Experimental Demonstration of an Optical Neural PDE Solver via On-Chip PINN Training

2025-01-01

AI Summary

Conventional neural PDE solvers rely heavily on backpropagation, incurring substantial computational overhead and posing significant challenges for hardware integration. Method: This work introduces the first on-chip all-optical Physics-Informed Neural Network (PINN) solver, eliminating gradient-based optimization entirely. It embeds physical constraints of partial differential equations directly into the optical-domain forward pass using a programmable Mach–Zehnder interferometer (MZI) mesh and an on-chip optoelectronic co-training architecture. Contribution/Results: Experimentally demonstrated on a silicon photonic chip, the solver achieves real-time solutions for the wave and heat equations with L₂ errors below 2%. It delivers three orders-of-magnitude higher inference throughput than GPU-accelerated implementations and reduces power consumption by over 95%. This work establishes, for the first time, the feasibility of a backpropagation-free, ultra-low-power, high-throughput all-optical neural PDE solving paradigm, opening a new pathway for photonic hardware acceleration in scientific computing.

Technology Category

• Machine Learning: Hardware-aware ML
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

Partial differential equation (PDE) is an important math tool in science and engineering. This paper experimentally demonstrates an optical neural PDE solver by leveraging the back-propagation-free on-photonic-chip training of physics-informed neural networks.

Problem

• Optical computing
• Partial differential equations
• Neural network training

Innovation

• Optical Neural PDE Solver
• Physics-Informed Neural Network (PINN) Training
• Light-based Computation