RENO: 5 MeV neutrino excess ("9σ")

[mfb]

"9σ" in quotation marks as it seems to include the statistical uncertainty only.

Article at physicsworld.com
The result was shown at the XXVII International Conference on Neutrino Physics and Astrophysics, the slides don't seem to be available and the abstract is not very helpful.

RENO measured the energy spectrum of reactor neutrinos, a spectrum that rises quickly from 1 to 3 MeV and then gradually drops off up to ~7 MeV. At around 5 MeV, the observed spectrum does not match the predictions with a discrepancy of up to 15%.

My personal guess: something went wrong in the model that predicts this shape. Without this theoretical shape as comparison there is no structure visible at all. Do we really trust the models of reactor neutrino spectra to be accurate within 10% everywhere? If I look at high-energy physics Monte Carlo simulations, where some spectra have deviations of more than a factor of 2 (even between simulations - it is known that it is a problem of the simulations, not the measurements), I'm not sure.

 

[Orodruin]

The 5 MeV bump is briefly discussed in the review https://arxiv.org/ct?url=http%3A%2F%2Fdx.doi.org%2F10%252E1016%2Fj%252Enuclphysb%252E2016%252E04%252E012&v=744e82cf . Of course, this was written before the Neutrino 2016 conference, but it does contain some discussion and references.

 

[ChrisVer]

. Without this theoretical shape as comparison there is no structure visible at all. Do we really trust the models of reactor neutrino spectra to be accurate within 10% everywhere?

Well more or less I guess that's the problem...
But I don't understand how with such deviations one can trust the neutrino oscillation results?

 

[mfb]

The neutrino mixing experiments don't rely on a theoretical flux value as far as I know. At least reactor- and accelerator-based experiments always have a near detector as well. Atmospheric measurements have different angles to get ratio measurements, and solar neutrino experiments can measure ratios between the neutrino types.

RENO is a good example: Divide the far detector signal by the near detector signal, and you are sensitive to mixing but the original flux and therefore the bump does not appear at all.