Recent KLOE results on rare KSKL processes M

Recent KLOE results on rare KSKL processes M. Martini INFN, laboratori di Frascati on behalf of the KLOE collaboration Manchester, 19/07/2007

Dafne: the Frascati f-factory Ø e+e– collider with 2 separate rings s 1020 Me. V M Ø peak 3 mb Ø Best performances in: v Lpeak = 1. 4 × 1032 cm 2 s 1 v Lint/day = 8. 51 pb 1 2. 5 fb 1 on tape @ √s=M ≈ 8× 109 produced 250 pb 1 @ √s=1000 Me. V + 4 scan points around the EPS 2007, Manchester M. Martini 19/07/2007

The KLOE experiment The KLOE design was driven by the measurement of direct CP violation through the double ratio: R = (KL p+p ) (KS p 0 p 0) / (KS p+p ) (KL p 0 p 0) and by the KL lifetime (4 m 3. 3 m) 90% He; 10% i. C 4 H 10 Stereo geometry 52140 wires lead/scint. fibers barrel-endcap 15 X 0 thickness 4880 PM 98% coverage p/p = 0. 4% x/y = 150 mm z = 2 mm vtx ~ 3 mm (Mp+p-)~1 Me. V E/E = 5. 7% / E(Ge. V) t = 54 ps / E(Ge. V) 140 ps vtx(KL p 0 p 0) ~ 1. 5 cm Beryllium beam pipe = 10 cm 0. 5 mm thickness Superconducting coil B=0. 6 T EPS 2007, Manchester M. Martini 19/07/2007

Kaon tagging KL “crash” b= 0. 22 (TOF) K S p +p - K S p -e +n KL 2 p 0 KS tagged by KL interaction in Em. C Efficiency ~ 30% (largely geometrical) KS angular resolution: ~ 1° (0. 3 in ) KS momentum resolution: ~ 2 Me. V EPS 2007, Manchester KL tagged by KS p+p- vertex at IP Efficiency ~ 70% (mainly geometrical) KL angular resolution: ~ 1° KL momentum resolution: ~ 2 Me. V M. Martini 19/07/2007

Talk layout Measurement of BR(KS gg) Direct search for KS e+e. Measurement of BR(KL peng) CPT test with Bell-Steinberger relation QM test in KSKL system EPS 2007, Manchester M. Martini 19/07/2007

Motivations for a new BR(KS gg) measurement It is a good test for Ch. PT (PRD 49 (1994) 2346) Experimental value of the BR changed along the years From 2003 it is known with a small error (3%) : BR(KS gg) = (2. 71 ± 0. 06 ± 0. 04) x 10 -6 due to a measurement of NA 48/1 collaboration Differs from Ch. PT O(p 4) by 30% (possible large O(p 6) contribution). NA 48/1 In NA 48, the KL gg background is a relevant component of the fit. In KLOE, the background from KL is reduced to 0 (tagging). First measurement of this decay with a pure KS beam. EPS 2007, Manchester M. Martini 19/07/2007

Strategy for BR(KS gg) measurement Data sample analyzed: 1. 6 fb-1 DATA - KS tagged from KL interacting in EMC (122 x 106 events) - 2 prompt photons required ( 496000 events) 2 p 0 Background is made of KS with 2 lost photons in the pipe or interacting in the calorimeter covering focusing quads (QCAL) events with in time hits on QCAL vetoed. cos(qgg**) BKG Background rejection from kinematic fit Event counting on the scatter plot Mgg vs qgg, where: cos(qgg**) SIG - qgg Opening angle between the two photons in the KS cms - Mgg Reconstructed gg mass cos(qgg**) EPS 2007, Manchester M. Martini 19/07/2007

KS gg: Fit results FCN/Ndof = 1. 2 To extract the number of signal, the 2 D-plot in data is fit using signal and background shapes from MC Nsig = 600. 3 ± 34. 8 cos(qgg**) (5. 8% stat. error) • • DATA -- MC all Signal Background Mgg (Me. V) EPS 2007, Manchester M. Martini 19/07/2007

KS gg: efficiencies and normalization After tagging, the events KS 2 p 0 events are used as normalization sample. The BR is then extracted as: • For the signal: e. SIG (tot| KL-crash) = e(presel) x e(veto) x e(c 2) = = (50. 8 ± 0. 6)% e(presel) (83. 2 ± 0. 2)% e(veto) (96. 5 ± 0. 4)% e(c 2) (63. 3 ± 0. 7)% • For the normalization sample, KS 2 p 0 events counted selecting 4 prompt photons: e 2 p 0 (tot | KL-crash) = (65. 0 ± 0. 2 stat ± 0. 1 sys)% N 2 p 0 / e 2 p 0 (tot | KL-crash) = 159. 8 Mevts Systematics due to application of data-MC correction curve for cluster efficiency. Cross checked with counting (3 -5) prompt photons (159. 5 Mevts) EPS 2007, Manchester M. Martini 19/07/2007

KS gg: final BR result Source +Syst (%) -Syst (%) Signal acceptance 0. 12 QCAL 0. 88 0. 51 c 2 cut 0. 44 c 2, qgg* scale from signal --- 0. 55 Fit procedure 0. 88 0. 44 Energy scale --- 1. 32 Norm sample 0. 15 Total +1. 33 -1. 65 c. PT NA 31 NA 48/99 NA 48/03 O(p 4) O(p 6) 2. 9 far from NA 48 result KLOE - The NA 48 measurement implied the existence of a sizeable O(p 6) counterterm in Ch. PT. Our number makes this contribution practically negligible EPS 2007, Manchester M. Martini 19/07/2007

Direct search for KS e+e. SM prediction is low but precise BR(KS e+e-) = 1. 6 10 -15 [Ecker, Pich 91] event selection (1. 32 fb-1 ) c 2 • KS tagged by KL crash • 2 tracks from IP to Em. C with Minv [e+e- hypo] > 420 Me. V c 2 -like variable based on: - TOF of the 2 particles KS p+p- mp K S p +p f p +p -p 0 K S e + esignal box - E/p - distance between track impact point and cluster centroid • P* (p hypo) in the KS rest frame 220 Me. V • Mmiss 380 Me. V to reject residual p+p p 0 EPS 2007, Manchester M. Martini Minv [e+e hypo] (Me. V) 19/07/2007

Upper limit on KS e+e§ Signal box MC optimization: (492 Minv 504) Me. V and c 2 20 Nobs = 3 and m. BKG = 7. 1± 3. 6 § we extract UL(msig) = 4. 3 @ 90% CL using bayesian approach § normalize to KS pp(g) counts in the same data set: UL(BR) = UL(msig) sig = presel g-rad pp = 0. 6 , Npp ~ 1. 5 108 pp BRpp sig Npp (E*g < 6 Me. V) = 0. 785 0. 888 0. 8 = 0. 558 KLOE preliminary: BR(KS e+e-(g)) < 2. 1 10 -8 @ 90% CL EPS 2007, Manchester M. Martini CPLEAR: < 1. 4 10 -7 19/07/2007

BR measurement of KL peng - We measure R=BR(Ke 3 g; E*g>30 Me. V, q*lep-g>20°)/BR(Ke 3(g)), using a 328 pb-1 2001 -2002 data sample; - Both IB and DE emission contribute to R; - Separation between IB and DE never measured; for the first time the DE contribution is measured; - E*g-q*ele-g reconstructed by kinematic closure based on cluster position and tracking EPS 2007, Manchester M. Martini 19/07/2007

KL peng: final results • Fit 2 D plot of E*g and q*e-g with the MC shapes we measure: R= BR(Ke 3 g; E*g 30 Me. V, q*e-g 200) BR(Ke 3(g)) theory [Gasser et al. , EPJ 40 C (2005)205 ]: R = (0. 96 ± 0. 01)% (uncertainty mainly due to the DE term) R = (0. 924 ± 0. 023 stat ± 0. 016 syst)% q*e-g (deg) According to Gasser et al. the spectrum can be parametrized as: MC E*g (Me. V) EPS 2007, Manchester M. Martini 19/07/2007

CPT test: Bell-Steinberger relation CPT test from unitarity based on KS-KL observables: (1 + i tan SW) [Re i Im ] 1 S f A*(KS f ) A(KL f ) f f + 0 t. S/t. L h+ 0* B(KL p+p p 0) + h+ B(KS p+p ) + g h+ B(KS p+p g) 00 h 00 B(KS p 0 p 0) kl 3 2 t. S/t. L B(KLl 3) [Re Re y i Im x+ ] 000 t. S/t. L h 000* B(KL p 0 p 0 p 0) 2 t. S/t. L B(KLl 3) [(AS+AL)/4 i Im x+] KLOE contributions: KS semileptonic asymmetry, UL on BR(KS p 0 p 0) Im x+ from a combined fit of KLOE + CPLEAR data before NA 48 and KLOE measurement Im( ) limited by h 000 main uncertainty now comes from h+-- trough +EPS 2007, Manchester M. Martini 19/07/2007

CPT test: Bell-Steinberger relation KLOE result (JHEP 0612: 011, 2006): Re (159. 6 1. 3) 10 5 Im (0. 4 2. 1) 10 5 CPLEAR: Re (164. 9 2. 5) 10 5 Im (2. 4 5. 0) 10 5 Assuming DG=0, i. e. no CPT in decay: -5. 3 10 -19 Ge. V < DM < 6. 3 10 -19 Ge. V at 95% C. L. EPS 2007, Manchester M. Martini 19/07/2007

QM coherence The decay time difference distribution for KS p+p , KL p+p allows to measure Dm and decoherence term S, L. PLB 642(2006) 315 From CPLEAR data: EPS 2007, Manchester In the B-meson system, BELLE: M. Martini 19/07/2007

CPTV and quantum gravity In presence of decoherence and CPT violation induced by quantum gravity (CPT operator “illdefined”) the definition of the particle-antiparticle states could be modified. This in turn could induce a breakdown of the correlations imposed by Bose statistics (EPR correlations) to the kaon state [Bernabeu, et al. PRL 92 (2004) 131601, NPB 744 (2006) 180]: |w| could be at most: KLOE result (w measured for the first time) Im w PLB 642(2006) 315 with L=2. 5 Re w fb-1: EPS 2007, Manchester M. Martini 19/07/2007

CPTV and quantum gravity KLOE has now ~2. 5 fb-1 data on disk Preliminary results based on 1 fb-1 : c 2/dof=29/31 S, L = 0. 009 0. 022 STAT 0, 0 = (0. 03 0. 12 STAT) × 10 -5 1 fb-1 published result (380 pb-1) S, L = 0. 018 0. 040 0. 007 0, 0 =(0. 10 0. 21 0. 04) × 10 -5 Preliminary results on w based on 1 fb-1 : EPS 2007, Manchester M. Martini published result (380 pb-1) 19/07/2007

Conclusions • KLOE has obtained new results on: - BR (KS gg) = (2. 27 ± 0. 14) 10 -6 - BR(KS e+e-(g)) < 2. 1 10 -8 @ 90% CL - KL peng: R = (0. 924± 0. 023 stat ± 0. 016 syst)% <X> = (-2. 3 ± 1. 3 stat ± 1. 4 syst), • Improved accuracy of CPT test with Bell-Steinberger relation • Several parameters related to CPT and QM tests are measured at KLOE, Re(w) and Im(w) for the first time. EPS 2007, Manchester M. Martini 19/07/2007

SPARES

Motivations. . . II - BR obtained by NA 48 from a fit to the Z vertex distribution (KL gg background is a relevant component in the fit) - In KLOE there is not background from KL gg so we can perform the first measurement of this decay with a pure KS beam - We can reach an accuracy of about 5 -6% , twice larger than NA 48 but with completely different systematics and background. . EPS 2007, Manchester M. Martini 19/07/2007

QCAL detector The QCAL tile calorimeters of KLOE are two compact detectors placed closed to the interaction point and surrounding the focalization quadrupoles. Their purpose is to increase the hermiticity of KLOE calorimeter. Each QCAL consists of a sampling structure of lead plates and 1 mm scintillator tiles. EPS 2007, Manchester M. Martini 19/07/2007

QCAL veto DTqcal distribution: Comparison between a KS p 0 p 0 and a KS p+p sample EPS 2007, Manchester M. Martini 19/07/2007

QCAL data/MC efficiency - In each period, the event fractions with Ng=2, 3, 4 have been fit with the following technique: we calculate the ratio: We found compatible value of R for the different DATA sample 2001 2002 2004 2005 R 0. 72 ± 0. 01 0. 83 ± 0. 01 0. 81 ± 0. 01 0. 78 ± 0. 01 EPS 2007, Manchester M. Martini 19/07/2007

QCAL data/MC efficiency results Using the results on R and the Ploss for the different DATA samples, we can correct the MC QCAL efficiency for the signal. Now we have also extract the efficiency on signal from MC: For the complete sample we found: EPS 2007, Manchester M. Martini 19/07/2007

Energy scale and efficiencies KL gg control sample selected to further check the energy scale on data-MC Signal and normalization sample free of KL gg Inclusive energy of the 2 g photons background Eg (Me. V) EPS 2007, Manchester M. Martini 19/07/2007

Inclusive photon energy barrel ecap 4 g, KS 2 p 0 • • Data -- MC barrel ecap 2 g, BKG EPS 2007, Manchester M. Martini 19/07/2007

Energy scale and efficiencies KL gg control sample selected to further check the energy scale on data-MC Signal and normalization sample free of KL gg KS gg MKL (Me. V) EPS 2007, Manchester MKS (Me. V) M. Martini 19/07/2007

Energy scale calibration… results We fit the distribution of reconstructed mass for data and MC at the end of analysis chain to check the energy scale calibration. Rt (cm) MKL data (Me. V) MKL MC (Me. V) (before calib) MKL MC (Me. V) (after calib) DMKL (Me. V) (Data – MC) (1 – 30) 496. 2 ± 0. 8 488. 7 ± 0. 4 495. 4 ± 0. 5 0. 8 ± 0. 9 (35 – 65) 495. 0 ± 0. 9 488. 1 ± 0. 5 494. 7± 0. 5 0. 3 ± 1. 0 (65 – 95) 494. 0 ± 1. 0 487. 1 ± 0. 6 493. 8 ± 0. 6 0. 2 ± 1. 2 (95 – 125) 494. 3 ± 1. 2 486. 1 ± 0. 7 492. 4 ± 0. 7 1. 9 ± 1. 4 (125 – 155) 492. 5 ± 1. 5 483. 3 ± 0. 9 490. 2± 0. 9 2. 3 ± 1. 8 (155 – 185) 484. 8 ± 4. 6 475. 6 ± 1. 9 481. 4 ± 1. 9 3. 0 ± 5. 0 After our scale correction, the data-MC scale agrees at (0. 2 ± 0. 2)% EPS 2007, Manchester M. Martini 19/07/2007

Energy scale systematics Since we still have a difference of (0. 2 ± 0. 2)% between data and MC on energy scale, we can extract a systematics varying of 0. 2% and 0. 4% Mgg from MC (signal and bkg). Variation Nsig BR Standard 600. 3 ± 34. 8 2. 27 ± 0. 13 x 1. 002 596. 8 ± 35. 4 2. 26 ± 0. 13 x 1. 004 591. 6 ± 35. 7 2. 24 ± 0. 14 We have a systematics of -0. 03 on BR value EPS 2007, Manchester M. Martini 19/07/2007

Systematics…. DTqcal distribution: Comparison between a KS p 0 p 0 and a KS p+p sample - From KSGG sample Ploss_mean = (3. 51 ± 0. 04)% - From KS 00 sample: Ploss_mean = (3. 55 ± 0. 06)% -From KS+- sample: Ploss_mean = (3. 29 ± 0. 05)% Ploss_win = (3. 31 ± 0. 05)% Summarizing, we can extract 2 different systematics: Dmean(KS 00 vs KS+-) = 0. 26 Dout-win(KS+- vs KS+-) = 0. 02 Summing up, we obtain 0. 26% EPS 2007, Manchester M. Martini 19/07/2007

Try to move qcal win To extract the systematics on QCAL cut, we varied the windows around the chosen cut Cut Nsig BR x 10 -6 Standard (-5: 5) 600. 3 ± 34. 8 2. 27 ± 0. 13 qcal = 96. 47% (-4: 4) 601. 3 ± 35. 1 2. 26 ± 0. 14 qcal= 97. 18% (-6: 6) 602. 2 ± 35. 0 2. 29 ± 0. 14 qcal= 95. 76% Systematics=+0. 02, -0. 01 on BR value EPS 2007, Manchester M. Martini 19/07/2007

Try to move c 2 FIT cut To extract the systematics on c 2 FIT, we slightly varied the value of the cut Cut Effi ana Nsig Nbkg BR x 10 -6 10 38. 3 ± 0. 7 373. 0 ± 25. 9 341. 0 ± 35. 6 2. 29 ± 0. 16 16 56. 2 ± 0. 7 530. 1 ± 33. 1 1079. 9 ± 50. 4 2. 28 ± 0. 14 18 60. 0 ± 0. 7 567. 5 ± 35. 1 1380. 5 ± 55. 4 2. 26 ± 0. 14 20 63. 3 ± 0. 7 600. 3 ± 34. 8 1678. 7 ± 62. 9 2. 27 ± 0. 13 22 66. 1 ± 0. 7 630. 1 ± 35. 3 1914. 9 ± 70. 4 2. 28 ± 0. 13 24 68. 2 ± 0. 7 650. 5 ± 38. 5 2295. 4 ± 76. 0 2. 28 ± 0. 14 Systematics=± 0. 01 on BR value EPS 2007, Manchester M. Martini 19/07/2007

Try to move bins in scatter plot To extract a systematics on fit procedure, we slightly change the bins on the 2 d distribution used in HMCLNL. Mgg bin Cos(qgg) bin Nsig Nbkg BR x 10 -6 30 50 600. 3 ± 34. 8 1678. 7 ± 62. 9 2. 27 ± 0. 13 25 45 603. 0 ± 35. 0 1676. 0 ± 63. 1 2. 28 ± 0. 13 35 55 596. 7 ± 33. 9 1682. 3 ± 62. 5 2. 26 ± 0. 13 25 55 606. 8 ± 35. 1 1672. 2 ± 62. 8 2. 29 ± 0. 14 35 45 597. 8 ± 34. 6 1681. 2 ± 61. 5 2. 26 ± 0. 13 Systematics= (+0. 02 ; -0. 01) on BR value EPS 2007, Manchester M. Martini 19/07/2007

Cumulative … c 2 FIT below DCH Using signal from KL gg, we can extract a systematics on c 2 fit, building a cumulative for data and MC and checking the ratio. To reject the bkg we use a preliminary cut with c 2 fit <50 and qgg<. 998 Using the value at c 2 fit =20, we obtain a systematics of EPS 2007, Manchester M. Martini -0. 41% 19/07/2007

Cumulative … qgg below DCH Using the same technique developed for c 2 fit, we can evaluate a systematics on qgg To reject the bkg we use a preliminary cut with c 2 fit <50 and qgg<. 998 Using the values at cos(qgg)=0. 999, we obtain a systematics of EPS 2007, Manchester M. Martini -0. 37% 19/07/2007

Fast simulation of Background To study the fit uncertainty as a function of MC statistics we have developed a method based on “hit or miss”. The procedure is only based on MC signal and background. Recipe for two components: Use the original 2 d-distribution from signal and bkg, to create 2 smoothed distribution; Use hit or miss to create N different distributions for signal and background of different “fast” MC statistics; Create a fake data distribution using signal and bkg from hit or miss with statistics as in data sample; Repeat the fit procedure N times increasing the “fast” MC statistics. EPS 2007, Manchester M. Martini 19/07/2007

Fast simulation: Mgg vs qgg MC original distributions MC from hit or miss Fake DATA EPS 2007, Manchester M. Martini 19/07/2007

Hit or miss Signal and bkg statistical error as a function of the “fast” MC statistics When we performed this study, we had 0. 5 fb-1 of full MC. Now we have 1. 1 fb-1 and we obtained a lower uncertainty. We have already practically reached the plateau region. EPS 2007, Manchester (01 -02) = 13. 9% , (04 -05) = 7. 5% (01 -02) = 12. 0% , (04 -05) = 6. 8% M. Martini 19/07/2007

Background enriched sample Using HMCLNL we fit the 2 D-plot for the bkg enriched sample (50 < c 2 < 500). We obtain the bkg weight “ ”, that is used to estimate the bkg in the signal region. • • DATA -- MC all Signal Background EPS 2007, Manchester M. Martini 19/07/2007

Background enriched sample Subtracting the obtained background to data, we obtain the expected number of signal events Procedure Nsig (upper band) BR x 10 -6 Standard 600. 3 ± 34. 8 2. 27 ± 0. 13 Bkg enriched sample 608. 2 ± 71. 4 2. 30 ± 0. 27 We obtain a results compatible with the number of signal events evaluated with the standard analysis. The larger error on Nsig(upper band) is dominated by the poissonian uncertainty on the number of signal events. EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: preselection EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: c 2 (I) EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: c 2 (II) EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: c 2 (III) EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: p* in region 1 EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: cut on Nprompt EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: cut on missing mass EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: analysis chain EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Ecrash vs b* in sidebands EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: optimization EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: UL evaluation EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: CPLEAR result SM prediction is small but precise: BR(KS e+e-)=1. 6 x 10 -15 (Ecker, Pich 91) leaving room for possible new physics effects to be detected. The most precise measurement up to now is done by CPLEAR using or Selection is done by performing a kinematical fit to the hypothesis: with 9 constraints. At the end: N(data)=0 N(MC)=0. 22 ± 0. 10 EPS 2007, Manchester BR(KS e+e-) < 1. 4 x 10 -7 (90% C. L. ) M. Martini 19/07/2007

KS gg: checking angular distribution Compare Data with tuned MC sample after fit. Inclusive distribution of the photon polar angles. • • DATA -- MC all Signal Background EPS 2007, Manchester M. Martini 19/07/2007

Direct search of KS e+e. SM prediction is small but precise: BR(KS e+e-)=1. 6 x 10 -15 [Nucl. Phys. B 336, 189, 1991] leaving room for possible new physics effects to be detected. The most precise measurement done by CPLEAR: BR(KS e+e-) < 1. 4 x 10 -7 (90% C. L. ) In KLOE we can perform a direct search of this decay using a pure KS beam. Data sample analyzed: 1. 32 fb-1 Starting normalization sample: 148 Mevts (KL-crash and KS p+p-) Preselection: - KS tagged by KL-crash - 2 tracks from IP with opposite curvature - Invariant mass in e+e- hypothesis Minv > 420 Me. V e(sig) = esig(KL-crash) x esig(presel. | KL-crash) 0. 3 x 0. 785 = 0. 24 After preselection: 1. 1 Mevts in Data sample EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Analysis strategy For signal identification, the calorimeter information is used to build a c 2 -like variable based on: - Sum and difference of (Tclu-To. F) of the 2 particles - E/p of both particles - Transverse distance between track impact point and the closest cluster Signal box We define a signal box in the plane: (c 2 vs Minv) Minv is evaluated in e+e- hypothesis Side-bands are defined in the invariant mass spectrum: - to define background normalization - to check Data/MC agreement after further cuts are applied EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Background composition From MC… The sources of background from MC are: A: KS p+p- mp B: KS p+p. C: f p+p-p 0 K S e + e- - KS background events enter preselection because of track resolution - p+p-p 0 events are selected with an accidental cluster satisfying the KL-crash algorithm The relative fraction of background in each region is: Bkg type Region 1 Region 2 Region 3 KS p+p pm 45. 8% 13. 2% 0. 6% KS p+p 53. 9% 65. 1% 2. 2% p+p p 0 0. 3% 21. 7% 97. 2% EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Background calibration in reg. 3 The scale factor for p+p p 0 component (background C) is evaluated in region 3, where the other decay modes give a negligible contribution: f. C 2001 -2002 2005 1. 99 ± 0. 06 2. 24 ± 0. 04 EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Background calibration in reg. 1 With f. C fixed, we can check MC prediction for A (KS p+p pm) and B (KS p+p ) background components. Fitting Minv spectra in region 1: 2001 -2002 2005 f. A 0. 59 ± 0. 01 0. 67 ± 0. 01 f. B 1. 80 ± 0. 01 2. 39 ± 0. 01 EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Background rejection in reg. 2 To reject p+p- and mp background events, we require : P*(p hyp) in KS rest frame > 220 Me. V epp, mp = 0. 014 esig = 0. 962 EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Background rejection in reg. 2 To reject p+p- and mp background events, we require : P*(p hyp) in KS rest frame > 220 Me. V epp, mp = 0. 014 esig = 0. 962 To reject p+p-p 0 contamination we require: Mmiss > 380 Me. V and Nprompt < 2 Where Mmiss is evaluated from momentum and tracks momentum (p hyp. ) eppp = 0. 001 esig = 0. 998 EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Sbox Optimization and UL - Optimization of signal box definition on Monte Carlo: (492 < Minv < 504) Me. V and c 2 < 20 obtained varying simultaneously Minv (± n m) and c 2. Only 2001 simulatio n so fa 2002 run of r for s optim used izati on The “optimized values” are chosen looking at: mbkg , msig and signal efficiency. - In the signal box: Nobs = 3 mbkg = 7. 1 ± 3. 6 - UL(msig) evaluated numerically with bayesian approach, taking into account background fluctuation (NIM 212 (1983) 319 -322) UL (msig) = 4. 3 @ 90% CL EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Radiative corrections Considering radiative corrections, there are two possible processes contributing to photon emission (not interfering): 1) KS e+e- + IB photon emission 2) KS gg* g e+e. Given the Minv cut, we actually measure the upper limit on: BR(KS e+e- (g) with E*g < 6 Me. V) A limit on the second process is spoiled out in this Minv range by a factor of 10 -8. The cut used in Minv corresponds to an efficiency correction of: This factor must be included in SIG. EPS 2007, Manchester M. Martini 19/07/2007

KS e+e-: Upper limit on BR Normalizing signal counts to KS pp(g) counts in the same data set: esig (tot | KL-crash)= esig(presel. |KL-crash) x ecut x ag-rad = 0. 785 x 0. 888 x 0. 8 = 0. 558 g-rad acceptance of the radiated photons for E*g < 6 Me. V epp (tot | KL-crash)= 0. 6 Npp 1. 5 x 108 KLOE preliminary EPS 2007, Manchester M. Martini 19/07/2007

CPT test: Bell-Steinberger relation + + g 000 (KS) + 0 kl 3 (KS) Main improvements: KS semileptonic asymmetry, UL KS p 0 p 0 Im x+ from a combined fit of KLOE + CPLEAR data before NA 48 and KLOE measuremnt Im( ) limited by h 000 main uncertainty now comes from h+-trough +EPS 2007, Manchester M. Martini 19/07/2007

Decoherence and CPT violation Deviations from QM could be due to Quantum Gravity effects which could cause pure state to evolve into mixed states: loss of quantum coherence. Modified Liouville – von Neumann equation for the density matrix of the kaon system: extra term inducing decoherence: pure state => mixed state Model of decoherence for neutral kaons => 3 new CPTV param. , b, g (NP B 241 (1984)) At most: CPLEAR and KLOE have tested this model in single kaon and entangled kaon pair systems, respectively EPS 2007, Manchester M. Martini 19/07/2007

Decoherence and CPT violation Usingle kaons from , Fit simultaneously at 90% CL PLB 364, 239 (1999) EPS 2007, Manchester M. Martini 19/07/2007

Decoherence and CPT violation KLOE fit I(Dt; p+p , p+p ) under the assumptions of complete positive (it holds for entangled systems) i. e. a=g and b=0 : only g is fitted KLOE 380 pb-1 Fit I(Dt; p+p , p+p ; g ) (complete positivity assumption) g is measured for the first time in the entangled kaon system EPS 2007, Manchester M. Martini 19/07/2007

QM coherence • Fit including Dt resolution and efficiency effects + regeneration • S, L Dm fixed from PDG KLOE result: PLB 642(2006) 315 with 2. 5 fb-1 ± 0. 8 10 -6 From CPLEAR data: In the B-meson system, BELLE coll. : EPS 2007, Manchester M. Martini 19/07/2007

CPT test using KS, L pen Semileptonic charge asymmetries provide CPT tests _ + + (KS, L p e n) AS, L = _ + + (KS, L p e n) If CPT holds, AS=AL =2 Re e AS AL signals CPT violation in mixing and/or decay with DS DQ AL = (3. 322 0. 058 0. 047) 10 3 [KTe. V 2002] AL = (3. 317 0. 070 0. 072) 10 3 [NA 48 2003] from KLOE (450 pb-1): AS = ( 1. 5 9. 6 2. 9 ) 10 3 With 2. 5 fb 1, AS 3× 10 3 Using AL = (3. 34 ± 0. 07)x 10 -3 from KTEV, from AS-AL: Use Re( ) from CPLEAR, x 5 improvement for error on Re(x-) From AS+AL: First determination of Re(y) independently of B-S relation EPS 2007, Manchester M. Martini 19/07/2007

Analysis of KL peng measurement of BR and the Direct Emission term in the g spectrum inclusive selection (328 pb-1 ): radiative sample selection: Eclu -Egeval (Me. V) • KLg vtx -> comparing To. F of KL and the g • KL tagged by KS p+p cluster time • (Emiss-|Pmiss|) in different mass • cluster position to close the kinematic and hypothesis to remove ~90% of bck evaluate Eg -> pn 2 = 0 = (p. K-pp-pe-pg)2 • To. F to separate e/p (after PID ~ 0. 7% contamination) 2 106 Ke 3 Signal Ke 3 g out of acceptance not radiative Ke 3 p+p-p 0 Km 3 MC Eclu(Me. V) EPS 2007, Manchester bck reduction: • this cut to remove not radiative Ke 3 • Eclu 25 Me. V to remove accidentals • NN trained with Em. C infos to remove Km 3 and p+p p 0 KL p+p p 0 control sample to check g efficiency, energy and vertex resolutions M. Martini 19/07/2007

KS gg: final BR result Various sources of systematics have been considered: Source +Syst (%) -Syst (%) Signal acceptance 0. 12 QCAL 0. 88 0. 51 c 2 cut 0. 44 c 2, q* gg scale from signal --- 0. 55 - c 2, qgg scale: check data-MC scale difference using KL gg Fit procedure 0. 88 0. 44 - Fit procedure: change bins size in 2 D distribution Energy scale --- 1. 32 Norm sample 0. 15 Total +1. 33 -1. 65 -QCAL: - Change Qcal veto definition - Ploss comparison with: KS p 0 p 0 and KS p+p samples. - c 2 cut: change c 2 cut definition - Energy* scale: correct data-MC energy scale calibration using KL gg c. PT NA 31 NA 48/99 NA 48/03 O(p 4) O(p 6) 3 far from NA 48 result, but confirming Ch. PT prediction KLOE - The NA 48 measurement implied the existence of a sizeable O(p 6) counterterm in Ch. PT. Our number makes this contribution practically negligible EPS 2007, Manchester M. Martini 19/07/2007

KS gg: kinematic fit To further reduce background a Kinematic fit is used with the following constraints: Before QCAL cut - PKS(KL-crash) = PKS(2 g) - Mgg = MKL - T = Rc for the two photons NDOF = 7 • • DATA -- MC signal + background MC Signal EPS 2007, Manchester M. Martini 19/07/2007

KS gg: kinematic fit To further reduce background a Kinematic fit is used with the following constraints: After QCAL cut - PKS(KL-crash) = PKS(2 g) - Mgg = MKL - T = Rc for the two photons NDOF = 7 • • DATA -- MC signal + background MC Signal EPS 2007, Manchester M. Martini 19/07/2007