A. Baldini: New Results from the MEG Experiment
In non-GUT theories, we can suppress LFV rates by lowering the right-handed neutrino mass. But in GUT it is independent on it and we cannot suppress it. For instance, in SO(10) GUT with type-II see-saw, there is testability of the model in the branching fraction of muon decays to electron-photon pairs.
A non-zero value of theta_13 is certainly a positive news in this panorama, since it inproves the testability of these models.
The signal is a two-body decay, an electron and photon back-to-back. Backgrounds are mainly due to accidentals. The detector is made of a solenoid spectrometer, scintillator bars, and liquid xenon (900 liters) read by PMTs. A monitor of photon energy is operational during run: it is a 17.67 MeV line from 7Li.
The data has been collected since 2008, but most of it is from the 2012-2013 run. They got a limit of 2.4E-12 in the muon to e-gamma BR in 2010. In this talk are given results based on 2011 data, which is double of the 2010 dataset.
With respect to the previous analysis, they treated differently the noise from the drift chamber, improving the angular resolution by 10%. They also revised the track fit algorithm, based on the kalman filter technique. They finally improved the elimination of pileup.
They blind the box in timing coincidence versus photon energy, and use sidebands in time for looking at the background energy spectrum; they also look at energy sidebands with zero timing difference, where at low energy they get radiative decays of the muon, measuring the timing resolution. So they can extrapolate in the signal box their timing resolution and the expected signal shape in time.
A maximum likelihood analysis is performed in the signal region, which accounts for signal and background pdfs, with a per-event error matrix from the kalman filter. THere are five obsevrables to characterize the decay: electron and photon energy, angles, and time.
Before opening the blind box, they do a fictitious likelihood analysis in a sideband in time, and find observed upper limits which are consistent with the sensitivity.
In the blind box they see no events. They combine the 2009-2011 datasets, and the best fit gives a signal consistent with zero. All in all, at 90%CL they have a upper limit of 5.7E-13. This is four times more stringent than the previous one. The new result is compatible with the old one, and they estimate the probability to observe a upper limit equal or lower than the one observed in 2011 of 24%.
The result removes a part of the models with values of theta_13 of 10 degrees. Due to the recent null LHC SUSY searches, one can imagine that the SUSY U(3) flavor symmetry is broken to U(2), which gives two families with higher squark masses and a lighter one. This has consequences in the lepton sector, but the result of MEG excludes most of this scenario.