Neural model predicts multi-port system responses from limited data
Neural Modal Decomposition: Architectural Priors from Observables
Machine Learning
Summary
Many engineered devices like radio-frequency cavities or photonic chips can be described by how they respond to signals at their ports. Predicting these responses normally requires complex calculations that grow quickly as the number of ports increases. The authors developed a neural network method that learns the underlying resonant patterns of these devices by looking only at the overall system responses. This method can then predict the behavior of devices with many ports, having only been trained on devices with fewer ports.
What this means in practice
- •For rf engineers: Predict multi-port electromagnetic device responses from limited training data to speed up design cycles.
- •For photonics designers: Model photonic device resonant behavior across varying port configurations using a single trained neural model.
Authors
Juho Park, Kaushik Sengupta
Abstract
Many engineering building blocks behave as multi-port linear time-invariant systems. RF cavities, photonic devices, and superconducting quantum chips, despite their different underlying physics, all share a common mathematical structure for their port-level response. Each entry of the response matrix is a sum of contributions from a small number of intrinsic resonant modes, the pole-residue form. A model capable of predicting such responses for arbitrary geometries and arbitrary port configurations, while simultaneously extracting the underlying eigenmode structure, would therefore establish a foundational design principle spanning all these domains. We propose a neural framework that learns this modal decomposition end-to-end, supervised only by system-level observables and without supervising the modal parameters themselves. The architecture decomposes into a port-independent pole predictor and two port-dependent coupling predictors whose outputs are combined entry-wise, separating intrinsic from port-dependent features. This factorization yields a single trained model that generalizes to port counts unseen during training, dissolving the $\mathcal{O}(N^2)$ scaling barrier of direct regression. Despite no modal supervision, the freely-parameterized poles converge to physically meaningful eigenmodes, verified by cross-validation against the AAA rational approximation algorithm. We instantiate the framework in radio-frequency electromagnetic surrogate modeling. A model trained only on 2-port data accurately predicts $N$-port responses unseen during training.