Methodology for physics-informed generation of synthetic neutron time-of-flight measurement data

Accurate neutron cross section data are a vital input to the simulation of nuclear systems for a wide range of applications from energy production to national security. The evaluation of experimental data is a key step in producing accurate cross sections. There is a widely recognized lack of reproducibility in the evaluation process due to its artisanal nature and therefore there is a call for improvement within the nuclear data community. This can be realized by automating/standardizing viable parts of the process, namely, parameter estimation by fitting theoretical models to experimental data. There are numerous candidate methods to approach this type of problem, but many rely on large, labeled datasets that are not accessible to the nuclear data evaluator. For a reaction cross-section, there are usually just a handful of datasets, none of which can be considered labeled because evaluators never have access to the exact solution (cross section). This work leverages problem-specific physics, Monte Carlo sampling, and a general methodology for data synthesis to generate unlimited, labeled experimental cross-section data of high-utility. The synthesized data is said to be of high-utility because it is statistically similar to the observed data. Heuristic and, where applicable, rigorous statistical comparisons to observed data support this claim. The methodology is split into two generative models. The first generates a realization of an energy-differential cross section for a given isotope. The second takes the output from the first as a determined input and generates noisy experimental observables (radiation detector signals) from the determined cross section realization. The latter is the primary development of this article and is based/limited to transmission measurements at Rensselaer Polytechnic Institute (RPI). The former leverages an existing method for model parameter sampling in the resolved resonance region (RRR), thus limiting the current demonstration to the RRR of incident neutron energies. An open-source software is published alongside this article that executes the complete methodology to produce high-utility synthetic datasets. The goal of this work is to provide an approach and corresponding tool that will allow the evaluation community to begin exploring more data-driven, ML-based solutions to long-standing challenges in the field. (c) 2023 Elsevier B.V. All rights reserved.