ReactorFold: Generative discovery of nuclear reactor cores via emergent physical reasoning

TL;DR

ReactorFold leverages language models to generatively explore nuclear reactor core designs, discovering novel configurations and expanding design space beyond traditional methods by internalizing physical relationships.

Contribution

It introduces a generative framework using language models for nuclear reactor design, enabling autonomous discovery of innovative and asymmetric core configurations.

Findings

Model adjusts Gd inventory to meet power constraints.

Discovers asymmetric configurations beyond traditional heuristics.

Accesses previously inaccessible design regimes.

Abstract

Designing nuclear reactor cores requires navigating large discrete design spaces governed by complex neutronic interactions. Traditional deterministic, metaheuristic, and machine-learning-assisted methods search within fixed, human-defined configuration spaces, limiting their ability to discover fundamentally new design topologies. Here we introduce ReactorFold, a generative framework that reformulates fuel-assembly design as a sequence modeling problem for language models. Using Monte Carlo data, parameter-efficient fine-tuning, and Direct Preference Optimization (DPO), the model learns the latent structure of a pressurized-water-reactor assembly and generates candidate layouts in a single forward pass. Notably, the DPO-aligned model exhibits emergent design-space expansion: despite being trained exclusively on configurations with a fixed number of gadolinium burnable absorber (Gd) rods, it autonomously adjusts Gd inventory to satisfy strict power-peaking constraints. The model also discovers high-performing asymmetric configurations that challenge conventional symmetric loading heuristics, accessing design regimes inaccessible to conventional search methods and demonstrating that language models can internalize causal physical relationships and transcend human-imposed design constraints.