Optimization through In-Context Learning and Iterative LLM Prompting for Nuclear Engineering Design Problems

2025-03-25

AI Summary

Multi-objective optimization of nuclear fuel assemblies—particularly balancing reactivity control and power distribution in boiling water reactor (BWR) lattice configurations—remains computationally intensive and heavily reliant on domain-specific algorithms and manual tuning. Method: This work proposes an iterative prompt optimization framework leveraging large language models (LLMs) as general-purpose optimizers. It requires only a natural-language problem description, a lightweight evaluator, and a structured parsing script—eliminating the need for explicit mathematical modeling, hyperparameter tuning, or custom algorithm design. Physical constraints and objective preferences are implicitly encoded via in-context learning. Contribution/Results: To our knowledge, this is the first application of LLMs as black-box optimizers in nuclear engineering design. Evaluated on BWR fuel assembly optimization, the method achieves superior Pareto-optimal solutions compared to conventional metaheuristic approaches, demonstrating robustness, scalability, and reduced reliance on expert-derived heuristics.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Prompt Engineering / Prompting
• Machine Learning: Optimization

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

The optimization of nuclear engineering designs, such as nuclear fuel assembly configurations, involves managing competing objectives like reactivity control and power distribution. This study explores the use of Optimization by Prompting, an iterative approach utilizing large language models (LLMs), to address these challenges. The method is straightforward to implement, requiring no hyperparameter tuning or complex mathematical formulations. Optimization problems can be described in plain English, with only an evaluator and a parsing script needed for execution. The in-context learning capabilities of LLMs enable them to understand problem nuances, therefore, they have the potential to surpass traditional metaheuristic optimization methods. This study demonstrates the application of LLMs as optimizers to Boiling Water Reactor (BWR) fuel lattice design, showing the capability of commercial LLMs to achieve superior optimization results compared to traditional methods.

Problem

• Optimizing nuclear fuel assembly designs using LLMs
• Replacing traditional methods with prompt-based optimization
• Applying LLMs to Boiling Water Reactor fuel lattice design

Innovation

• Iterative LLM prompting for optimization
• Plain English problem description
• In-context learning surpasses metaheuristics