FieldSeer I: Physics-Guided World Models for Long-Horizon Electromagnetic Dynamics under Partial Observability

2025-12-04

AI Summary

This work addresses the long-horizon electromagnetic field dynamics prediction problem for 2D TE waveguides under partial observability. We propose a geometry-aware world model that integrates structural/material map guidance, source-action conditional generation, and a symmetric logarithmic-domain GRU architecture, enabling closed-loop rollout and physics-domain interactive simulation on FDTD-based benchmarks. Our key contribution is the first demonstration of real-time geometric editing—after prefix observation—without reinitializing data assimilation, substantially enhancing interactivity and design adaptability in digital twin systems. Evaluated across three canonical photonic scenarios, the model consistently outperforms GRU and deterministic baselines across 200 reproducible simulations, achieving markedly improved fidelity in suffix-field prediction. This establishes a new, efficient, and editable dynamic modeling paradigm for inverse design of photonic devices.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We introduce FieldSeer I, a geometry-aware world model that forecasts electromagnetic field dynamics from partial observations in 2-D TE waveguides. The model assimilates a short prefix of observed fields, conditions on a scalar source action and structure/material map, and generates closed-loop rollouts in the physical domain. Training in a symmetric-log domain ensures numerical stability. Evaluated on a reproducible FDTD benchmark (200 unique simulations, structure-wise split), FieldSeer I achieves higher suffix fidelity than GRU and deterministic baselines across three practical settings: (i) software-in-the-loop filtering (64x64, P=80->Q=80), (ii) offline single-file rollouts (80x140, P=240->Q=40), and (iii) offline multi-structure rollouts (80x140, P=180->Q=100). Crucially, it enables edit-after-prefix geometry modifications without re-assimilation. Results demonstrate that geometry-conditioned world models provide a practical path toward interactive digital twins for photonic design.

Problem

• Forecasts electromagnetic field dynamics from partial observations
• Generates closed-loop rollouts in physical domain with source and structure conditions
• Enables geometry modifications without re-assimilation for photonic design

Innovation

• Geometry-aware world model forecasts EM dynamics
• Training in symmetric-log domain ensures stability
• Enables edit-after-prefix geometry modifications without re-assimilation