HALO: A Physics-Aware LLM Agent Framework for Nanophotonic Design
Abstract
Language models have recently been applied to nanophotonic design, but it remains unclear whether they can reliably translate optical objectives into simulation-ready designs, execute electromagnetic analysis, and revise decisions from numerical feedback. We introduce HALO, a physics-aware framework that couples language-model planners with typed design specifications, electromagnetic simulation, diagnostic evaluation, and optional reuse of prior failure trajectories in an iterative design loop. We further introduce HALO-Bench, a 52-task benchmark spanning lab-derived, paper-derived, and open-ended nanophotonic design tasks under a shared evaluation protocol. We compare three planner configurations: a Fixed Structured Workflow, an Autonomous Structured Agent using the same simulation interface, and an Autonomous Coding Agent that directly writes and executes simulation code. The Fixed Structured Workflow is the most token-efficient and exhibits no observed code- or path-level failures, while autonomous coding can achieve higher task success with stronger models at the cost of additional operational failures. We also study reuse of prior failed trajectories. On targeted multi-round tasks, retrieved failure feedback reduces both iterations to first success and total token use. These results clarify the tradeoffs between explicit interfaces, autonomous execution, and reusable design experience in scientific agents.
Related
- Designing an AI Agent That Does Not Tutor: Design Rules for Socratic Probing in Introductory Physics
- Solver Agent: an Agentic AI Framework for Theoretical Physics Computations Applied to F-theory Uplifts of O3-planes and S-folds
- Photonics-GCCE: group collaborative-competitive evolution multi-agent framework for universal and autonomous optical design
- A Physics-based and Geometry-aware Reduced-order Framework for Accelerated Part-Scale Thermomechanical Simulation of Additive Friction Stir Deposition Process
- Fast Nanophotonic Inverse Design using Precomputed Numerical Green Functions
BibTeX
@misc{zhang2026halo,
title = {{HALO: A Physics-Aware LLM Agent Framework for Nanophotonic Design}},
author = {Yubo Zhang and Jinlin Xiang and Zijun Zhao and Yang Zhao and Eli Shlizerman and Arka Majumdar},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2608.28877},
eprint = {2608.28877},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.28877},
}
RIS
TY - UNPB
TI - HALO: A Physics-Aware LLM Agent Framework for Nanophotonic Design
AU - Yubo Zhang
AU - Jinlin Xiang
AU - Zijun Zhao
AU - Yang Zhao
AU - Eli Shlizerman
AU - Arka Majumdar
PY - 2026
DA - 2026/09/07
PB - arXiv
DO - 10.48550/arxiv.2608.28877
UR - https://arxiv.org/abs/2608.28877
AB - Language models have recently been applied to nanophotonic design, but it remains unclear whether they can reliably translate optical objectives into simulation-ready designs, execute electromagnetic analysis, and revise decisions from numerical feedback. We introduce HALO, a physics-aware framework that couples language-model planners with typed design specifications, electromagnetic simulation, diagnostic evaluation, and optional reuse of prior failure trajectories in an iterative design loop. We further introduce HALO-Bench, a 52-task benchmark spanning lab-derived, paper-derived, and open-ended nanophotonic design tasks under a shared evaluation protocol. We compare three planner configurations: a Fixed Structured Workflow, an Autonomous Structured Agent using the same simulation interface, and an Autonomous Coding Agent that directly writes and executes simulation code. The Fixed Structured Workflow is the most token-efficient and exhibits no observed code- or path-level failures, while autonomous coding can achieve higher task success with stronger models at the cost of additional operational failures. We also study reuse of prior failed trajectories. On targeted multi-round tasks, retrieved failure feedback reduces both iterations to first success and total token use. These results clarify the tradeoffs between explicit interfaces, autonomous execution, and reusable design experience in scientific agents.
ER -