YongPyong-High1 015 Joint Winter Conference on Particle Physics, String and Cosmology Search for + l + X 0 with hadronic tagging method at elle Chanseok Park (Yonsei Univ.) pcs437@yonsei.ac.kr 015-01-30 High1 015-01-30 cspark Yonsei, YHEP 1
Contents elle Experiment Motivation X 0 candidate Analysis procedure Hadronic tagging method Preliminary result Summary High1 015-01-30 cspark Yonsei, YHEP
elle Experiment High1 015-01-30 cspark Yonsei, YHEP 3
elle Experiment High1 015-01-30 cspark Yonsei, YHEP 4
Motivation At elle, there are searches that have invisible particle in the final states. eg) + l ν, 0 ν ν, K ν ν, D(*) τ ν Neutrino is not detected at elle detector, from this we have interesting assumption What if massive particle can substitute neutrino? From now, we call it X 0 Possible mass range that X 0 can have 0 m(x 0 ) m( + ) m(l + ) High1 015-01-30 cspark Yonsei, YHEP 5
Why + l + X 0? + D 0 l + X 0 * We don t consider 3body decay ( moment is not clear ) + τ + X 0 * additional neutrino produced Hard to search We consider + l + X 0 ( l = e, μ ) For m(x 0 ) : 0.1 1.8 GeV/c High1 015-01-30 cspark Yonsei, YHEP 6
High1 015-01-30 cspark Yonsei, YHEP 7 X 0 candidate : LSP (RPV) ~ ~ ~ 13 0 1 0 6 1 1 1 1 ) ( 8 ~ X l b u l b u l i i M M M M M M m m p m f g l i R L i When we assuming r-parity violation, one lightest neutralino can be produced from meson decay via slepton or squark. We can give bounds for unknown parameters PRD 65, 015001
X 0 candidate : Large Extra Dimension PL 489, 367-376 * RH-neutrino in large extra dimension might candidate of X 0. * When there are δ additional dimensions, decay width via H or W. (Γ W >> Γ H ) * Decay width is proportional to R δ. M Pl : 4D Planck scale M * : (4+δ)D fundamental Planck scale R : size of additional dimension High1 015-01-30 cspark Yonsei, YHEP 8
+ l + X 0 Sample for analysis mode Mass of X Amount + e + X 0.1, 0., 1.8 GeV,000,000 events for each mass of X + μ + X 0.1, 0., 1.8 GeV,000,000 events for each mass of X ackground MC Separately generated! Signal MC We have 18 kinds of X for different mass Mode Process Amount Generic MC, qq 5 streams Rare b s, d 50 streams Ulnu X u lν 0 streams eνγ + eνγ 1000 streams μνγ + μνγ 1000 streams π + K 0 + π + K 0 500 streams π 0 eν + π 0 eν 300 streams π 0 μν + π 0 μν 300 streams High1 015-01-30 cspark Yonsei, YHEP 9
Hadronic tagging method Signal lepton Tagged Signal D K Y(4S) Ex) D + (K - π + π + ) π - π - Invisible >96% of Y(4S) with nothing else produced one -meson is completely reconstructed from known b c decays without ν * 615 + channels are used for reconstruction. * Low efficiency, high purity Good way to reconstruct modes with invisible particle High1 015-01-30 cspark Yonsei, YHEP 10
+ l + X 0 - event selection Particle Identity L e > 0.9 L μ > 0.9 Track quality Dz < cm Dr < 0.5 cm Continuum suppression cosθ thrust < 0.9 for + e + X cosθ thrust < 0.8 for + μ + X E ECL Quality of tagged- meson ΔE < 0.05 GeV M bc > 5.7 GeV/c O N > e -6 0.5 0 p l sideband E ECL Sideband linded Region 1.8.3 3.0 p l (GeV/c) E ECL : Remaining energy of ECL calorimeter (tagged- & signal lepton) p l : signal lepton s momentum in the signal rest frame High1 015-01-30 cspark Yonsei, YHEP 11
+ l + X 0 PDF modeling Fitting Signal Fitting ackground p l peak changes by mass of X p l cut should be optimized for each mass of X Fitting Peaking ackground High1 015-01-30 cspark Yonsei, YHEP 1
+ l + X 0 obtain U.L. of.f.. F. N obs sig G est N( ) *.F. is obtained by Feldman-Cousins method * Signal region is optimized. N obs : Number of Data in the signal region (counting) G est : expected background in the signal region, decided by background PDF, data distribution in the p l sideband ε sig : decided by signal PDF. High1 015-01-30 cspark Yonsei, YHEP 13
+ l + X 0 - preliminary result e mode μ mode Sideband Signal region + e + X M(X) : 1.8 GeV Sideband Signal region + μ + X M(X) : 1.8 GeV High1 015-01-30 cspark Yonsei, YHEP 14
+ l + X 0 E ECL sideband calibration + e + X + μ + X There are some disagreement between Data and MC, about p l >. GeV/c for E ECL sideband region(0.5 < E ECL < 1.0 GeV). Get Calibration Factor!! High1 015-01-30 cspark Yonsei, YHEP 15
High1 015-01-30 cspark Yonsei, YHEP 16 + l + X 0 ounds for parameters ~ ~ ~ 13 0 1 6 1 1 1 1 ) ( 8 ~ R L i b u l b u l i i M M M m m p m f g l ) ( ).(. ) ( 8 6 1 1 1 1 0 ~ ~ ~ 13 0 X l l i b u b u l i m m m p m f g X l U L m m M M M i i R L i This is parameter we give bounds!
Summary * We search for + l + + X 0, where X 0 have 0.1 ~ 1.8 GeV mass range * Hadronic tagging method enables effective background suppression * + l + X 0 has preliminary results, and analysis enters the final steps * e + e - -factory experiments has an advantage for this study. High1 015-01-30 cspark Yonsei, YHEP 17
ACKUP High1 015-01-30 cspark Yonsei, YHEP 18
+ l + X 0 - skim procedure S KIM PATH Hadronic Tagging LX_SKIM ANALYSIS_CODE L X _ S K I M 1 charged particle not used in Full_recon call it c (Charge of c) x (Charge of tagged ) = -1 Momentum of c(la frame) >1.0 GeV High1 015-01-30 cspark Yonsei, YHEP 19
e mode dz, dr, deltae -log(nboutput) distributions High1 015-01-30 cspark Yonsei, YHEP 0
e mode Mbc, Eecl, cos(thrust), pl distributions with basic cut High1 015-01-30 cspark Yonsei, YHEP 1
μ mode dz, dr, deltae -log(nboutput) distributions High1 015-01-30 cspark Yonsei, YHEP
μ mode Mbc, Eecl, cos(thrust), pl distributions with basic cut High1 015-01-30 cspark Yonsei, YHEP 3
Fitting to obtain PDFs (MC samples) - 1D ML fit for p l was done (1.8~3.0 GeV/c) - Cuts for all remaining variables are same - Using simple function as much as possible Some modes in Ulnu are scaled High1 015-01-30 cspark Yonsei, YHEP 4
e-mode signal region(e ecl <0.5 GeV) All Other cuts are applied to signal region : Gaussian Ulnu :.G. Rare : Exp eνγ :.G. π0eν : G +.G. High1 015-01-30 cspark Yonsei, YHEP 5
μ-mode signal region(e ecl <0.5 GeV) All Other cuts are applied to signal region : Gaussian Ulnu : Gaussian μνγ :.G. Rare : Exp + Argus πk0 : G + G π0μν : G +.G. High1 015-01-30 cspark Yonsei, YHEP 6
Signal (left : e mode, right : μ mode) For E ecl <0.5 & 1.8< p l <3.0 Signal is fitted with Gauss+Gauss+.G M X : 0.1 GeV M X : 1.0 GeV M X : 1.8 GeV High1 015-01-30 cspark Yonsei, YHEP 7
Expectation of ranching Fraction Statistical uncertainty PDG uncertainty PDF uncertainty Data in sideband MC(signal)/MC(side) Estimated # of G Signal efficiency and uncertainty Estimate.F.(or upper limit) for observed events when open box. High1 015-01-30 cspark Yonsei, YHEP 8
Expected Upper Limit. F. U. L.( Yield ) N( ) sig 1. Relative uncertainty of ε sig. Estimated G and uncertainty 3. # of observed events POLE U.L. Uncertainty from PDG(F), PDF, systematic, etc High1 015-01-30 cspark Yonsei, YHEP 9
Optimization Study using the criterion of est Upper Limit Mean of U.L. 6 n0 6 Yield n0 U. L. ( G est Poisson( G ; n) Poisson( G est est ; n;1000) N( ) ; n;1000) sig - n : # of observed events in signal region. - Yield_{U.L.} : U.L. of Yields using POLE program - Poisson : # of values of 1,000 events have Poisson dist High1 015-01-30 cspark Yonsei, YHEP 30
+ l + X 0 - Optimization G : Fit p l sideband extrapolate PDF G est Data side S( MC) S( MC) sig side Feldman-Cousins method 1. Relative uncertainty of ε sig. Estimated G and uncertainty 3. # of observed events POLE U.L. Uncertainty from PDG(F), PDF, systematic, etc High1 015-01-30 cspark Yonsei, YHEP 31
+ l + X 0 - Optimization Mean of upper limit of branching fraction based on MC for each p l criteria + e + X M(X) : 1.8 GeV/c + μ + X M(X) : 1.8 GeV/c High1 015-01-30 cspark Yonsei, YHEP 3
Summary Table (e-mode) M(X) pl cut G_est Efficiency( ) Observed event U.L. (10^{-6}) 0.1 (GeV).5 < pl <.70 0.44±0.01 1.13±0.14 0.41 0..5 < pl <.70 0.44±0.01 1.1±0.14 0.43 0.3.55 < pl <.68 0.8±0.134 1.08±0.13 0.70 0.4.55 < pl <.68 0.8±0.134 1.06±0.13 0.75 0.5.5 < pl <.70 0.44±0.01 1.08±0.13 0.5 0.6.5 < pl <.70 0.44±0.01 1.07±0.13 0.54 0.7.5 < pl <.70 0.44±0.01 1.11±0.14 0.45 0.8.51 < pl <.6 0.436±0.190 1.07±0.13 0.54 0.9.51 < pl <.6 0.436±0.190 1.01±0.13 0.69 1.0.51 < pl <.6 0.436±0.190 0.97±0.1 0.81 1.1.47 < pl <.57 0.615±0.51 0.99±0.1 0.54 1..45 < pl <.53 0.636±0.57 0.97±0.1 0.57 1.3.43 < pl <.51 0.738±0.303 0.98±0.1 0.45 1.4.41 < pl <.51 0.985±0.410 1.0±0.1 0.15 1.5.39 < pl <.46 0.843±0.374 0.95±0.1 1 4.80 1.6.37 < pl <.43 0.816±0.380 0.94±0.11 1 4.88 1.7.34 < pl <.39 0.805±0.389 0.89±0.11 1 5.17 1.8 High1 015-01-30.31 < pl <.36 0.941±0.455 cspark 0.90±0.11 Yonsei, YHEP 7.10 33
Summary Table (μ-mode) M(X) pl cut G_est Efficiency( ) Observed event U.L. (10^{-6}) 0.1 (GeV).58 < pl <.68 0.439±0.111 1.18±0.14 1 4.6 0..58 < pl <.68 0.439±0.111 1.19±0.15 1 4.3 0.3.58 < pl <.68 0.439±0.111 1.18±0.14 1 4.6 0.4.58 < pl <.68 0.439±0.111 1.19±0.15 1 4.34 0.5.58 < pl <.68 0.439±0.111 1.15±0.14 1 4.37 0.6.58 < pl <.68 0.439±0.111 1.13±0.14 1 4.45 0.7.56 < pl <.63 0.46±0.116 1.13±0.14 0.35 0.8.54 < pl <.61 0.485±0.140 1.14±0.14 1 4.37 0.9.5 < pl <.60 0.605±0.187 1.14±0.14 1 4.3 1.0.49 < pl <.58 0.838±0.70 1.13±0.14 1 4.04 1.1.49 < pl <.58 0.838±0.70 1.18±0.14 1 3.87 1..48 < pl <.53 0.594±0.194 1.06±0.13 0.37 1.3.45 < pl <.50 0.731±0.33 1.03±0.13 0.8 1.4.4 < pl <.48 0.994±0.307 1.10±0.13 5.75 1.5.40 < pl <.47 1.33±0.371 1.11±0.14 5 10.64 1.6.37 < pl <.4 1.05±0.87 1.05±0.13 4 9.66 1.7.34 < pl <.39 1.164±0.308 1.05±0.13 1 3.93 1.8 High1 015-01-30.31 < pl <.37 1.574±0.40 cspark 1.1±0.14 Yonsei, YHEP 1 3.7 34
+ l + X 0 E ECL sideband calibration 1.8 < p l <.65 1.8 < p l <.45 1.8 < p l <.3 E ECL cut : 0.5 < E ECL <.0 GeV (ecause we want more statistics) Data/MC ratio is fitted to linear function Ratio function : R(p l ) = p 0 + p 1 ( p l - 1.8 ) when p 0 and p 1 is parameter To fit well, we apply error to bins where no events (but MC exist) High1 015-01-30 cspark Yonsei, YHEP 35
+ l + X 0 E ECL sideband calibration Originally we use Data & MC ratio in p l sideband region to scale expectation of G So we use this ratio fitting function to scale G expectation. Calibration factor R* is used for scaling. We use ratio fitting function when fitting range 1.8 < p l <.65 GeV/c Old : G est Data side S( MC) S( MC) sig side New : G est R* Data side S( MC) S( MC) sig side High1 015-01-30 cspark Yonsei, YHEP 36