NUclear Reaction model improvement with Bayesian Statistics – NURBS

The wide range of nuclear physics applications, and especially applications using nuclear reactions, can be distinguished into two major regimes. First, there is the low-energy regime of, e.g., nuclear power plants. For such applications, essentially all relevant nuclear data can and have been measured. Second, there are high energy applications like hadron therapy (up to ~250 MeV), spallation neutron sources and ADS (~500 MeV up to ~2 GeV), cosmogenic nuclides (large range of energy, peaked at ~1 GeV), but also neutrino studies (hundreds of MeV up to tens of GeV), and antimatter experiments (from at rest up to several GeV).
Due to the higher energies and the fact that several projectile types are relevant, measuring all required nuclear data is impossible and nuclear models are needed. Such (stand-alone) models, a combination of an intranuclear cascade model, e.g., INCL++6, and a de-excitation model, e.g., Abla, are able to predict the outcome of a spallation reaction. In such reactions, numerous light particles, e.g., mesons, neutrons, or light charged particles, are ejected. As the result of the interaction, a lighter target nucleus (or more if fission or multifragmentation occur) is produced.
Such models have been greatly improved the last two decades and we have entered a new and more demanding period: quantification of the uncertainties and errors. We propose here a project in which not only uncertainties and errors of the INCL++6/Abla simulations will be quantified, but we will also improve the models thanks to a parallel optimization of some parameters and feedback of these results. The information needed to determine the uncertainties and errors of the results and parameters is provided by Bayesian statistics where experimental data play the role of likelihood.