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 ID Waveform from 55MeV photon

COBRA Solenoid superconducting solenoid gradient B-field (0.49-1.27 T) very thin conductor and cryostat wall (0.2X0) Drift Chamber segmented radially (16 sectors) helium:ethane (50:50) opened-frame very thin cathode foil with pads 2x10-3 X0 along positron trajectory Timing Counter 2-layers of scintillators - scintillator bars (outer) - scintillator fibers (inner)

MEG Hidden box on (Teγ, Eγ) BG Accidental: Teγ-sideband Radiative muon decay: Eγ-sideband BG Eγ spectrum Blind box Teγ resolution

MEG Likelihood Three event types: signal, radiative μ decay (RMD), accidental BG Observables xi : Eγ, Ee, Teγ, θeγ, φeγ Fit parameters: Nsignal, NRMD, NBG PDF Signal: measured detector response function RMD: theoretical spectrum convoluted with measured detector response BG: measured BG spectrum Confidence interval Feldman-Cousins method with profile likelihood ratio ordering

PDFs PDF PDF Radiative muon decay (RMD)

MEG 1999: @ PSI 2006: 2007: 2008: 2009: (DAQ time~35 ) Run2008 @LP2009 (B<2.8 10-11 (90% C.L.)) 2010: (DAQ time ~56 ) run2009 Run2009 ICHEP2010 B<1.5 10-11 (90% C.L.) 2011: ( DAQ ~104 ( )) Run2009+2010 EPS-HEP2011 (PRL )

MEG run2009 ( ) (ICHEP2010) Run2010 (2010 2009 2 ) Likelihood 2009+2010

Run 2009/2010 Run2009 DC (HV ) ~43 days # of μ stops (run2009) 5 10 13 μ-stops μ-stop rate: 2.9 10 7 s -1 Total μ-stop: 6.5 10 13 2week Run2010 μ Accelerator maintenance # of μ stops (run2010) 5 10 13 μ-stops 2week ~67 days μ-stop rate: 2.9 10 7 s -1 Total μ-stop: 1.1 10 14 CEX calibration run run2009

LXe (γ detector), DC (e + tracker) MEG γ (p-γ reaction AmBe) DC (Millipede CR

(Two turn method: ) likelihood ΔΦ - ΔE

Performance Summary γ energy γ position γ timing e + mom. e + angle e + timing γ-e + timing γ-e + angle Vertex position Trigger eff. μ stop rate DAQ time Total μ stops Run2009 Run2010 1.9% (w>2cm) 1.9% (w>2cm) 5mm (u,v) / 6mm (w) 5mm (u,v) / 6mm (w) 96ps 67ps 0.59 (core) 0.61 (core) 6.7mrad (Φ,core), 9.4mrad (θ) 7.2mrad (Φ,core), 11.0mrad (θ) 107ps 107ps 146ps (core) 122ps (core) 13.1mrad(Φ), 14.5mrad(θ) 14.0mrad(Φ), 17.1mrad(θ) 1.5mm(Z), 1.1mm(Y,core) 2.0mm(Z), 1.1mm(Y,core) 91 92 2.9 10 7 s -1 2.9 10 7 s -1 35days 56days 6.5 10 13 1.1 10 14 Resolution in sigma

= BG toy MC 2009 2010 2009+2010 B= 3.3 10-12 (median) B = 2.2 10-12 (median) B = 1.6 10-12 (median)!

Selection (Ee-Eγ): Θeγ<178.4 Teγ <0.278ns Blue curves: PDF contour (1, 1.64, 2-σ) (cosθeγ-teγ): 51<Eγ<55MeV 52.34<Ee<55MeV likelihood Teγ-sideband (off-time) Angle-sideband (off-angle) Off-Time B < 2.2 10-12 Off-Angle

Run2009 Θeγ<178.4 Teγ <0.278ns 51<Eγ<55MeV 52.34<Ee<55MeV Blue curves: signal PDF contour (1, 1.64, 2-σ) Events with highest signal likelihood (S/(0.1R+0.9B)) are numbered.

Run2009 Likelihood Fit : 48<Eγ<58MeV, 50<Ee<56MeV, Teγ <0.7ns, θeγ <50mrad, Φeγ <50mrad BG (best fit) Total Best fit Nsignal = 3.4 B = 3.2 10-12 RMD (best fit) Signal (best fit) Signal (UL 90%C.L.) Total w/ signal UL ICHEP2010 (Nsig=3.0)

Run 2009 Confidence Interval 1.7 10-13 < B < 9.6 10-12 (90%C.L., incl. sys.) CL curve: confidence level allowed region B = 0 is marginally excluded, but significance is not high. (p-value is only 8%)

Run2010 Θeγ<178.4 Teγ <0.278ns 51<Eγ<55MeV 52.34<Ee<55MeV Blue curves: signal PDF contour (1, 1.64, 2-σ) Events with highest signal likelihood (S/(0.1R+0.9B)) are numbered.

2009 Run2010 2010

Run2010 Likelihood Fit : 48<Eγ<58MeV, 50<Ee<56MeV, Teγ <0.7ns, θeγ <50mrad, Φeγ <50mrad BG (best fit) RMD (best fit) Signal (best fit) Total Signal (UL 90%C.L.) Total w/ signal UL Best fit Nsignal = -2.2 B = -9.9 10-13

Run2010 Confidence Interval B < 1.7 10-12 (90%C.L., incl. sys.)

2009+2010 Combined B < 2.4 10-12 (90%C.L., incl. sys.) Combined data set is analyzed as a single data set (Note that PDF is on an even-by-event basis.)

2009+2010 Combined Profile likelihood ratio N.B. likelihood curves are not directly used in confidence interval calculation Compatibility bw/ 2009 and 2010 ~15%

RMS of Δ(ΔNLL) (not used in final calculation) Confidence interval (frequentist-bayesian hybrid) Uncertainties of PDF parameters: fluctuating param. in toy MC generation Uncertainties of NRMD, NBG : profiling B upper limit ~2%

Data set B fit Sensitivity LL UL 2009 2010 2009+2010 3.2 10-12 3.3 10-12 1.7 10-13 9.6 10-12 -9.9 10-12 2.2 10-12 - 1.7 10-12 -1.5 10-12 1.6 10-12 - 2.4 10-12 Run2009+2010 B (μ + e + γ) < 2.4 10-12 (90%C.L.) (1.2 10-11 ) 5!

MEG Run2011 2011 6 Multi buffer scheme DAQ HV drift chamber TC run2010 Run2012 run2011 +α

MEG ( BG ) Efficiency improvement with thinner DC cables / optimum cable layout Better positron reconstruction with Reduced noise (adaptive filter/new DC HV) Z-measuring fiber counters Calibration with monochromatic positron beam (Mott scattering) Positron timing Eγ resolution with precise calibration and better reconstruction algorithms Beam intensity optimization...

(2011) 10-13 2008 2009 2010 2011 2012 BG 2012 data UL (2009) ~5 10-13 10-13 2, 3 data UL (2009+2010) Band: range of possible improvements of detector performance...?

MEG μ eγ R&D PMT ( ) H22-26 ( ) MPPC(α ) MPPC PMT active target

Run2009, 2010 5 (B (μ + e + γ) < 2.4 10-12 (90%C.L.)) arxiv:1107.5547, to appear in PRL Run2011 10-13 2012 ~5 10-13 R&D

MEG is further constraining new physics! MEG (this work) MEG (this work) B-physics constraint g-2 discrepancy MEG (this work) Recent T2K result favors large θ13!

Backup