Multinucleon and final-state effects in neutrino energy reconstruction
In plain words
Oscillation experiments must reconstruct each neutrino's energy from the particles leaving the nucleus, but some energy goes into undetected neutrons, knocked-out nucleons or reabsorbed pions. The bias this causes must be known to about one percent and is not.
Precise statement
Quantify the bias and spread of reconstructed versus true neutrino energy, $E_{\mathrm{reco}} - E_{\mathrm{true}}$, for calorimetric reconstruction on 40Ar (DUNE, $E_\nu \sim 0.5 \text{ to } 5\,\mathrm{GeV}$) and lepton-kinematics reconstruction on 16O (Hyper-Kamiokande, $E_\nu \sim 0.6\,\mathrm{GeV}$), from 2p2h events, nucleon and pion final-state interactions, and undetected neutrons, to 1 percent of the energy scale. The answer is a model-independent bound or a validated model, tested on electron-scattering data with known beam energy (for example e4nu) and on near-detector data.
What would settle it
Electron-scattering and neutrino near-detector data that constrain the reconstructed-energy smearing on argon and water to 1 percent, reproduced by one event generator without retuning.