From low- to high-energy nuclear data evaluations

Evaluation of neutron cross sections between 0eV and 20MeV is based on several aspects of nuclear physics such as nuclear reaction and structure models and microscopic and integral measurements. Most of the time, the evaluation process is separately done in the resolved resonance range and the continuum. It may give rise to non-physical mismatches of cross sections and large uncertainties at boundaries. It also leads to an absence of cross correlations between high-energy domain and resonance range. In addition, integral experiments are sometimes only used to check central values (evaluation is “working fine” on a dedicated set of benchmarks). Eventual reduction of uncertainties on cross sections is not straightforward: “working fine” could be mathematically turned into reduced uncertainties. This paper will present several ideas that could be used to avoid such effects. They are based on basic physical principles, recent advances in terms of covariance evaluation methodologies, intensive use of Monte Carlo methods and High Performance Computing (HPC) and on some newly introduced models. A clear connection is made between resonance and continuum energy ranges