An Agentic Framework for Autonomous Metamaterial Modeling and Inverse Design

2025/06/07 by Darui Lu, Lu, Darui, Jordan M. Malof +3 · 5 citations
Computer Science · Materials Science · Physics and Astronomy · #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #FOS: Physical sciences #Materials Science (cond-mat.mtrl-sci) #Metamaterials and Metasurfaces Applications #Neural Networks and Reservoir Computing #Photonic Crystals and Applications

paper · pdf · doi:10.48550/arxiv.2506.06935

openalex publication_date 2025/06/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Recent significant advances in integrating multiple Large Language Model (LLM) systems have enabled Agentic Frameworks capable of performing complex tasks autonomously, including novel scientific research. We develop and demonstrate such a framework specifically for the inverse design of photonic metamaterials. When queried with a desired optical spectrum, the Agent autonomously proposes and develops a forward deep learning model, accesses external tools via APIs for tasks like simulation and optimization, utilizes memory, and generates a final design via a deep inverse method. The framework's effectiveness is demonstrated in its ability to automate, reason, plan, and adapt. Notably, the Agentic Framework possesses internal reflection and decision flexibility, permitting highly varied and potentially novel outputs.

Citations

• Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents
• MetasurfaceViT: A generic AI model for metasurface inverse design
• A multi-agentic framework for real-time, autonomous freeform metasurface design
• AIDE: AI-Driven Exploration in the Space of Code
• Agent Laboratory: Using LLM Agents as Research Assistants
• MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering
• The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
• Neural Operators for Accelerating Scientific Simulations and Design
• Tree of Thoughts: Deliberate Problem Solving with Large Language Models
• Self-Consistency Improves Chain of Thought Reasoning in Language Models
• Inverse deep learning methods and benchmarks for artificial electromagnetic material design
• Highly accurate protein structure prediction with AlphaFold
• Neural-adjoint method for the inverse design of all-dielectric metasurfaces
• Language Models are Few-Shot Learners
• Nanophotonic Particle Simulation and Inverse Design Using Artificial Neural Networks

Cited by

• A NEAT Approach to Evolving Neural-Network-based Optimization of Chiral Photonic Metasurfaces: Application of a Neuro-Evolution Pipeline
• Advancing Machine Learning Optimization of Chiral Photonic Metasurface: Comparative Study of Neural Network and Genetic Algorithm Approaches
• Towards autonomous quantum physics research using LLM agents with access to intelligent tools
• Agentic Exploration of Physics Models
• Autonomous agentic design for photonics

Related

• A multi-agentic framework for real-time, autonomous freeform metasurface design
• CoSP: Reconfigurable Multi-State Metamaterial Inverse Design via Contrastive Pretrained Large Language Model
• Probabilistic representation and inverse design of metamaterials based on a deep generative model with semi-supervised learning strategy
• Data driven approaches in nanophotonics: A review of AI-enabled metadevices
• Map Optical Properties to Subwavelength Structures Directly via a Diffusion Model