Comprehensive estimation of nuclide production cross sections using a phenomenological approach

Iwamoto, Hiroki   ; Meigo, Shinichiro   ; Sugihara, Kenta*

Nuclide production cross sections are crucial in nuclear research, development, space exploration, and astrophysical investigations. Despite their importance, limited experimental data availability restricts the practicality of phenomenological approaches to comprehensive cross-section estimation. To address this, we propose a Gaussian process-based machine learning (ML) model capable of transferring knowledge from elements with abundant data to those with limited or no experimental data. Our ML model not only enables comprehensive cross-section estimations for various elements but also demonstrates predictive capabilities akin to physics models, even in regions with scarce training data.