aiqbench analysis
This research introduces a physics-augmented graph transformer that significantly accelerates patch-antenna design by embedding electromagnetic simulation constraints directly into the neural network's training loop. By using surface currents as an intermediate supervisory signal, the model achieves high-fidelity radiation pattern prediction and inverse design without incurring additional inference costs. The approach demonstrates strong generalization on large-scale benchmarks, effectively bridging the gap between computational physics and scalable machine learning for hardware optimization.
Physics-Augmented Graph Transformers for Patch-Antenna Forward and Inverse Design
arXiv:2610.05004v1 Announce Type: new Abstract: Full-wave electromagnetic (EM) simulation enables accurate patch-antenna analysis but is computationally expensive for large-scale forward prediction and inverse design. We present a mesh-native, physics-augmented graph-learning framework that treats radiation-pattern prediction as signal reconstruction on an irregular surface mesh. For the forward problem, a GPS graph transformer is trained with Physics-Augmented Intermediate Supervision (PAIS), an auxiliary node-level objective that predicts complex surface currents, the physical intermediate linking geometry to radiation. PAIS improves multiple GNN backbones at no inference-time cost, while shuffled-current and non-physical controls show the gain comes from physical correspondence. Direction-conditioned decoding and a differentiable radiation-integral consistency loss further exploit this structure. On an 80,000-sample CST benchmark, GPS+PAIS reaches MSE 0.17 / PSNR 19.67, generalizes to a PCA split, and transfers zero-shot to canonical patches. For inverse design, surrogate-filtered diffusion beats nearest-neighbor retrieval by 32% relative MSE.
Abstract:Full-wave electromagnetic (EM) simulation enables accurate patch-antenna analysis but is computationally expensive for large-scale forward prediction and inverse design. We present a mesh-native, physics-augmented graph-learning framework that treats radiation-pattern prediction as signal reconstruction on an irregular surface mesh. For the forward problem, a GPS graph transformer is trained with Physics-Augmented Intermediate Supervision (PAIS), an auxiliary node-level objective that predicts complex surface currents, the physical intermediate linking geometry to radiation. PAIS improves multiple GNN backbones at no inference-time cost, while shuffled-current and non-physical controls show the gain comes from physical correspondence. Direction-conditioned decoding and a differentiable radiation-integral consistency loss further exploit this structure. On an 80,000-sample CST benchmark, GPS+PAIS reaches MSE 0.17 / PSNR 19.67, generalizes to a PCA split, and transfers zero-shot to canonical patches. For inverse design, surrogate-filtered diffusion beats nearest-neighbor retrieval by 32% relative MSE.
6 pages, 3 figures, 3 tables. Accepted for oral presentation at the 2026 IEEE International Workshop on Machine Learning for Signal Processing (MLSP 2026), Atlanta, USA. Code, dataset and Colab demo: this https URL
Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
arXiv:2610.05004 [cs.LG]
(or arXiv:2610.05004v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.05004 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 192 words
Related coverage
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · Sep 4, 2026
Machine Learning Assisted Inverse Design of Pixelated mmWave Patch Antennas
arXiv · Aug 25, 2026
Lattice-Based Cryptography Explained
Quantum Zeitgeist · Aug 15, 2026
Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers
Hugging Face · Aug 18, 2026