This research introduces a novel Optimization by Prompting approach that leverages large language models to solve nuclear engineering design challenges without complex mathematical formulations.

• Successfully applied to nuclear fuel assembly configurations and Boiling Water Reactor fuel lattice design
• Requires no hyperparameter tuning or sophisticated optimization algorithms
• Enables engineers to describe optimization problems in plain language
• Demonstrates how LLMs can iteratively explore design spaces to find optimal solutions

This advancement matters for engineering by making sophisticated nuclear design optimization more accessible, potentially accelerating innovation in energy systems while reducing the expertise barrier for complex engineering problems.

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