EUCLID: A New Approach to Constrain Nuclear Data via Optimized Validation Experiments using Machine Learning

• Published in: EPJ Web of conferences Vol. 284; p. 15006
• Main Authors: Hutchinson, J., Alwin, J., Clark, A.R., Cutler, T., Grosskopf, M.J., Haeck, W., Herman, M.W., Kleedtke, N., Lamproe, J., Little, R.C., Michaud, I.J., Neudecker, D., Rising, M.E., Smith, T., Thompson, N., Vander Wiel, S., Wynne, N.
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Les Ulis EDP Sciences 2023
• Subjects: Design optimization; Elastic scattering; Errors; Inelastic scattering; Machine learning; Research facilities
• ISSN: 2100-014X; 2101-6275; 2100-014X
• DOI: 10.1051/epjconf/202328415006

Compensating errors between several nuclear data observables in a library can adversely impact application simulations. The EUCLID project (Experiments Underpinned by Computational Learning for Improvements in Nuclear Data) set out to first identify where compensating errors could be hiding in our libraries, and then design validation experiments optimized to reduce compensating errors for a chosen set of nuclear data. Adjustment of nuclear data will be performed to assess whether the new experimental data—spanning measurements from multiple responses—successfully reduced compensating errors. The specific target nuclear data for EUCLID are 239Pu fission, inelastic scattering, elastic scattering, capture, nu-bar, and prompt fission neutron spectrum (PFNS). A new experiment has been designed, which will be performed at the National Criticality Experiments Research Center (NCERC).