Research map / Probabilistic and scientific modeling / Differential equations and PDEs
NeurOLight: A Physics-Agnostic Neural Operator Enabling Parametric Photonic Device Simulation
Jiaqi Gu, Zhengqi Gao, Chenghao Feng, Hanqing Zhu, Ray Chen, Duane S Boning, David Z. Pan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use NeurOLight instead of prior neural models | Photonic device field prediction · Tunable MMI and Etched MMI | test prediction error: 53.8% lower prediction error | Improved |
| use NeurOLight instead of other prediction models | Photonic device field prediction · Etched MMI | prediction error: 42% lower error | Improved |
| use superposition-based mixup | NeurOLight · Multi-source inference on MMI devices | prediction fidelity | Improved |
| use only single-source training examples without mixup (ablation) | NeurOLight · Multi-source inference on MMI devices | generalization | Hurt |
| use superposition-based mixup with reduced training data | NeurOLight · MMI devices with 10% training data | generalization: 10% training data | Improved |
| replace fixed-domain encoding with joint PDE encoding using permittivity and wave prior (ablation) | NeurOLight · Unseen simulation settings | test error | Improved |
| add raw wavelength and domain information to the joint PDE representation (ablation) | NeurOLight · Photonic device simulation | test error | Hurt |
| use raw PDE parameters rather than a compact joint representation (ablation) | NeurOLight · Photonic device simulation | test error | Hurt |
| use inadequate Fourier modes (ablation) | NeurOLight · Terahertz optical field prediction | test error | Hurt |
| use all Fourier modes (ablation) | NeurOLight · Photonic device simulation | test error | Hurt |
| add a parallel convolution path (ablation) | NeurOLight · Photonic device simulation | runtime: 20% runtime penalty | Hurt |
| use a convolutional FFN instead of simple BatchNorm-GELU (ablation) | NeurOLight · Photonic device simulation | test error | Improved |
Later papers that compared with this method
1 later accepted paper reports NeurOLight as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- PACE: Pacing Operator Learning to Accurate Optical Field Simulation for Complicated Photonic Devices (NeurIPS 2024 · better on 3 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PACE: Pacing Operator Learning to Accurate Optical Field Simulation for Complicated Photonic Devices (NeurIPS 2024 · similarity 0.91)
- Physics-Constrained Comprehensive Optical Neural Networks (NeurIPS 2024 · similarity 0.84)
- Layered-Parameter Perturbation for Zeroth-Order Optimization of Optical Neural Networks (AAAI 2025 · similarity 0.83)
- Neural Tangent Knowledge Distillation for Optical Convolutional Networks (NeurIPS 2025 · similarity 0.83)
- Fast Reconstruction of Exact Maxwell Dynamics from Sparse Data (NeurIPS 2026 · similarity 0.82)
- A fast algorithm to simulate nonlinear resistive networks (ICML 2024 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
