Identification of Effective Integral Experiments for Nuclear Data–Induced Uncertainty Reduction Through Sparse Modeling

Proposed here is a novel method involving integral experiments effective in reducing nuclear data–induced uncertainty. The method is formulated by the extended bias factor method and the L1 norm–based sparse modeling to address computational difficulty in combinatorial optimization. The pseudo-design parameters only sensitive to specific microscopic reactions are defined and can be used to identify integral experiments used for the validation of nuclear data. The method is applied to two pseudoparameters: neptunium-237 capture and bismuth-209 inelastic scattering cross sections, while considering integral experimental data used in ADJ2017 together with sample worth measurements made at the Kyoto University Critical Assembly. The results indicate that the proposed method successfully identifies a small subset of effective integral experiments.