PowerPoint プレゼンテーション

Size: px

Start display at page:

Download "PowerPoint プレゼンテーション"

きょうすけだいほうじ
6 years ago
Views:

1 The result of the MEG experiment with the full dataset (+ MEG II status) Daisuke Kaneko, on behalf of the MEG collaboration

2 Contents 1. µ eγ Decay 2. MEG Instruments 3. Analysis and Result 4. MEG II experiment

3 1. μ eγ Decay

4 Flavor of Particles 4 Quark ~ ~ u c ~t u c t Lepton ~ ~ e µ e µ τ ~ τ ~ CKM Matrix d ~s ~ b d s b ~ ν e ν e ~ ν µ ν µ PMNS Matrix ~ ν τ ν τ

5 μ + e + γ : undiscovered decay 5 Forbidden in Standard Model (Lepton flavor conservation law) It is possible, with neutrino oscillation, probability is < no exist practically μ e γ via ν-oscillation Promising theories beyond SM predict accessible probability see-saw mechanism SUSY-GUT etc. μ e γ via SUSY particle

θ 13 value (already discovered to be ~9 ) L. Calibbi et al. Phys. Rev.

6 10-12 ~10-14 is predicted 6 before MEG MEG Antusch et al., J. HEP 2006(11), 090 (2006) SU(5) + seesaw different colors correspond different θ 13 value (already discovered to be ~9 ) L. Calibbi et al. Phys. Rev. D 74, (2006) SO(10) + seesaw green : PMNS case, red : CKM case tanβ = 10, as function of M 1/2

7 History of μ e γ search Discovery of μ 1947 First search with cosmic-ray μ is not an excited state of e 1950s μ eγ search with accelerator 1970s search with meson factories Concept of lepton flavor Rumor of discovery, but not true Crystal Box MEGA LAMPF

8 Signal & BackGround 8 Signal 52.8 MeV = m μ /2, back-to-back, at the same time. BackGrounds Raditive Muon Decay (RMD) μμ + e + νν μμ νν e γγ ACCidental BG (ACC) R BG R μ2 δe e (δe γ ) 2 δω/4π δt - e + from normal µ+ decay - γ from RMD or annihilation of e + Type Eγ Ee+ Time Angle Signal 52.8 MeV 52.8 MeV Te = Tγ 180 RMD <52.8 MeV <52.8 MeV Te = Tγ 180 ACC <52.8 MeV 52.8 MeV uniform no correlate

9 2. MEG Instruments

10 Location of experiment 10 Kanton Aargau Switzerland CERN

PSI experimental hall 11 target E UCN main ring

28 MeV/c ± 5-7% 150 mstr proton therapy πe5 area

11 PSI experimental hall 11 target E UCN main ring cyclotron target E neutron Wien filter control magnets πe5 spec. at entrance Intensity 10 8 /s Momentum Solid angle 28 MeV/c ± 5-7% 150 mstr proton therapy πe5 area Beam Transport Solenoid Spot size Angular divergence V: 15mm H: 20mm V: 450 mrad H: 120 mrad

BTS & Target 12 Beam Transport Solenoid Degrader no-scale He-cooled Superconducting magnet to conduct µ beam on target hole Φ1cm cross-marker μ + stopping target

12 BTS & Target 12 Beam Transport Solenoid Degrader no-scale He-cooled Superconducting magnet to conduct µ beam on target hole Φ1cm cross-marker μ + stopping target 199 A, 2.4 T (nominal) Requirement Must stop µ +, but must not interrupt e + Put thin film with angle Design 8 cm 20 cm ellipse 20.5 slant angle Stacked PE & PS, 205 µm

13 MEG detector 13 COBRA Magnet Timing counter Drift chamber z y x μ + beam target y x z Liquid xenon detector

14 Liquid Xe γ-ray detector 14 Liquid Xenon? Rare-gas scintillator Fast, Many photon Heavy as a liquid Homogeneity No self-absorption Many applications in high-energy experiments Difficulty in application Handle low-temp liquid (T~165 K) Control pressure (ΔP < 0.01 atm ) Detect Ultra-violet light (λ ~ 175 nm) phase diagram of xenon Hamamatsu R9869 Photo-multiplier

window at γ- ray entrance face Cooled with pulse tube

15 LXe detector design 15 Inside of the detector Characteristics Total 900 l LXe C-shaped cryostat 846 PMTs on 6 face Honey-comb window at γ- ray entrance face Cooled with pulse tube refrigerator 2 kinds of purification systems equipped 200 W pulse tube refrigerator

LXe detector γ-ray calibration 16 Main γ calibrations A. Cockcroft-Walton (CW) accrlerator target of Li 2 B 4 O 7 14.8, 17.6 MeV B.

16 LXe detector γ-ray calibration 16 Main γ calibrations A. Cockcroft-Walton (CW) accrlerator target of Li 2 B 4 O , 17.6 MeV B. Neutron generator Ni(n,γ)Ni reaction 9.0 MeV C. Charge exchange π - + p π 0 + n π 0 γ + γ CW accel. PSI πe5 p ビーム Li target BGO detector LXe detector H 2 ターゲット μ + beam π - beam neutron source + Ni

π 0 calibration 17 2γ from the reaction π 0 γ + γ By selecting back-to-back γ pair, concentrated energy γ can be selected. Most important calibration, since 55 MeV is near signal.

17 π 0 calibration 17 2γ from the reaction π 0 γ + γ By selecting back-to-back γ pair, concentrated energy γ can be selected. Most important calibration, since 55 MeV is near signal. BGO detector is small and movable, to scan all acceptance of LXe. Energy [MeV] MeV 55MeV opening angle [ ] LH2 target NEW LXe Detector γ γ BGO π - beam COBRA Timing counter

18 γ-ray resolutions 18 42% 58% σ up 2.3% σ up 1.6% Fit 55MeV peak with response function considering Correlation of 2γ angle and energy Difference of noise condition Detector acceptance is divided into small parts and fit each. When γ-ray convert at shallow part of the detector, energy resolution is worse Position resolution is evaluated with lead collimator. to be 5 mm σ in u, v direction and 6 mm σ in w direction.

19 COBRA magnet 19 Uniform B-field e + emitted in θ~90 θ vs radius of track reduce pile up Gradient B-field low momentum e + s are isolated Characteristics Combination of SC magnets with different bore size Thin that γ-ray to transmit Cooled by GM refrigerator Compensation magnet which reduce field at LXe detector

20 Drift chamber e + tracker 20 Interaction of e + and matter: Multiple scattering Worsens angular resolution Pair annihilation Generate γ-ray background High-rate tolerance: High rate μ + s in beam eventually decay into e + s. Low mass tracker 16 modularized detector in φ direction Detector locate only at large R drift cell

e + track reconstruct 21 Hit detection by waveform analysis Reconstruct hit in each cell Ratio of charge on each side Detail z-position by vernier

21 e + track reconstruct 21 Hit detection by waveform analysis Reconstruct hit in each cell Ratio of charge on each side Detail z-position by vernier Connect neighboring hits First fit by circle Main fit of track Kalman Filter algorism is used (Fit error in each event is utilized in final physics analysis )

22 Positron observables 22 Positron energy resolution (σµ e ) is obtained by fitting spectrum of normal μ decay with response function. 1 theoretical spectrum 2 acceptance function 3 resolution function 1 3 Double turn method is adopted to evaluate energy, position and angular resolutions. Independently propagate 1 st and 2 nd turn of genuine track. 2 Resolutions are largely affected by operation condition of DCH, but roughly Ee ~ 300keV, θe φe ~ 10mrad, ye ~ 1.3 mm, ze ~ 3.0mm

e+ timing counter 23 φ counter BC404 scintillator 4 4 79.

23 e+ timing counter 23 φ counter BC404 scintillator cm3 15 bars on each side PMT read-out on both end (Fine mesh type) Assembled φ counter (one of two) z counter BCF-20 scintillation fiber Total 256 pcs. APD readout at one end ( z counter is not used) Roll of timing counter Precise measurement of e + hit time Provide information for trigger

24 Timing reconstruction 24 From the PMT hit time at TIC both end, hit position and time at TIC bar is calculated. Emission timing of positron needs track information (L track ). te tt TIC = tt IN + tt OUT LL bar 2 2vv zz TIC = vv 2 (tt IN tt OUT ) t TIC t in t out Time-walk effect of PMT is corrected in tt IN, tt OUT. Final timing observable is defined as, tt eeee = tt LXe rr γγ rr μμ cc tt TTTTTT LL track cc

25 Timing resolution 25 Timing resolution is evaluated with RMD data, where all the γ-ray detector, positron detector, trigger are the same as the data for μeγ physics data. (Eγ, Ee correlation on teγ need to remove) RMD events peak σ eγ = 122 ± 4 ps Accidental resolutions for each component σtγ ~ 65 ps σte ~ 100 ps

26 Efficiencies 26 γ-ray detection efficiency 62.5±2.3%, for γ from target aiming at detector acceptance Loss: material between (COBRA, cryostat wall, PMT etc) leakage of electro-magnetic shower positron detection efficiency 48% from Monte Carlo simulation. It is not needed in physcis analysis Trigger efficiency After improvement in 2011 trigger rate13hz Live Time ratio 99% Selection efficiency 97%

27 3. Analysis and Result

28 History of MEG 28 design 2000 construction data taking 1999 PSI proposal Approval Old 2007 Detector Complete New 2010 Nucl. Phys. B x (90%CL) 2011 Phys. Rev. Lett. 107, x (90%CL ) 2013 Phys. Rev. Lett. 110, x (90%CL ) 加速器休みメンテナンス等新しく解析に用いたデータ通算データ量 93 TB 通算 DAQ 時間 288 日通算 run 数 (~2000 event/run) 通算静止 μ + 数

29 Event selection 29 Firstly, apply pre-selection in order to obviously accidental events. Then, detailed calibration is done on passed events Final event selection is defined as, 48 < Eγ < 58 MeV 50 < Ee < 56 MeV teγ < 0.7 ns θeγ < 50 mrad φeγ < 75 mrad Region teγ < 1.0 ns is blinded at first. Parameter for physics analysis is determined by outside (sideband) events. Signal events will concentrate around here, if exist. RMD

30 Likelihood analysis 30 Definition of MEG likelihood function L NN sig, NN RMD, NN ACC, tt = ee NN NN obs! CC(NN RRRRRR, NN AAAAAA, tt) NN obs (NN sigss xx ii, tt + NN RMDRR xx ii + NN ACCAA xx ii ) ii=1 extended likelihood PDF constraint term NN = NN ssssss + NN RRMMMM + NN AAAAAA tt: Target parameter NN oooooo : Event number in window xx : (EE γγ, EE ee, tt eeee, θθ eeee, φφ eeee ) SS, RR, AA: (Probability Density Function) CC : Constrain NN RMD NN ACC around expectation in side band Best fit value is defined by such that maximized likelihood function Confidence interval is determined with Feldman-Cousins approach, setting Nsig as the main parameter, and profiling out the others.

31 Probability to find the observable to be the value PDF when Signal, RMD, AccBG happens. 31 Determined from sideband data (partially Monte Carlo simulation) All known correlations between observables, detector position etc. are corrected. event-by-event PDF Shape of function changes, according to Error in reconstruction Position in detector Correlation (ns) θθ eeee tt eeee EE ee EE γγ (mrad) (MeV) (MeV) φφ eeee 0 75 (mrad) 緑 : Signal 赤 : RMD 桃 : ACC 青 : sum Examples in certain events

32 Target Position 32 Target Δz t r e + non-scale When r ~ 10 cm Δz t ~ 1 mm, Δ φ e Drift Chamber Δφe ~ 10 mrad (φe reso. 10mrad) There are 2 methods 1. Optical method next page 2. Software method Utilize correlation of apparent hole position depends on position direction. true hole y true target Δz t Δy assumed target z t e + ΔY ~ tan(φ) ΔP + offset

Target measure 2 33 0 1 5 6 3 4 Measure target with theodolite. Conventionally fit is done with plane, but expanded to paraboloid fit.

33 Target measure Measure target with theodolite. Conventionally fit is done with plane, but expanded to paraboloid fit data can be seen as plane, but 2012, 2013 data has large strain. Horizonal Cross marker Plane fit Paraboloid fit Hole position Vertical 年

Deformation & countermeasure 34 For detail investigation 3D laser scan was performed in 2015.

34 Deformation & countermeasure 34 For detail investigation 3D laser scan was performed in As the result, deformation of complex shape was found, but around the beam-spot, paraboloid is a good approximation. Countermeasure : 1. In trac reconstruction, set start point of e + (=μ stopped point ) to be fitted paraboloid (previously fitted plane) 2. Remaining uncertainties = position local shape are taken into account as nuisance parameters paraboloid Result of 3D scan

35 Target uncertainty 35 Shift center of φeγ PDF for Signal event PDF. Parallel shift Paraboloid of 3D scan Δμμ φφ = Δ μμ φφ eγγ pp, φφ e + ss[δ FARO φφ eγγ xx e, yy ee Δ para φφ eγγ xx e, yy ee ] p : Parallel shift parameter s : Local shape parameter 0 ~ 1 Paraboloid ~ 3D scan p and s are independent for each year, Δ FARO is scaled to match with curvature of paraboloid fit. p is constrained by Gaussian dist. centered at 0 (error 300 (500) um) s is constrained in [0,1] for 2013, narrower region for previous years. Impact on sensitivity: Sensitivity is worsened by13% in sensitivity. This is largest systematics, and the others occupy only 1%.

positron AIF recognition γ estimation LXe detector γ observation (Annihilation in Flight) Δθ AIF Δφ AIF AIF point target Drift CHamber Δθ AIF correct Δφ AIF Δt AIF AIF pair random AIF pair 36 Tag one

36 positron AIF recognition γ estimation LXe detector γ observation (Annihilation in Flight) Δθ AIF Δφ AIF AIF point target Drift CHamber Δθ AIF correct Δφ AIF Δt AIF AIF pair random AIF pair 36 Tag one of the sources of γ-ray, positron AIF A. Recognize interrupted e + track in drift chamber B. Estimate γ-ray momentum from that before AIF C. Calculate angle difference between estimation and observation

AIF reduction and impact 37 Sharp peak in Δθ AIF, Δφ AIF distribution is really tagged AIF events. Cut events near peak. Precise shape of Δt AIF distribution is difficult to obtain.

37 AIF reduction and impact 37 Sharp peak in Δθ AIF, Δφ AIF distribution is really tagged AIF events. Cut events near peak. Precise shape of Δt AIF distribution is difficult to obtain. It is used only for rough cut. Method : 1. Fit 2D distribution Δθ AIF, Δφ AIF with combination of 2D Gaussian function. (2 peak and 1 base component.) 2. Remove events within 0.7σ from either of the peaks, as they are likely to be AIF Accidental BG. Impact : No significant improvement in sensitivity. Insurance for AIF event to come near center of window.

38 Normalization A constant to convert event number 38 and μ + e + γ branching ratio B μμ + e + γγ = Γ μμ+ e + γγ Γ TOTAL = NN sig kk Norm. factor k is considered to be a number of events multiplied with detector acceptance and detection efficiency, There are independent 2 ways, Michel positron way and RMD way. Final value is given by combining two. Both ways do not need e + detection efficiency. For all statistics of MEG data, k = 1.71±

39 Search sensitivtiy 39 median Sensitivity % CL Upper Limit Arrows are limit from time sideband ( -2.0ns, +2.0ns) , Histogram of upper limits of many Toy MCs which do not contain signal. Data set k ( ) Sensiti vity ( ) (90% CL) Previous publication( ) Sensitivity was Understandable, considering the changes in analysis.

Event distribution full data 40 Excess of the signal is not seen. 2009-2013 cosθ < -0.99963 (90% ε signal ) teγ < 0.

40 Event distribution full data 40 Excess of the signal is not seen cosθ < (90% ε signal ) teγ < ns (90% ε signal ) 51 < Eγ < 55.5 MeV (74% ε signal ) < Ee < 55 MeV (90% ε signal ) Contours show averaged signal PDF (1σ,1.64σ,2σ)

41 tt eeee EE γγ EE ee θθ eeee Data set Fit result full data data sum ACC RMD Data and projected PDF agree well signal (500events) best fit BB ( ) φφ eeee Indication for signallikelihood R sig SS(xx ii ) RR sig = 0.07RR(xx ii ) AA(xx ii )

42 Confidence interval 42 Consistent with no signal assumption Data set BB 90% UL ( ) Sensitivit y ( ) B(μμ + ee + γγ) < (90% CL) In previous result, with data. Consistent including change in analysis. CL curve with data (Ratio of ToyMC with λλ pp MC < λλ pp data )

43 Move of the observables 43 High rank event in either (current/previous) of results are plotted. Previous Current We tested MC experiment to simulate move of observables and compared upper-limits. Data located around the center of the MC distribution.

44 Fit result constrain 44 Usual likelihood function contains constraint term for N RMD と N ACC to be near to the estimation from sideband. CC NN RMD, NN ACC, tt = exp NN RMD μμ RMD 2 2 exp NN ACC μμ ACC 2 2σσ RMD 2 cc( tt) 2σσ ACC In order check the BG distribution in analysis window, fit without constrain term were tested expect ±41.2 N fit no constr ±103 ACC standard fit ±37.7 expect ±33.8 N fit no constr ±59.1 RMD standard fit ±28.4

45 4. MEG II experiment

46 MEG II experiment 46 Upgrade aiming at 10 times higher sensitivity of MEG Main features 2.3 times stronger beam target not easy to deform Replace PMT of inner face of LXe with MPPC Unified, larger volume, stereo wired drift chamber Pixelated timing counter with SiPM read out New detector to tag RMD AccBG Expected sensitivity is

47 MEG II status 47 Xenon detector Drift chamber Timing counter 組み立て中一部張らせているワイヤーが見える片側のみ半分の列数をもつプロトタイプ RDC counter LYSO結晶+プラシンの検出器と可動式のマウント約4000個の紫外線に感度のあるMPPCが取り付けられているところ

48 MEG II prospects 48 Specification MEG I MEG II Beam intensiy (/s) Resolutions Eγ(%, w>2 / w<2) 2.4/ /1.0 γ pos. (mm, u/v/w) 5/5/6 2.6/2.2/5 Ee (kev) θeγ/φeγ (mrad) 9.4/ /3.7 teγ (ps) Efficirncies (%) trigger >99 >99 γ e R&D 2013 Upgrade proposal approve assembly 2017 upgrade complete start data taking DAQ 3 years sensitivity

49 Summary 49 MEG experiment is searching for μ + e + γ, evidence of the physics beyond the standard model of particle. MEG I experiment has been finished and we published final result Eur. Phys. J. C, 76(8), 1-30 New limit is 30 times more stringent than MEGA experiment. MEG II experiment is aiming

50 おわり

近縁の CLFV 探索 51 μ-e 転換 (N μ - N e - ) 現在の上限値は SINDRUM-II 実験から B<7 10-13 (N =Au) 新しい実験の準備が進んでいる COMET, DeeMe, Mu2e μ eee 崩壊 PSI にて Mu3e 実験が準備中これら 2 つのチャンネルは

51 近縁の CLFV 探索 51 μ-e 転換 (N μ - N e - ) 現在の上限値は SINDRUM-II 実験から B< (N =Au) 新しい実験の準備が進んでいる COMET, DeeMe, Mu2e μ eee 崩壊 PSI にて Mu3e 実験が準備中これら 2 つのチャンネルは μeγ とは異なるタイプの相互作用も可能で μeγ と相補的関係にある L = mm μμ κκ + 1 Λ 2 μeγ と共通 μμ RR σσ μμμμ ee LL FF μμμμ + κκ 1 + κκ Λ 2 μμ LLγγ μμ ee LL ( ff LL γγ μμ ff LL ) μeγ には無い項

52 キセノン補助システム 52

53 エレクトロニクス 53 典型的なエレキチェーン Sensor Active splitter Trigger : FPGAを用いて高速な事象再構成を行いトリガー情報を作る条件 : γ 線エネルギー γ-e + の時間差 γ-e + の方向 trigger rate ~13 Hz DRS : PSI で開発された波形取得装置サンプル速度 1.4GHz (DCH は 0.7GHz) Trigger DRS Online computers data size ~1 MB/event (compressed) MIDAS システム採用 : データの取得スローコントロールを管理するシステム PSI が開発トリガーシステムの構成

54 事象再構成 : 概要 54 再構成でのデータの流れガンマ線キセノン検出器 γ 位置ドリフトチェンバー e + 飛跡陽電子タイミングカウンターヒット時間 γ 時間 Eγ e + 時間角度差 θeγ φeγ 時間差 teγ

55 γ 線位置時間 55 位置 ( キセノン中で最初に反応した点 ) a. 中心付近の光子数の分布を χ 2 フィット b. フィット結果の補正シャワーの大きさ斜め入射時間 ( キセノン中で最初に反応した時間 ) 1 点から等方的にシンチレーション光が放たれていると仮定 r i Ω i 2 χχ time = ii tt PMT,ii rr ii vv tt LXe σσ tt NN phe,ii 2 2 χχ pos = ii NN pho,ii ccω ii (uu, vv, ww) σσ ppppp NN pho,ii 2 - 和は 50 光電子以上の PMT についてとる

56 γ 線エネルギー 56 エネルギー各 PMT の波形の和から計算される PMT ごとの光子の伝搬時間は差し引いておくそれぞれの PMT の重みは次を考慮する sum 波形 PMT のゲインと量子効率 ( 光電子の収集率も含む ) PMT がカバーする立体角面ごとの補正係数放出点から光電面を見込む立体角 γ の位置による不均一性の補正 pile-up unfolding 複数 γ 線のパイルアップへの対処シンチレーション光の空間分布 sum 波形のピークサーチ

57 ミッシングターン復元 57 陽電子がチェンバーを複数回通過する場合それぞれの周回が別の陽電子として識別されてしまう事があった一つの陽電子による分かれた軌跡を識別し復元する手法を導入した正しい原点偽の原点 X [cm] Run 51847, Event ドリフトチェンバー 1 st trun 2 nd trun 効果 2 周目を認識できなかったためイベント選別から漏れてしまったイベントの回復約 4% のイベント増加 AccBG イベントの出現と消滅はほぼ同数のため BG 数に対する影響は無い e Z [cm]

58 θeγ, φeγ, teγ 58 μ 粒子の初期位置 (rr μμ ) は飛跡がターゲットと交わる点とする y γ の放出角度 nn γγ = rr γγ rr μμ rr γγ rr μμ φ 角度差 (0 だと完全に反対向き ) z θ x θθ eeγγ = ππ θθ ee φφ eeee = ππ + φφ ee θθ γγ φφ γγ 時間差 tt eeee = tt LXe rr γγ rr μμ cc tt ee

59 PMT 再構成 59 PMT ごとのヒット再構成 constant fraction 法からヒット時間フィルターした波形を積分して光子数を得る Amplitude (mv) % Amplitude (mv) ns height Time (nsec) Time (nsec) raw high-pass

60 LXe 検出器 PMT の較正 60 増倍率 ( ゲイン ) 量子効率 (QE) LED を一定の強度で点灯させる σσ NN 2 = μμ NN + σσ 0 2 (N: 光電子数 ) QQ = GG NN で電荷の関係に直すと σσ QQ 2 = GG(μμ QQ + σσ 0 2 ) α 線源 ( 241 Am) が付いたワイヤー α 線イベントで測定された光電子数と MC シミュレーションで予想される光子数の比から QE を計算する

61 新 γ 線位置補正年レーザー測量機を用いて検出器の内外壁 PMT 取付用の構造体を測量した結果 x 軸 : 1 mrad, y 軸 : 5 mrad 程度の回転他図面からのズレが見つかった対策 PMT の取付方法 + 温度変化に基づいた位置の補正を行う ( キセノンの重量による変形は無視できる ) 修正されるガンマ線位置の平均値は角度の不確かさと同程度 ( 約 4mrad) 補正による u,v 位置の移動 (10 倍 )

ドリフトチェンバー位置合わせ 62 Optical method - 測量器各年の run 開始前精度 0.2-0.3mm (x,y) 1.

3mm (x,y,z) Software method - Millipede alignment 宇宙線カウンタ (CRC) を用いた特殊

62 ドリフトチェンバー位置合わせ 62 Optical method - 測量器各年の run 開始前精度 mm (x,y) mm (z) - レーザートラッカーと corner cube 2011 年から精度 0.3mm (x,y,z) Software method - Millipede alignment 宇宙線カウンタ (CRC) を用いた特殊 run 精度 0.15mm 磁石は off CRC - Michel positron alignment 通常の陽電子 track と fit の残差が小さくなるよう最適化 CRC

63 規格化因子の計算 63 Michel 法通常崩壊陽電子の個数 kk Michel = NNMichel Michel ff PP Michel Michel εε signal ee Michel EEEE εε AA signal signal signal γγ εεtrg εεsel ee εε trg RMD 法 trigger 数輻射崩壊の個数 e + 検出効率 γ 受入効率 trigger 効率 signal 選別効率 signal kk RMD = NNRMD RMD B εε Signal ee RMD EEEE εε εε trg RMD ee εε εε sel trg signal RMD εε sel e + 検出効率 trigger 効率 signal 選別効率どちらの方式も陽電子が検出されているイベント数からスタートするため陽電子検出効率は既に含まれている

64 平均 PDF のフィットとの比較 64 角度変数を 1 次元化 event-by-event でない PDF を用いる別解析と結果を比較したフィット結果は主方式と同様シグナルの有意な超過は無い同じデータを別の方法で解析した上限値は多数の MC の分布の中心付近に位置する別方式 Θ (stereo angle) PDF とデータ本方式

65 High rank events 65 Rank Run Event Pair Rsig t [ps] th Ee [MeV] Eg [MeV] [mrad] ph [mrad] cos AIF

MEG μ + e + γ ( ) ( MEGA) = (BSM) MEG μ + e + γ ( : a few ) 180 γ μ + e +

MEG μ + e + γ ( ) ( MEGA) = (BSM) MEG μ + e + γ ( : a few ) 180 γ μ + e + MEG ( ) 2011 9 10 MEG μ + e + γ ( ) (1.2 10-11 MEGA) = (BSM) MEG μ + e + γ ( : a few 10-13 ) 180 γ μ + e + μ eγ @MEG DC μ + (590MeV 1.3MW @ ) γ: e + : MEG (900L) (VUV) 846 PMT : PMT : PMT PMT (DRS4) particle