Preliminary results on Ks 3p0 search...... M. Martini and S. Miscetti Analysis outlook: Comparison Data-MC, starting from DST dk0,mk0 Data 450 pb-1 MC all avaible statistics (NeuKaon) Kinematic fit procedure Definition of pseudo-c2 to improve S/B c22p c23p Study of the background shape Data vs MC Adjusted simulation to reproduce observed bkg rate Track veto, E gamma cuts Preliminary determination of upper limit for BR M.Martini 1
Standard Ks tag using KL interaction on EMC ( Kcrash ) Events filter Standard Ks tag using KL interaction on EMC ( Kcrash ) 30% (24 x 106 with 4 cluster norm.) Tuning of acceptance cuts (6 pb-1 Data, MC) looking for highest e while retaining 0 candidates in events with 6 neutral clusters: |T-R/c| = 3.5 st ; Ecut = 7 MeV ; = 22,5° 58% (starting sample 39728 events) MC BKG 10Kevents MC Ks 3p0 Kinematics fit applied using Ks momentum estimated by Kcrash (and ) by requiring 4-momentum conservation on the Ks side (c2fit). After applying a reasonable cut on the c2fit (< 30) a sizeable quantity of bkg events still remains. In order to improve signal/noise we have constructed two pseudo-c2 to discriminate between Ks2p0; Ks3p0 M.Martini 2
Kcrash + 4 prompt clusters Construction of the c22p c22p is built selecting 4 out of 6 clusters which better satisfy the kinematics of KS into 2 pions decay The kinematics parameters used are: Mass distributions Opening angle between pions in Kaon C.M. Frame Four-momentum conservation KL KS All this is done using the reconstructed cluster parameters before applying the kinematic fit procedure Calibration of the c22p is done using Kcrash + 4 g (i.e. Golden Ks 2p0 sample) Kcrash + 4 prompt clusters •• DATA -- MC M.Martini 3
c23p At the moment, the c23p is based only on the 3 Data MC At the moment, the c23p is based only on the 3 reconstructed pion masses M.Martini
Data MC comparison of c22p vs c23p 2001: a surprise! In the data, a new category of BKG events (not simulated by the “standard” Kcrash MC) appears. This simulation takes into consideration only KL decaying after a cylinder bigger than DCH and smears the KL MC direction with the KL crash resolution observed in data. M.Martini
c22p vs c23p 2002 sample Data MC Ks2p0 MC Ks 3p0 The MC has been adjusted inserting a calibrated quantity of “fake” Kcrash. Requiring in the MC reconstruction the same cuts used in the tag definition ( E , b*) MC calibration performed without cutting on c2fit Kcrash fakes ( 3 %) are dominated by Ksp+p- , KL3p0 events, with pions Interacting in Qcal, beam-pipe MC Ks 3p0 Data MC Ks2p0 M.Martini
Comparison “Data-MC” c22p, no c2fit cut All c23p c23p > 80 c23p < 80 c23p < 200 After MC adjustment Result of normalization Weight factors Kcrash MC 0.417 Kcrash fake 0.979 M.Martini
Comparison “Data-MC” c23p, no c2fit cut All c22p c22p > 40 After MC adjustment 14<c22p < 40 c22p < 14 M.Martini
Definition of the Signal box Up Cup Sbox CSbox Down Cdown M.Martini
Sbox Up Down CSbox Cup Cdown Comparison Data-MC NO CUTS on c2fit The sample without any c2FIT is used to check the reliability of the “adjusted” simulation on reproducing the rate in the signal and control boxes. NO CUTS on c2fit BOX Name Ndata Nmc kkm Sbox 304±17 313.6±16.6 Up 456±21 493.7±21.8 Down 356±19 401.1±13.6 CSbox 5141±72 5241.1±47.9 Cup 10523±103 11005.4±69.7 Cdown 22948±151 22276.0±97.0 M.Martini
Next cuts ... TRKveto We count only tracks coming from IP (numTRK) No cuts applied c22p > 40 All c22p 14<c22p < 40 c22p < 14 We count only tracks coming from IP (numTRK) r(PCA) < 4 cm Z(PCA) < 10 cm Veto events with numTRK>0 M.Martini
E(Ks) = Eks(KineFit)- Eg Study of DECUT After finding the 4 satisfying the Ks2p0 kinematics (by c22p) we evaluate the residual energy on the Ks side: E(Ks) = Eks(KineFit)- Eg --- MCBG no 2fit cut MCBG 2fit < 30 MCBG 2fit + TRKveto MCSIG KCRASH STANDARD MC KCRASH FAKES End of analysis: 2fit < 30 TRKveto Signal Box M.Martini
Next cuts ... DECUT (Rejecting DE < 9 MeV) TRKveto+DECUT+c2<30 TRKveto+DECUT TRKveto All c22p c22p > 40 14<c22p < 40 c22p < 14 M.Martini
Sbox Up Down CSbox Cup Cdown Comparison Data-MC Comparison Data-MC after applying c2fit<30, TRKveto and DECUT BOX Name Ndata Nmc kkm Sbox 5±2 3.1±1.3 Up 0±0 0.0±0.0 Down 16±4 22.3±3.1 CSbox 494±22 515.8±14.7 Cup 29±5 19.3±2.9 Cdown 5422±74 5485.5±47.9 M.Martini
VERY PRELIMINARY Calculation of the Upper Limit... Efficiencies (after Kcrash tag) Number of events with 1 Kcrash tag and 4 prompt g (Norm. Sample) 0.31 (PDG) Upper limit determined using a classical frequentist method and using for BKG the average found –1 sigma (conservative approach): Mean Expected Background= BKG used = 1.7 Events observed = 5 With a1 weak amplitude for K0 to decay into I=1 final states To be compared with NA48: No assumption on CPT 1.4*10-6 Assuming CPT conservation 3.0*10-7 VERY PRELIMINARY M.Martini
What needs to be finish for paper… Maximization of upper limit on MC For example with DECUT=100 MeV e3p=19% , UL could improve by a factor of 1.5 check of “any” dependence on and run conditions Check efficiency as a function of each cuts Complete determination of upper limit, - compare with NA48 Constrain |000| 90% CL M.Martini
Future plans for next running period Use the MC/Data sample as a benchmark to maximize upper limit and definition of Sbox, tag used (for istance, hardening energy cuts DECUT 100 MeV, Ekcrash > 200 MeV efficiency lowers from 0.3x0.26 to 0.24x0.19 but the Ncandidates greatly decrease). Extend the Ks tag to KLpmn,pen,ppp, decays looking at charge vertecies on DCH ( Tag efficiency increases of 1.5) If the accidental background remains at the level of 2002 running we can still hope to see 0 candidates in 2 fb-1 M.Martini
Future plans for next running period Assuming 0 candidates from 2 fb-1 Ncand (2.3/7.6) x Eff (26/19) x Norm 1/6 = 0.069 BR < 0.069 * 2.2x10-7 = 1.5x10-8 90% CL M.Martini
Additional information M.Martini
Comparison “Data-MC” c2fit, no c2fit cut Result of normalization Weight factors Kcrash MC 0.417 Kcrash fake 0.979 Comparison chi2 fit M.Martini
A NEW SIMULATION OF Klcrash (AcciK) To understand these events whenever no Kcrash is found by the standard Kcrash MC we add the possibility to find a Kcrash applying the standard data cuts (E and b*) Running this new Kcrash simulation on 2002 MC we find other 1320 entries with respect of to the 9657 events already simulated in the 6 prompt clusters sample. There are three different sources of these new BKG events: AcciKcrash K crash by accidental (1) T0stolen Golden cluster by accidentals (2) Klpipe K crash by KL daughters inside Rt = 25 cm (3) Ngam=6 Ngam=5 Ngam=4 Total events 1320 13677 5314 1 17 4 2 130 255 7 3 139 217 28 2*3 77 64 1*2 --- 1*2*3 1+2+3 207 420 35 Type of fakes M.Martini
Rate normalization Comparison “Data-MC” c2 2p, no c2 cut All c23p c23p > 80 Normalization with 6 g rate reasonable in the overall plot but missing to reproduce the observed rate in these regions c23p < 80 c23p < 200 Rate normalization M.Martini
Check reliability of the “adjusted” simulation when a c2<30 cut is Comparison Data-MC Check reliability of the “adjusted” simulation when a c2<30 cut is applied. BOX Name Ndata Nmc kkm Sbox 18±4 18.3±3.9 Up 1±1 0.0±0.0 Down 54±7 51.9±4.9 CSbox 820±29 871.2±19.2 Cup 32±6 20.6±3.0 Cdown 13278±115 12294.7±71.8 Table chi2 cut M.Martini
Normalization plot Adjusted simulation Normalization kk1 c23p<80 Data MC KcraMC AcciK Normalization kk1 c23p<80 Normalization kk2 c23p<200 MC KcraMC AcciK Data Normalization plot M.Martini
Comparison “Data-MC” c2 fit, c2<30 Chi2 fit chi2<30 M.Martini
Comparison “Data-MC” c2 3p, c2<30 All c22p c22p > 40 Normalization with kkm values 14<c22p < 40 c22p < 14 Chi3 proj chi2<30 M.Martini
Definition of c2 2p c2 2p is the c2 that we build searching the best combination of 4 out of 6 clusters which represents a KS 2p0. The best combination is the one minimizing: M.Martini
Before splash c2 3p At the moment, the c2 3p is based only on the 3 reconstructed pion masses Data MC Comparison Data-Mc before splash filter Splash filter consists of: - Ngam = 6 Emean < 40 MeV Mmean < 40 MeV - Ngam = 4 Emean < 50 MeV Mmean < 50 MeV M.Martini
After splash c2 3p Comparison Data-Mc after splash filter M.Martini
Adjusted normalization ID1 = Data ; ID2 = MC(Kcrash) ; ID3 = MC(AcciK) ID1 = a1ID2 + a2ID3 Where: Ndata = Number of entries of the c2 fit plot for data Nmc = Number of entries of the c2 fit plot for mc Normalization from two different plots: kk1 Coefficients calculated from c2 2p with c2 3p less then 80 kk2 Coefficients calculated from c2 2p with c2 3p less then 200 kkm The average value between kk1 and kk2 M.Martini
Comparison “Data-MC” c2 3p, no c2 cut All c22p c22p > 40 Normalization with 6 g rate 14<c22p < 40 c22p < 14 Chi3 rate norm M.Martini
Comparison “Data-MC” c2 3p, no c2 cut All c22p c22p > 40 Normalization with kk1 values 14<c22p < 40 c22p < 14 Chi3 kk1 norm M.Martini
Comparison “Data-MC” c2 3p, no c2 cut All c22p c22p > 40 Normalization with kk2 values 14<c22p < 40 c22p < 14 Chi3 kk2 norm M.Martini
Data events Candidates... M.Martini At the end of analysis we have: 5 candidates from “Data” @ 450 pb-1 3 expected from “MC” @ 150 pb-1 Nrun: 24051 Nev: 7110735 NTracks: 0 ---------------------------------------------- Reconstructed pions masses: M1 = 114.5806 MeV M2 = 125.6421 MeV M3 = 108.1511 MeV ---------------------------------------------- Chi2 fit: 17.32738 Chi2 pair: 17.72474 Chi3 pair: 73.09442 Number of kcrash: 1 Ekcra: 210.6075 MeV Beta Kcra: .2319278 Number of clusters: 9 ---------------------------------------------- Clusters parameters: Cluster Energy (MeV) Nsigma Angle 1 122.7134 .3378006 1.265235 2 112.8406 1.413125 .7238826 3 87.42661 1.504362 2.373147 4 39.42066 1.709174 2.366150 5 118.7905 1.757564 2.336169 6 35.05759 1.826631 2.369749 ----------------------------------------------- Data events M.Martini
Preliminary calculation of the Upper Limit... The efficiencies are: The number of events with 4 g from data are: To calculate the BR we can use: Only assuming the Kcrash is the same in 2p and 3p BR formula M.Martini
Systematics on background evaluation: Dependence on c2fit The events in the Up zone have very high c2fit and the region becomes quickly empty We fit the rate Data/MC obtained with different values of c2fit cut: c2fit < 500 c2fit < 400 c2fit < 300 c2fit < 200 c2fit < 100 Data/mc vs chi2 M.Martini
Systematics on data/bkg for different cuts Distribution of the Data/MC ratio for all regions, (including also signal BOX) for all used c2fit cuts No cuts TRKveto CONSERVATIVELY assuming a systematic error of ±20% on the absolute amount of expected background TRKveto+DECUT Data/mc syst M.Martini
Prevision on BR M.Martini x10-9 Black dots= Current statistics Red dots=2fb-1 MC Using F.C.: Actual BKG=1.5-2.5 2fb-1 BKG=6-10 18*10-9 observed 0.75*10-9 M.Martini