parameterized surface modeling

This work proposes a network-oriented modeling and control framework for metasurfaces, treating them as wave-routing components within wireless communication systems. Inspired by network layering principles, the approach leverages graph theory to model multi-metasurface systems and integrates heuristic and path-search algorithms to optimize control strategies. Innovatively mapping the metasurface control problem onto a network-layer architecture, the framework establishes a standardized interface compatible with Omnet++ simulation and seamless integration into communication system workflows. By unifying networked control methodologies for metasurfaces, the proposed framework demonstrates significant potential in enhancing data rates, energy efficiency, privacy preservation, and environmental awareness, thereby laying a foundational groundwork for AI-driven intelligent networks of the future.

Inverse design of diffractive metasurfaces faces challenges including strong nonlinearity in the structure–optical response mapping, susceptibility to local optima, and high computational cost. Method: This work introduces, for the first time, conditional diffusion models into metasurface inverse design, enabling continuous wavelength-band generation and integrating dual initialization paradigms—direct sampling and gradient-based optimization. A physics-consistent dataset is constructed using Rigorous Coupled-Wave Analysis (RCWA), and RCWA-guided posterior sampling is jointly employed with gradient optimization for efficient co-solution. Contribution/Results: The method designs uniform beam splitters and polarization beam splitters within 30 minutes, achieving low reconstruction error and strong generalization across wavelengths and target specifications. All code and datasets are publicly released.

Metamaterial design faces fundamental challenges including high geometric complexity and strongly nonlinear structure–property mappings. To address these, we propose the first end-to-end generative design framework tailored for metamaterials. Our method introduces MetaDSL—a domain-specific language enabling unified, human-readable and machine-parsable design representation—complements it with MetaDB, an open-source database containing over 150,000 samples, and establishes MetaBench, a comprehensive benchmark covering structural reconstruction, inverse design, and property prediction. The framework integrates parametric modeling, 3D electromagnetic simulation, multi-view rendering, and fine-tuned vision-language models, and features a CAD-inspired interactive interface. Experiments establish the first baseline performance of vision-language models in metamaterial design, achieve— for the first time—joint modeling of structure, representation, and property, and demonstrate the framework’s efficacy in efficient inverse design and physically consistent generation.

To address the high computational cost of iterative optimization and the ill-posedness (non-uniqueness) of inverse electromagnetic design, this paper proposes an end-to-end generative approach based on conditional diffusion models that directly maps target microwave scattering cross-section spectra to structured dielectric geometries. We introduce a feature-linearly modulated 1D U-Net architecture to model the highly nonlinear spectrum-to-geometry mapping, while explicitly capturing solution-space diversity via the stochastic sampling mechanism inherent to diffusion processes—thereby mitigating non-uniqueness. On unseen targets, our method achieves a median relative error below 19% (best case: 1.39%), reduces design time from hours to seconds, and significantly outperforms conventional optimization methods such as CMA-ES. The framework is efficient, diverse in solution generation, and scalable to broader electromagnetic inverse design tasks.

To address coverage and efficiency bottlenecks induced by wireless channel dynamics, this work systematically investigates three reconfigurable intelligent surface (RIS) paradigms—two-dimensional RIS, three-dimensional stacked intelligent metasurfaces (SIM), and flexible intelligent metasurfaces (FIM)—for intelligent propagation environment control. We first comparatively analyze their underlying physical mechanisms and network gain models. We propose a novel wave-domain analog computing architecture for SIM and introduce a deformation-driven diversity gain principle for FIM. Leveraging electromagnetic metamaterial design, programmable RF arrays, and multimodal flexible actuation, we jointly optimize full-wave electromagnetic simulations and channel modeling. Experimental results demonstrate: (i) a 12 dB SNR improvement for SIM in the millimeter-wave band; (ii) a 3.8 dB diversity gain for FIM under time-varying channels; and (iii) the construction and validation of a four-dimensional deployment framework for conventional RIS. This work establishes theoretical foundations and practical implementation pathways for intelligent wireless environment reconstruction.

Existing reconfigurable electromagnetic structures (REMS)—including reconfigurable intelligent surfaces (RIS) and reconfigurable reflectarrays (RRAs)—suffer from a fundamental trade-off between computational efficiency and physical fidelity in modeling, forcing control algorithms to rely on oversimplified, inaccurate surrogate models. This work proposes a unified, physics-informed modeling framework that integrates circuit-theoretic descriptions with far-field electromagnetic interaction characterization. Leveraging only a single full-wave simulation, the framework enables rapid prediction of the complete far-field radiation response for arbitrary tunable element configurations. It rigorously satisfies Maxwell’s equations and consistently incorporates mutual coupling, polarization effects, dielectric/conductor losses, nonreciprocal responses, and thermal noise. The resulting model achieves accuracy comparable to full-wave simulation while accelerating computation by over two orders of magnitude. Furthermore, it enables the first real-time, high-fidelity multi-user beam and null synthesis algorithm capable of joint beamforming and null-steering optimization.

This work addresses the limitations of conventional metasurface inverse design, which relies on time-consuming full-wave simulations and struggles to balance spectral accuracy with manufacturability, while existing generative approaches often lack precise conditional control and yield impractical structures. To overcome these challenges, the study introduces a physics-guided conditional diffusion model that integrates target reflection spectra via Feature-wise Linear Modulation (FiLM) and embeds a pre-trained electromagnetic surrogate model to impose spectrum-level physical regularization. The proposed method enables efficient, high-fidelity generation of manufacturable metasurfaces tailored to specific absorption performance, supports one-to-many design for a single target, achieves an average spectral mean squared error of 0.0006 and a band-alignment accuracy of 0.958 across the 2–18 GHz range, requires only ~30 seconds per generation—dramatically accelerating the design cycle from months—and has been experimentally validated.

This work addresses the limitations of conventional metasurface inverse design, which relies on time-consuming full-wave simulations and suffers from insufficient control accuracy and structural diversity in existing generative approaches. The authors propose a generative inverse design framework based on a progressively growing Wasserstein GAN, incorporating feature-wise linear modulation to achieve high-precision spectral conditioning. Physical consistency is ensured through a surrogate model–guided spectral alignment loss, while geometric diversity is enhanced via determinantal point process regularization. Evaluated over the 2–18 GHz band, the method achieves an average mean squared error of 0.0052, a diversity score of 0.8730, a band-alignment accuracy of 0.8533, and an effective design generation rate of 89.57%, significantly outperforming current state-of-the-art techniques.

This work addresses the challenge of achieving high-speed, parallel, and energy-efficient signal processing directly in the electromagnetic wave domain to jointly support communication, sensing, and computing tasks. The authors propose a stacked intelligent metasurface (SIM) physical neural network that, for the first time, systematically integrates the feature extraction capabilities of neural networks with the intrinsic computational advantages of electromagnetic wave propagation. By stacking metasurface layers and modeling wave-domain dynamics, the architecture enables end-to-end trainable signal processing within the physical propagation process, facilitating multimodal functionality on a single hardware platform. Experimental results demonstrate SIM’s versatility across communication, sensing, and computing applications, exhibiting exceptional parallelism, ultra-low latency, and minimal power consumption, thereby establishing a new paradigm for wave-domain native intelligent hardware.

This work addresses the challenge that metasurface inverse design heavily relies on domain experts to construct solver-compatible workflows and that existing language-driven approaches struggle to transfer knowledge across tasks. The authors propose an agent-based framework featuring context-level skill evolution, wherein a large language model–driven coding agent collaborates with a persistently evolving skill library and a physics-simulation–based deterministic evaluator. This enables cross-task, self-evolving optimization without modifying either the underlying model or the solver. Evaluated on in-distribution tasks, the method improves success rate from 38% to 74%, increases the达标 rate (task-completion metric) from 0.510 to 0.870, and reduces the average number of trials to 2.30. Furthermore, it demonstrates preliminary transferability to unseen task families.

Conventional multilayer intelligent metasurfaces struggle to meet practical wireless communication demands due to structural complexity, high computational overhead, and severe interlayer power attenuation. This work proposes two dual-layer stacked architectures—MF-SIM and FILM—that jointly optimize signal processing flexibility and power efficiency directly in the electromagnetic wave domain. By integrating meta-fiber interconnects with a flexible layered design and performing co-optimization within MIMO and multi-user systems, the approach explicitly characterizes the trade-offs among functionality, configuration, and performance. Case studies demonstrate that the proposed architectures substantially reduce both power loss and optimization complexity while preserving excellent signal processing capabilities, thereby offering an efficient and viable pathway for intelligent metasurface deployment in 6G networks.