1. First observation of K+ → p+p0e+e-
Mauro Raggi Laboratori Nazionali di Frascati
NA48/2 rare decay session
CERN 14/07/2011

2. Outline Motivations Data and MC samples Selection cuts
BG estimates and BG cuts
Data MC comparison
BR error estimate

3. K± → p±p0e+e- theory
The interference thanks to the possibility of measuring the plane of polarization of the e+e- pair is splitted into 3 terms:
IB/E, IB/M and E/B.
Where IB/M cancels when integrated over f as in ppg, while E/B is only non-zero if CP violation is allowed.
The only theoretical paper on ppee is currently:
H. Pichl, ``K --> pi pi0 e+ e- decays and chiral low-energy constants,'’ Eur. Phys. J. C 20, 371 (2001) [arXiv:hep-ph/ ].
L. Cappiello, O. Cata, G. D'Ambrosio and D. Gao are currently working on a revised theoretical paper on ppee.
There is no observation of the decay at the moment

4. Theory different contributions
From private communication with D’Ambrosio

5. Analysis targets
Implement and test the IB generator to be used in the NA62FW
Measure the IB BR to better than 5% precision
Measure the plane asymmetry f to better than 2%
Try to spot for deviations from the IB spectrum
Explore the sensitivity of NA62 setup to other components
Establish a trigger recipe to collect ppee in the NA62 experiment (S. Giudici is investigating)

6. Data and MC samples used
Data sample: whole data
Total Kaon flux ~1x1011
MC background generated run by run in SS0
43.2M p±p0g corresponding to 6x1012 K
??M p-p0D corresponding to ??x1011 K
94.5M p±p0p0D corresponding to 2.3x1010 K
MC p+p0e+e- signal
Only 100K events generated with pure phase space

7. Pre-selection cuts This set of cuts is common to ppee and ppd:
SEVT_NCLUSTER <8
SEVT_NTRACK <8
SEVT_NVTX <8
SEVT_NTRACK >3
SEVT_NCLUSTER >3
SEVT_NVTX >0
Loose tracks and clusters quality cuts
NgoodTr ≥3
ngoodcl ≥3
simele ≥2
The same set of cuts has been used to run a data filter (ss0 only) which reduces the data size by factor 100.

8. Basic signal selection cuts
Signal signature cuts:
N good Tracks = p + 2e
N good Clusters = g + 2e
N associated tracks ≥ at least 2e
N non associated clusters = g
Only 1 vertex with cm < ZV < 8000 cm
PID definitions and cuts:
Electron = track with 0.9< E/p < 1.1
Pion = track with E/p < 0.85
Photon = cluster without an associated track
N of electrons = 2 with total electron charge = 0
N of pions = 1
N photons = 2

9. BG rejection cuts General cuts
abs(Mp0 - Mp0PDG) < 10 MeV
COG < 2.5 cm
abs(MK - MKPDG) < 10 MeV
abs(ETOT - PK) < 6 GeV
Mee > 1MeV (physical lower bound)
Cut against gamma conversion (rejects p+p0g)
sqrt(DXe(DCH1)^2-DYe(DCH1)^2) >?? cm
Radiative clusters cut (p+p0D):
Distance extr. e- and gamma cluster >2 cm

10. First signal evidence ~5000 candidates p+p0e+e- in the peak
Are those geniune p±p0e+e-?

11. Background estimates

12. K→pp0p0D background K+→p+p0p0D = p+p0e+e-g need a lost g
MK, COG and PK cuts are effective
BR(p+p0p0D) = 2x(2.07±0.06)x10-4

13. K+→p+p0D background K →p+p0D= p+e+e-g need an additional g
Can profit by accidentals or radiated g from p or e
MK, COG and PK cuts are effective

14. K+→p+p0g background Very dangerous BG K→pp0gIB with g→e+e-
Same final state of the signal
Pass all the mass cuts (mp0, MK) PK and COG cuts
Can get rid of it using dedicated cut on distance of the tracks at DCH1

15. Summary of BG contribution
Decay
%BG
p+p0D
2.1±0.03
p+p0p0D
4.7±0.3
p+p0gIB
< 0.1
Accidental
Total BG
6.8
This is calculated on the whole data sample

16. Wrong sign events data MC
p±p0p0D MC
Wrong signs events have total electron charge of +2 or -2
The mass distribution of WS events is well reproduced by MC of
p±p0p0D only

17. BG estimate data comparison
p+p0D
p+p0p0D
Data points

18. How to reduce p±p0p0D (Mee)
p±p0p0D BG shows a clear peak for Mee above 100MeV
The ppee IB BR becomes so small at so high Mee mass to allow us to have statistics above 100 MeV
Can apply a cut at Mee>100 to reject ~90% of the residual
p±p0p0D
Mee distribution
Data
p±p0p0D

19. An easy handle…Meeg Meeg mass
Can easily get rid of all the Daliz background just cutting on Meeg
Data
MC sum
Introduce a bias in the Mee the variable you want to study
I can’t still evaluate the bias due to absence of MC matrix element

20. Data MC comparison (mass, mee, Zvtx)

21. MK values data and MC MKdat=493.331 DPDG=-0.346 MeV
Still not on the PDG mass.
Need to insert the blue tube corrected directions for the track
Thanks to Evgueni for the suggestion
MK data
MK MC
MKdat= DPDG= MeV
s(MK) dati= MeV
MKMC= DPDG= MeV
s(MK)MC= MeV

22. Mass of the p0 data and MC Mp dati= 135.331 MeV DMPDG= 0.364 MeV
Mp0 data
Mp0 MC
Mp dati= MeV DMPDG= MeV
sMp dati= 2.55 MeV
Mp MC= MeV DMPDG= MeV
sMp MC= 2.10 MeV

23. BR extraction roadmap

24. Normalization channel p+p0D
Br(p+p0)=(20.66±0.08)·10-2; Br(p0)D=(1.174±0.035 ) ·10−2
Br(p+p0)D= (2.43±0.07)·10−3
Same trigger chain 3 track, same tracks p+e+e-
p±p0D mass plot
No significant BG found
Precision to flux measurement limited to 3.% due to p0 Dalitz BR

25. Montecarlo simulation signal
Montecarlo for p+p0e+e- does not exist in CMC
I wrote a pure phase space generator
Can provide a guess of the signal acceptance
Ask Giancarlo for ppee IB matrix element
Acceptance I’m calculating up to now is pure phase space therefore not correct!

26. L2 trigger efficiency measurement
Trigger type used
Bit 1 1VTX (SEVT_TRGW)
Bit 2 2VTX (SEVT_TRGW)
Bit 4 C-PRE (SEVT_TRGW)
Bit 5 MB1TR-P (SEVT_TRGW)
Trigger definition used for efficiency
Control trigger C-PRE
Signal trigger: 1Vtx or 2Vtx or MB1TR-P
To be improved
Got the recipe by Evgueni K3p measurement

27. Trigger efficiency estimates
Used a p±p0p0D with lost gamma and p±p0D for the measurement
Signal BG
p+p0D
1VTX
88.42± 0.08
2VTX
6.29±0.06
1VTX or 2VTX or MB1TR-P
97.5±0.5
96.92±0.04
No evidence for differences in trigger efficiency
Maximum trigger correction is order 1%
Should be done using PU for the control trigger.
Need to measure mee dependence of trigger efficiency

28. Naïve estimate of BR uncertainties
Source
Value
Statistics
1.7%
Background subtraction
0.4%
Normalization (BR p0->Dalitz)
3.0%
Total Uncertainty
3.5%
With the present sample of ~5000 events the total error is already dominated by Dalitz decay BR.
Changing the normalization will introduce difficult systematics due to trigger efficiency which will increase a lot the work to reach a result
Neglecting the trigger efficiency will now change the error by only 0.1%!

29. Measuring the f angle f is the angle in between electron and pion pair
Still trying to understand if this is a CP violating observable.
Measurement uses both calorimeter and spectrometer due to p0 momentum p+ e+ e- p0

30. f Asymmetry sensitivity study
Source
Raw asym values
N sigma
Signal data
± 0.015
2.48
Signal phase space MC
± 0.02
0.02
BG p±p0p0D
± 0.05
0.93
BG p±p0D
0.14 ± 0.22
0.65
Based on partial data set and MC (numbers are not final)
No statistically relevant asymmetry is observed in data
Phase space MC show no asymmetry in the detector and in the reconstruction
Contribution of BG asymmetries to data asymmetry are:
BG(p±p0D)xAsym(p±p0D)=0.14*0.02 =
BG(p±p0p0D)xAsym(p±p0p0D)=0.045*0.047 =

31. Conclusions First evidence of p+p0e+e-
~5000 events found in
BG is order 6%
Easy handles to reduce BG have been identified
ppee MC simulation is not available in CMC007
Pure phase space MC wrote (not enough)
Got a recipe from Evgueni to implement the new generator
First attempt to measure the f angle
Sensitivity is order 2% and no detector bias found