Answers are grounded in this paper and cite their sources.
Modeling and simulation problems in the nuclear field often concern complicated systems whose optimizations are complex and multi-objective in nature. The optimization of nuclear systems within these high-dimensional simulation spaces presents difficult and computationally expensive challenges. This study introduces a framework that pairs Monte Carlo N-Particle (MCNP) simulations of a nuclear assay system with machine learning (ML) surrogate modeling and multi-objective Bayesian optimization. Rather than replacing MCNP or other computational tools, our approach leverages ML to guide the selection and focus of the simulation efforts, ensuring that the MCNP simulations explore the most promising regions of the design space. This methodology enables the systematic identification of detector configurations aligned with specific operational goals, preserving the rigor of physics-based modeling while enhancing the efficiency and objectivity of the design process.
No takes yet. Share an insight, caveat, or question.
Reynolds et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: