P Massarotti Charged kaon lifetime preliminary results Paolo

P. Massarotti Charged kaon lifetime: preliminary results Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Summary: ØLength strategy ØLast check ØTime strategy ØEfficiency evaluation ØResolution function Øt± preliminary measurement ØConclusions Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Length measurement … pre cis em eas ure me nt so nl y K± ALL Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Strategy Ø Signal selection • Self triggering muon tag • K track on the signal side • Decay vertex K Kmn tag K Li vtx m T Øvertex efficiency Li = step length Øresolution functions needed wrt the proper time of the Kaon Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Proper time distribution The main distortion of the slope comes from the badly reconstructed Kaon vertexes We evaluate on Monte. Carlo this effect Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Background analysis Five families: Ø “golden”: good vertex (~ 69. 1%) Ø kaon hits associated to daughter track (~ 20. 6%) Ø daughter hits associated to kaon track (~ 5. 2%) Ø early pion decay (~ 1. 4%) Ø K± broken track(~3. 7%) Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Families proper time resolutions All (but one) the dists are centered within 200 ps good track K± broken track kaon hits associated to daughter track Daughter P* with kaon mass hypothesis Mean: -7. 6 ns daughter hits associated to kaon track early pion decay Kaon broken tracks Cut at 100 Me. V/c Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Systematic from P* cut Daughter P* All (but one) the dists are good track with kaon mass hypothesis centered within 200 ps Kaon broken tracks K± broken track kaon hits associated to daughter track Cut at 70 Me. V Mean: Cut at 100 Me. V -7. 6 ns daughter hits associated to kaon track early pion decay Cut at 130 Me. V/c Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Systematic checks Source of systematic uncertainties Range stability Bin stability Efficiency correction Beam Pipe wall Drift Chamber wall Self trg tag efficiency Cut on P*K Systematic uncertainties (ps) ± 60 ± 20 The systematic uncertainty ± 10 as a function ± 10 of q in the DC ± 15 is negligible -20 ± 15 Total Systematic uncertainty of the order of 70 ps Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Reducing range stability syst. To reduce the systematic error given by the stability of the fit as a function of the range used we have to study the resolution functions. These ones are defined as the difference between T*reco and T*true But where c is a global correction factor. We will evaluate this correction factor bin by bin in order to reduce the systematic uncertainty Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Decay length result Weighted mean between t+ and t- t = (12. 367± 0. 044 stat± 0. 070 sys) ns 0. 024 KLOE Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Time measurement K+ X 0 X gg Use neutral vertex only in order to obtain a completely independent measurement Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Time measurement Self triggering muon tag Considering only kaon decays with a p 0 K X p 0 X gg we look for the neutral vertex asking Ø clusters on time: (t - r/c)g 1 = (t – r/c)g 2 Ø p 0 invariant mass Ø agreement between kaon flight time and clusters time Paolo Massarotti E , x , t p. K tm KLOE General Meeting 2006 Kmn tag x. K ± l. K t 0 p K 0 Eg, tg, xg 20 December 2006

Time measurement: proper time distribution C I T S I T E W E V A TO A Using the time technique, T S taking. E into account the different G reconstruction efficiency, we have R A half of the statistic given L N by the length technique E H Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Time measurement: add kaon decays before DC inner wall It’s possible to reconstruct kaon neutral vertices before DC inner wall in the same way in which the neutral vertex can be reconstructed in the drift chamber: we look for neutral clusters in the calorimeter then, taking into account the energy tm loss in the material, using tag information we reconstruct the kaon helix in the region before DC inner wall and we look for a neutral vertex. Paolo Massarotti KLOE General Meeting 2006 p. K Kmn tag t 0 x. K p. K 0 Eg, tg, xg 20 December 2006

Time measurement: proper time distribution The region in red is given by kaon decays before the DC inner wall Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Time measurement: resolution Resolution function for charged kaon decays before DC inner wall kaon decays inside DC Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Time measurement: resolution Before DC inner wall Inside DC Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation Our goal is the evaluation of the neutral vertex reconstruction efficiency before the DC inner wall. In the DC the efficiency is defined as the ratio of the events in which both the charged and the neutral vertices are reconstructed over the events in which the charged vertex is reconstructed In the region before the DC inner wall we can construct a virtual charged vertex given by the secondary track and the kaon helix Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation Kaon helix secondary track F meson Virtual charged vertex In the region before the DC inner wall the efficiency to reconstruct a neutral vertex is defined as the ratio of the events in which both the virtual charged and the neutral vertices are reconstructed and the events in which the virtual charged vertex is reconstructed Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Virtual vertex technique: two cuts applied After two cuts applied, one on the secondary P* in kaon mass hypotesis (100 Me. V) and one on the number of hits associated to the secondary track (25), we obtain very good time of flight resolutions which are comparable with the resolution given by the usual vertex technique Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Virtual vertex tecnique: two cuts After two cuts applied, one on the secondary P* in kaon mass hypotesis (100 Me. V) and one on the number of hits associated to the secondary track (25), we obtain very good time of flight resolutions which are comparable with the resolution given by the usual vertex technique Virtual vertex tecnique Virtual and normal vertex tecnique Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: MC true and MC reco comparison Nvc = usual charged vertex or virtual charged vertex Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: the normalization question Nvcp 0 = usual charged vertex or virtual charged vertex both with a p 0 Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: ratio (slope) P 1 fit Slope (7. 7± 1. 4)*10 -4 c 2/ndf = 43/44 Paolo Massarotti Nvcp 0 = usual charged vertex or virtual charged vertex both with a p 0 KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: ratio (const) P 0 fit (99. 85±. 12)*10 -2 Con c 2 /ndf= 72/45 Con (99. 81±. 12)*10 -2 c 2/ndf = 46/30 Nvcp 0 = usual charged vertex or virtual charged vertex both with a p 0 Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: In order to take into account also the correlation between the tag requirement and the K± X 0 signal events, which are characterized by low momentum charged daughter, we use as normalization sample for the true efficiency the number of tag events. On the other hand the charged vertex sample already takes into account this correlation. So we have to compare and Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: MC true and MC reco comparison Nvc = usual charged vertex or virtual charged vertex Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: ratio Slope(-2. 3± 2. 6)*10 (-1. 3± 3. 1)*10 -4 -4 cc 22/ndf == 83/44 72/29 Nvc = usual charged vertex or virtual charged vertex Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Efficiency evaluation: data and MC comparison MC reco Data Nvc = usual charged vertex or virtual charged vertex Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Fit procedure We make the fit in the region between 8 and 35 ns. To fit the proper time distribution we construct an histogram , expected histo, between 5 and 45 ns, in a region larger than the actual fit region to take into account border effects. The number of entries in each bin is given by the integral of the exponential decay function, which depends on one parameter only, the lifetime, convoluted with the efficiency curve. Then a smearing matrix accounts for the effects of the resolution. We also take into account the tiny correction to be applied to the efficiency given by the ratio of the Monte. Carlo data-like and Monte. Carlo kine efficiencies. nbins Nexpj = S Csmearij × eicorr × Nitheo i=1 Paolo Massarotti KLOE General Meeting 2006 20 December 2006

MC t+ measurement: ° MC reco • Fit Best value between 8 and 35 ns (more than 2 lifetimes…) t +MC = (12. 411 ± 0. 048) ns Paolo Massarotti c 2/ndf = 22. 6/26 KLOE General Meeting 2006 T*(ns) Pc 2 =65. 7% 20 December 2006

MC t+ measurement: ° MC reco • Fit region between 8 and 35 ns t +MC = (12. 411± 0. 048) ns Paolo Massarotti T*(ns) The fit reproduces the dist. very well also outside the fit region ! KLOE General Meeting 2006 20 December 2006

MC t+ residual evaluation T*(ns) p 0 = 100. 8± 109. 4 c 2 /ndf = 21. 75/ 25 Paolo Massarotti p 1 = -4. 6± 4. 0 Pc 2 = 65% KLOE General Meeting 2006 20 December 2006

MC t+ normalized residual evaluation T*(ns) p 0 = 0. 40± 0. 58 c 2 /ndf = 21. 75/ 25 Paolo Massarotti p 1 = -. 02±. 03 Pc 2 = 64% KLOE General Meeting 2006 20 December 2006

Data t+ measurement On Data the fit doesn’t converge in the whole region because there are problems in the region between 8 and 12. . . We make the fit in the region between 14 and 38 ns. To fit the proper time distribution we construct the expected histo between 12 and 45 ns in order to take into account border effects. Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Data t+ measurement ° MC reco • Fit ° Data • Fit T*(ns) Fit between 14 and ns 38 (about 2 lifetimes…) t +Data = (12. 378 ± 0. 066) ns Paolo Massarotti c 2 /ndf = 38. 2/22 Pc 2 = 2. 4% KLOE General Meeting 2006 20 December 2006

Data t+ measurement ° Data • Fit between 14 and ns 38 (about than 2 lifetime…) t +Data = (12. 378 ± 0. 066) ns Paolo Massarotti T*(ns) c 2 /ndf = 38. 2/22 Pc 2 = 2. 4% KLOE General Meeting 2006 20 December 2006

Data t+ residual evaluation p 0 = 197± 141 c 2 /ndf = 38. 4/ 22 Paolo Massarotti p 1 = -7. 2± 4. 6 Pc 2 = 2. 8% KLOE General Meeting 2006 T*(ns) 20 December 2006

+ Data t normalized residual evaluation T*(ns) p 0 =. 80±. 83 p 1 = -. 031±. 031 c 2 /ndf = 36. 4/ 21 Pc 2 = 2. 1% Paolo Massarotti KLOE General Meeting 2006 20 December 2006

The question of the 8 -12 ns region The value of the lifetime obtained on Data is the same value of the simulation, so we can compare Data and Monte. Carlo proper time distribution, corrected for their efficiency curves in order to understand the problem that we found on Data in the region between 8 and 12 ns Paolo Massarotti KLOE General Meeting 2006 20 December 2006

The question of the 8 -12 ns region Looking at the ratio of the Data and Monte. Carlo distributions we see two slope, the first one not consistent with zero… P 11 = (-7. 6± 1. 2)10 -3 P 12 = (6. 4± 8. 4)10 -4 c 2 /ndf = 40/32 Paolo Massarotti 1 KLOE General Meeting 2006 2 20 December 2006

The question of the 8 -12 ns region Looking at the ratio of the Data and Monte. Carlo distribution we see two slope, the first one not consistent with zero… The “crucial” point is at 12 ns. We think that this discrepancy is given by a not good tuning of the procedure of the virtual charged vertex on Data (further studies of cuts applied) Anyway the technique is very promising. Paolo Massarotti KLOE General Meeting 2006 20 December 2006

To Do LENGTH: Resolution Check Memo Update To Bless… TIME Tuning of the virtual vertex procudure K- measurement (CPU time only) Systematic checks Paolo Massarotti KLOE General Meeting 2006 20 December 2006

Conclusions We have obtained, for the Length measurement t. K±=(12. 367± 0. 044 stat± 0. 070 sys)ns in agreement with PDG 2005 fit. We have obtained, for the Time measurement t. K+=(12. 378± 0. 066 stat)ns in agreement with PDG 2005 fit. Paolo Massarotti KLOE General Meeting 2006 20 December 2006

p 0 bias Short T* Bad reconstruction No tag Kmn tag tm p. K t 0 x. K p. K Long T* Tag p. K 0 tm Eg, tg, xg Paolo Massarotti KLOE General Meeting 2006 Kmn tag t 0 x. K 0 Eg, tg, xg 20 December 2006