Presentation is loading. Please wait.

Presentation is loading. Please wait.

MEG Experiment Data Analysis: a status report

Similar presentations


Presentation on theme: "MEG Experiment Data Analysis: a status report"— Presentation transcript:

1 MEG Experiment Data Analysis: a status report
Fabrizio Cei INFN & University of Pisa On behalf of the MEG Collaboration CNS1, Roma, Dec 3rd 2012 03/12/2012 Fabrizio Cei

2 Outline Brief remind of the MEG experiment; Analysis strategy;
Results ; Advances in analysis; Re-analysis of 2009 data; Sensitivity & sideband analysis for 2011 data; Other analyses; Perspectives. 03/12/2012 Fabrizio Cei

3 Brief remind of the MEG Experiment
03/12/2012 Fabrizio Cei

4 Goal and signal - back-to-back topology; Search for m  eg LFV decay.
Target sensitivity BR  wrt normal muon decay. Previous Upper Bound: BR  1.2 x (MEGA, 2001) Signature: - back-to-back topology; - energy equally shared between e+ and g; - simultaneous emission. e+  g qeg = 180° Ee = Eg = 52.8 MeV Te = Tg 03/12/2012 Fabrizio Cei

5 (radiative muon decay, RMD)
Background accidental   e n n   e g n n ee  g g eZ  eZ g physical m  e g n n (radiative muon decay, RMD) e  g n e+ + n RRD = Rm* BR(m → enng)  0.1 Racc g 03/12/2012 Fabrizio Cei Fabrizio Cei 5

6 The Detector World’s most intense continuous muon beam
pE5 (up to 108 stopped m+/s); Photon detection by a Liquid Xenon calorimeter; Positron detection by a Drift Chamber spectrometer for momentum and by scintillation bars for timing measurements. 03/12/2012 Fabrizio Cei

7 Summary of MEG performances
2009 2010 2011 Gamma Energy (%) 1.9% (w > 2 cm) 1.7% (w > 2 cm) Gamma Position (mm) 5 (u,v),6(w) 5 (u,v), 6 (w) Positron Momentum (%) 0.59 (core 80%) 0.61 (core 79%) 0.59 (core 86%) Positron Angle (mrad) 6.7 (F,core), 9.4 (Q) 7.2 (F, core), 11.0 (Q) 6.5 (F,core), 10.6 (Q) Vertex Position (mm) 1.5 (Z), 1.1 (Y) 2.0 (Z), 1.1 (Y) 1.9 (Z), 1.3 (Y) Gamma-Positron timing (ps) 146 (core) 126 (core) 133 Gamma efficiency (%) 58 59 Trigger efficiency (%) 92 95 03/12/2012 Fabrizio Cei

8 Analysis strategy: blind + likelihood
Signal and bkg optimization done in sidebands Timing sidebands Eg sideband Events in the blind box ( 0.2%) are hidden up to the end of optimization procedure RMD events 03/12/2012 Fabrizio Cei

9 MEG likelihood analysis
The most dangerous bck is measured ! Likelihood function 03/12/2012 Fabrizio Cei

10 PDF’s 03/12/2012 Fabrizio Cei

11 Normalization where: 107 pre-scaling factor TRG = 22: Michel events trigger (only DCH track required) TRG = 0: MEG events trigger 03/12/2012 Fabrizio Cei

12 Results 03/12/2012 Fabrizio Cei

13 Data sets for 2009 & 2010 Muons on target 2009 Muons on target 2010
Statistics for 2010 about twice that for 2009 Stable detector conditions (LXe, DC, TC) Optimized degrader Improved electronics timing Slightly lower DC efficiency because of higher noise. 03/12/2012 Fabrizio Cei

14 Sensitivity evaluation
Upper limit averaged over a sample of 1000 toy MCs, generated with 0 signal events and using background rates and spectra measured in sidebands. Confirmed by applying the analysis procedure to events falling in timing and angular sidebands. 03/12/2012 Fabrizio Cei

15 Unbinned maximum likelihood fit 2009
Three independent analyses performed by different groups gave consistent results. 03/12/2012 Fabrizio Cei

16 Unbinned maximum likelihood fit 2010
03/12/2012 Fabrizio Cei

17 Combination 2009-2010 Factor five improvement wrt previous limit !
Published result: PRL 107,171801, (2011) Factor five improvement wrt previous limit ! Sensitivity 1.6 x 10-12; difference due to 2009 fluctuation. 03/12/2012 Fabrizio Cei

18 Advances in analysis 03/12/2012 Fabrizio Cei

19 New Kalman filter A new Kalman filter was developed recently, with important features: More accurate measurement model; Track model based on GEANE package  well linked with the GEANT-based MC simulation of the detector; Accurate extrapolation to the Timing Counter bars  stricter correlation between DCH and TC information; Use of timing information from TC to refine hit position  refine track fitting; More precise evaluation of track length (important for relative timing resolution) and uncertainties on kinematic variables; Compatibility with a (possible) active target. Several tests performed. General results: 15% higher efficiency of track reconstruction (5% for Ee > 50 MeV); Resolutions consistent with the old code; Lower high energy tail in positron energy spectrum. 03/12/2012 Fabrizio Cei

20 New vs Old algorithm 1) Ee Difference of reconstructed positron
variables for positrons reconstructed by both the OLD and the NEW algorithm. Shown XNEW – XOLD No significant biases observed. Ee feg qeg Teg 03/12/2012 Fabrizio Cei

21 New vs Old algorithm 2) Relative timing resolution 
Positron energy spectrum: Black = OLD Red = NEW 5% higher efficiency and reduced tails 03/12/2012 Fabrizio Cei

22 New waveform reconstruction algorithm
Pile-up important at high rate  needed a pile-up identification algorithm. Difficult to identity on single PMT waveform because of fluctuations  needed to sum waveforms of several (all) PMTs. Simple charge sum and/or light distribution not enough: - difficult to extract energy of main pulse; - shape depends on interaction depth. New strategy: - operate a first-order event reconstruction; - use timing and position reconstructions and PMT calibration constants to correctly weigh the single PMT waveforms and perform the sum; - fit the sum waveform with a multi-pulses template and identify as “main pulse” that closest in time with the event time; - remove other pulses and determine the energy of the main pulse. Final results: - signal reconstruction efficiency increases from 58 to 63%. - reduced dependence of energy spectra on event depth. 03/12/2012 Fabrizio Cei

23 Pileup elimination 1) Original sum waveform
Fitted with 3-peak template Black line: main pulse Blue dashed lines: removed pulses Start of integration window to determine main pulse energy. 03/12/2012 Fabrizio Cei

24 Pileup elimination 2) 55 MeV photons from p0 decay
BKG event by event comparison NEW-OLD Better reconstruction of shallow events After applying pileup elimination algorithm, pileup flagged events look very similar to no pileup events. 03/12/2012 Fabrizio Cei

25 Analysis of 2009 data with improved algorithms In progress
03/12/2012 Fabrizio Cei

26 Sensitivity & Normalization
Sensitivity with toy MC: BR = 3.1 x 10-12 Consistent with timing sidebands: BR = (3 ÷ 3.2) x 10-12 Previous result was BR = 3.3 x (7% improvement, consistent with expectations) Normalization factor: k = 1.19 x (Michel) k = 1.18 x (RMD) Previous calculation was 1.1 x 1012 03/12/2012 Fabrizio Cei

27 Event ranking Events ranked according to the variable
Independent analyses have compatible event ranking (many events in common with limited re-shuffling). Ranking not used in likelihood fit. 03/12/2012 Fabrizio Cei

28 The top score event Magenta: BKG Red: RD Green: Signal Blue: Total
OLD Algorithms pdf’s NEW Algorithms pdf’s 03/12/2012 Fabrizio Cei

29 Event distributions Lines: 1, 1.64, 2 s contours
51 MeV < Eg < 55 Mev && 52 MeV < Ee < 55 MeV Qeg > 178o && |Teg| < 280 ps Lines: 1, 1.64, 2 s contours 03/12/2012 Fabrizio Cei

30 Top ranked events: New vs Old
Small displacements Black: Old ranking, events 1-8 Red: New ranking, events 1-8 Blue Lines: Connection between events present in both lists. 03/12/2012 Fabrizio Cei

31 Signal region likelihood fit; Confidence Level curve; Limits; ...
Analysis in progress: Signal region likelihood fit; Confidence Level curve; Limits; ... 03/12/2012 Fabrizio Cei

32 2010 re-analysis re-processed with new algorithms.
2010 data sample ( 2 larger than 2009 sample) re-processed with new algorithms. Pdf preparation under way. Complete re-analysis starting soon: - sensitivity; - normalization; - sideband checks; - signal region fit. 03/12/2012 Fabrizio Cei

33 Sensitivity & sideband analysis for 2011 data
03/12/2012 Fabrizio Cei

34 Data 2011 Number of stopping muons in 2011.
Data statistics more than doubled: 2011 > ( ) 1.85 x 1014 Smooth running conditions (except for a cryo-plant damage in November); Improved LXe calibration (NaI replaced by BGO); Higher DAQ Efficiency and live time (> 95 %); Resolutions/efficiencies comparable or better than that of previous years. All data processed with old and new algorithms for further checks. 03/12/2012 Fabrizio Cei

35 Normalization 2011 Confirmed by an independent calculation based on Muon Radiative Decay events 03/12/2012 Fabrizio Cei

36 Sensitivity 2011 Two independent evaluations (different analysis groups, different pdf’s). Median =  using k factor Sensitivity = 1.26 x 10-12 1.35 x 10-12 03/12/2012 Fabrizio Cei

37 Example of sideband fit
We are ready for unblinding 2011 data and plan to do it after finishing checks on 2009/2010 samples Sideband fit agrees with sensitivity 03/12/2012 Fabrizio Cei

38 Future perspectives 03/12/2012 Fabrizio Cei

39 Data collected in 2012 Higher number of collected muons than in 2011; still 20 days of DAQ. 03/12/2012 Fabrizio Cei

40 Detector status Presented on Nov. 30th . No significant deterioration
since end of Sep. If we will have beam time, we plan to run for three months during 2013 without opening the detector. 03/12/2012 Fabrizio Cei

41 Projected sensitivity
Stopped m+ per year Evaluated by toy MC and likelihood analysis Expected final sensitivity:  5 x 10-13 03/12/2012 Fabrizio Cei

42 Other searches 1) Study of RMD events and measurement of RMD branching ratio and Michel parameters. Paper in preparation. Dots: data Black Line: theoretical prediction with Pol = 0.9 Grey boxes: uncertainties 03/12/2012 Fabrizio Cei

43 Other searches 2) Measurement of muon beam polarization with Michel e+. Paper in preparation. Search for more exotic muon decays (mediated by pseudo-scalar particles or with massless Majorons) Best fit: P = 0.89  0.03 RED: DATA BLUE: MC,POL = 0 q q RED: DATA BLUE: MC,POL = 1 03/12/2012 Fabrizio Cei

44 Summary The analysis of MEG experiment data is going on to have shortly a new publication, with a larger sample of data and higher quality results. 2009 Data: reprocessed with new algorithms; analysis under way; 2010 Data: reprocessed with new algorithms; analysis close to start; 2011 Data: ready for unblinding after conclusions of analysis; Combined result to be presented at the Winter Conferences and published; 2012 Data: sample larger than 2011  expected relevant contribution to the final result. 2013: Three months of data taking planned ( 0.5 x 2012 statistics). Expected final sensitivity  (5 ÷ 6) x at the end of DAQ. 03/12/2012 Fabrizio Cei

45 On demand 03/12/2012 Fabrizio Cei

46 Background m  e g n n accidental   e n n physical   e g n n
ee  g g eZ  eZ g Racc  (Rm)2 * (DQ)2 * (DEg)2 * DT * DEe O Muon rate to be used is a trade off between expected number of signal events and background level; O Used 3 x 107 m+/s; O Sensitivity is limited by accidental background; O High resolution detectors and a continuous muon beam are mandatory. physical m  e g n n (radiative decay, RMD) e  g n e+ + n RRD = Rm* BR(m → enng)  0.1 Racc g 03/12/2012 Fabrizio Cei Fabrizio Cei 46

47 Part of Total  60 physicists 03/12/2012 Fabrizio Cei

48 Overview of calibration system
Needed to ensure: Required precision; Long term detector stability; Continuous checks for a detector based on innovative technology (LXe). 03/12/2012 Fabrizio Cei

49 A sideband event vs pdf’s
Magenta: BKG Red: RD Green: Signal Blue: Total 03/12/2012 Fabrizio Cei


Download ppt "MEG Experiment Data Analysis: a status report"

Similar presentations


Ads by Google