Physics-Guided Sequence-Based Generative Framework for Acoustic Metamaterial Inverse Design

2026-06-08

AI Summary

This work addresses the challenges in inverse design of broadband acoustic metamaterials, particularly the difficulty in matching multi-frequency responses and preserving geometric fidelity and structural connectivity. To tackle these issues, the authors propose MetaSeq, a novel framework that, for the first time, encodes metamaterial architectures as structured sequences and formulates inverse design as a sequence-to-sequence generation task mapping target acoustic responses to corresponding structural sequences. The approach integrates supervised pretraining with physics-informed reinforcement learning to effectively handle the one-to-many nature of inverse design, while ensuring feasibility through solver-based validation and complexity-aware sampling. In COMSOL simulations, MetaSeq reduces response error by 45% compared to five baseline methods, while maintaining high geometric accuracy and structural connectivity.

Abstract

Acoustic metamaterial (AMM) inverse design is particularly challenging for broadband target responses due to acoustic dispersion: a structure that matches the desired response at one frequency may deviate at others, and modifying geometry to improve one sub-band often perturbs neighboring sub-bands. Yet existing broadband inverse-design approaches are either constrained by predefined templates, or rely on image representations that fail to preserve the geometric precision and structural connectivity required by acoustic structures. We present MetaSeq, a physics-guided, sequence-based generative framework for acoustic metamaterial inverse design. At its core, MetaSeq introduces a language that represents each AMM as a structured sequence, rather than as a pixel grid or fixed template. This representation preserves precise geometry, explicitly encodes connectivity, and casts inverse design as a sequence-to-sequence task from target response to structure sequence. MetaSeq further constructs a balanced, high-fidelity dataset with efficient calibration and complexity-based sampling. To address the one-to-many nature of inverse design, MetaSeq combines supervised pretraining with reinforcement learning fine-tuning guided by a physics-based solver and validity checker. Extensive evaluations against COMSOL and five baselines show that MetaSeq reduces response error by 45% over the best baseline.

Problem

• acoustic metamaterial
• inverse design
• broadband response
• acoustic dispersion
• geometric precision

Innovation

• sequence-based representation
• physics-guided generative model
• acoustic metamaterial inverse design
• reinforcement learning fine-tuning
• structural connectivity encoding