)
- evaluation of systematic error in progress
Scaling the values of signal and background to 2 fb-1 we expect:
- 16 events, of which 9 background
- 60% statistical accuracy on BR(KSp+p-p0)
M. Martini, K rare decays
18
Paving the road for KS gg
BR differs from CHPT O(p4) by 30%,
useful to fix one O(p6) counterterm
Projections based on
– 150 pb-1 of 2001 background MC
– 10K events of signal MC
With 2 + 0.5 fb-1 we expect
– 500x106 KS events tagged by Klcrash
– N(KSgg, tagged) = 500x106 x 2.8x10-6
= 1400 events
– acceptance 0.4
– Nsig = 560 events
2 fb-1: with good background rejection  ~ 4% statistical error
M. Martini, K rare decays
19
Paving the road for KS gg
bkg
A.U.
– No recover splitting and
A.U.
• Strategy of analysis
MC distributions, no data yet
N. of events in A.U.
signal
large angular acceptance
– Kinematic fit to exploit two
body kinematics
• Distribution of kinematic
• Background separation
looks promising
bkg
signal
Recons.
variables after fit
After fit
M. Martini, K rare decays
20
Conclusion
A direct search for KS3p0 decay has been performed using the whole
statistics collected at KLOE during 2001- 2002 data taking.
We set an upper limit on the branching ratio at:
BR(KS3p0) < 1.2x10-7 @ 90% C.L.
We have started the direct search of the KSp+p-p0.
For the moment we have analyzed only 740 pb-1 of data.
Now we are analyzing the other 300 pb-1 already on disk.
The prospects at 2 fb-1 is promising.
A statistical accuracy of 60% can be reached.
We are paving the road to study KSgg.
With 2 fb-1 we can reach a statistical error of 4% and contribute
to test the cPT prediction for the branching ratio.
M. Martini, K rare decays
21
Backup
Search for KS p 0p 0p 0
Adding in quadrature all the sources of systematic error, we obtain:
 SEL ( 2001) = ( 91.8  0.2 stat  0.3 sys ) %
Using these results and the efficiency on trigger and cosmic veto, we
can calculate the number of events of the normalization sample:
(
)
N K L - crash and K S  2p 0 = ( 37.8  0.2)  106
This value enters directly in the upper limit calculation.
M. Martini, K rare decays
Search for KS p 0p 0p 0
The c22p is built selecting 4 out of 6 clusters
which satisfies better the kinematics of KS  2p0
The parameters used are:
mass distribution
opening angle between pions in
KS C.M. frame
The calibration is done
using KS2p0 sample (see
next slide)
4-momentum conservation
The c23p is based only on the 3 “best reconstructed”
pion masses
M. Martini, K rare decays
Search for KS p 0p 0p 0
In the construction of c2
we use a different sigma
for each sample.
DATA and MC
(OLDMC, NEWMC)
(2001 ,2002 ).
DATA
MC
Mp
Mp
MC
DATA
DE
M. Martini, K rare decays
DE
Search for KS p 0p 0p 0
To better calibrate data and MC, we have also questioned how well
the MC reproduces the amount of double shower fragments and
double accidental clusters. To understand and calibrate this we have
divided the MC KL-crash events into 2 further classes:
 2A: events of Ks2p0 in overlap with 2 accidental (~ 60% )
 2S: events of Ks2p0 with 2 splitted clusters or 1 accidental + 1
splitted cluster (~ 35%)
To do this, we perform a 3 components fit (2S, 2A and fake events)
M. Martini, K rare decays
Search for KS p 0p 0p 0
c22p
c22p
DATA
2S
c23p
c22p
c23p
c22p
Fake
2A
c23p
M. Martini, K rare decays
c23p
Search for KS p 0p 0p 0
Back
Summing up 2001-2002 for each MC, we can compare DATA with the
two different MC productions.
NEW
OLD
DATA s data
MC
s MC
DATA s data
MC
s MC
282
17
283
18,9
5037
71
4870
61,5
452
21
413
26,9
10132
101
9962
146,9
Down
326
18
381
14,2
CDown
22309
149
22636
266,8
282
17
280
17,6
5037
71
4761
59,3
452
21
424
27,1
10132
101
9993
145,6
UP
Cup
326
18
379
13,9
22309
149
22728
269,6
Sbox
CSbox
A reasonable data-MC comparison is found for both samples
at the beginning of the analysis.
M. Martini, K rare decays
Search for KS p 0p 0p 0
 DATA
-- MC ALL
ALL
We apply a track veto
to reject events with
tracks coming from
IP.
c23p
14<c22p<40
c22p<14
c23p
c22p>40
We reject events with:
rPCA < 4 cm
|ZPCA| < 10 cm
c23p
M. Martini, K rare decays
c23p
Search for KS p 0p 0p 0
 DATA
-- MC ALL
TRK veto
+
DE/sE
ALL
c23p
c22p central
c23p
M. Martini, K rare decays
c22p down
c23p
c22p up
c23p
Search for KS p 0p 0p 0
 DATA
-- MC ALL
END OFANA
ALL
c23p
c22p central
c23p
M. Martini, K rare decays
c22p down
c23p
c22p up
c23p
Search for KS p 0p 0p 0
Comparison between DATA and MC after the optimization
procedure.
DATA
s data
MC
s MC
EV TOT
s tot
Sbox
2
1
3,125
0,8
17
4
CSbox
520
23
446,5
10,1
2402
49
UP
0
0
0
0,0
0
0
Cup
4
2
3,2
0,8
17
4
Down
3
2
2,45
0,8
11
3
CDown
326
18
388,5
9,6
1961
44
Nobs = 2
Bexp = 3.13 ± 0.82
M. Martini, K rare decays
Search for KS p 0p 0p 0
Source
DBexp/Bexp
MC calibration
1.6 %
Fakes definition
1.8 %
Track veto
4.8 %
Energy resolution
6.6 %
Energy scale
6.7 %
c2FIT
5.0 %
TOTAL
11.5 %
Adding in quadrature all the contribution found, we obtain:
Bexp = 3.13 ± 0.82stat ± 0.37sys
M. Martini, K rare decays
Search for KS p 0p 0p 0
For each sample we generate 3 poissonian centered
around N2S, N2A
and Nfake. Weighting each
distribution for the proper calibration factor, obtained
with the 2D-fit, we sum them up to build our PDF
for the background.
The obtained PDF is reasonably similar to a Gaussian
with average 3.15 and with equivalent to the ext. stat.
Folding the previous PDF with a Gaussian
distribution centered around 0 and with a width
equivalent to the whole systematic uncertainty we
obtain the PDF for bkg that we use on the upper limit
calculation.
The RMS include statistic and systematic uncertainty.
M. Martini, K rare decays
Search for KS p 0p 0p 0
Back
THE UNIFIED APPROACH: when Bexp is greater than Nobs,
the classical method does not provide a perfect coverage. In this
condition is better to use Feldman-Cousin-Neyman method based on
likelihood ratio.
We observe Nobs = 2 candidates on data and we estimate:
Bexp = 3.13 ± 0.82stat± 0.37sys
In our case, the exp. background is greater than Nobs.
With this counting we obtain the following UL @ 90% C.L.:
Classical
Neyman
N3p = 2.42
N3p = 3.45
M. Martini, K rare decays
Search for KS p +p -p 0
Current values for the BR
Decay amplitude is composed of CPC (3x10-7) and CPV (1.2x10-9)
Selection requiring KL-crash, 2 tracks from IP (zero net charge), 2 prompt neutral
clusters
• Data 740 pb-1 (2001+2002+ some 2004 runs)
• Assuming BR=3x10-7  230 signal events produced
• Major bkg: charged kaons, dalitz, ’
• bkg reduced using: TCA , DTcut, P*
• Use kinematic fit to select events kinematically closed (c2 cut)
• veto events with additional neutral prompt clusters below acceptance.
The set of cuts designed grants SIG 1.38% with very high background rejection
We expect to measure this BR integrating the whole collected data so far.
M. Martini, K rare decays
Search for KS p +p -p 0
M. Martini, K rare decays
Back
Search for KS p +p -p 0
M. Martini, K rare decays
Back
Search for KS p +p -p 0
M. Martini, K rare decays
Back
Search for KS p +p -p 0
M. Martini, K rare decays
Back
Search for KS p +p -p 0
Summary of signal efficiency:
M. Martini, K rare decays
Search for KS p +p -p 0
Because of Data/MC discrepancies,
especially in DTOF, we want to obtain
Nbkg from sidebands in data after cuts
on DTOF and p*.
Assuming no correlation beween c2
and EFree,
A/B=C/D
Study systematics by inverting cuts,
varying acceptances
M. Martini, K rare decays
Search for KS p +p -p 0
M. Martini, K rare decays
Search for KS p +p -p 0
M. Martini, K rare decays
Conclusion
With the data collected during 2001-2002 data taking, KLOE has:
- determined the best upper limit on KS3p0
- measured the main KL BR’s with 0.5% accuracy
- measured in two independent ways the KL lifetime
with 0.5 % accuracy
 Important contribution to the measurement of Vus
Next in line:
- Direct search for BR(KSp+p-p0)
- Final result on KSpen BR
- Analysis of KL semileptonic form factor slopes
KLOE expects to collect 2 fb-1 by the end of 2005:
thus allowing to improve the search for rare KS decays and KSKL interference
studies.
M. Martini, K rare decays