Large-scale nonlinear optical computing with incoherent light via linear diffractive systems

2026-04-01 · Source: cs.NE updates on arXiv.org · Field: Science & Research — Physical Sciences & Chemistry, Engineering & Applied Sciences, Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

A new framework demonstrates that linear diffractive optical systems can perform large-scale nonlinear function approximation using spatially incoherent or partially coherent light. This approach addresses the challenge of weak optical material nonlinearity and high intensity requirements typically associated with optical nonlinear computation. By preceding the diffractive processor with intensity-only input encoding, the system can execute up to one million distinct nonlinear functions in a single forward pass, with outputs spatially multiplexed and read by densely packed detectors. The accuracy of this nonlinear function approximation is quantified based on parallelism, number of diffractive layers, and trainable features. A proof-of-concept experiment, utilizing incoherent illumination from an LCD, validated the system through a model-free in situ learning strategy that jointly optimizes diffractive profiles and detector geometry, accounting for hardware imperfections.

Key takeaway

For optical computing engineers designing high-throughput systems, this research indicates that diffractive processors offer a viable path to massively parallel nonlinear computation, even with incoherent light sources. You should consider integrating intensity-only input encoding with optimized diffractive layers to overcome traditional material nonlinearity limitations, potentially simplifying system design and reducing power requirements for advanced optical processors.

Key insights

Linear diffractive optics can achieve large-scale nonlinear computation with incoherent light via intensity-only input encoding.

Principles

• Linear optics can approximate nonlinear functions.
• Incoherent light enables scalable optical processing.

Method

The method involves intensity-only input encoding followed by an optimized diffractive processor. A model-free in situ learning strategy jointly optimizes diffractive profiles and detector readout geometry for experimental validation.

In practice

• Perform 1M nonlinear functions in one pass.
• Use LCDs for incoherent illumination.
• Optimize diffractive profiles in situ.

Topics

• Diffractive Processors
• Nonlinear Optical Computing
• Incoherent Light
• Function Approximation
• Massively Parallel Computation

Best for: AI Scientist, AI Hardware Engineer, Research Scientist

Related on AIssential

• Photonics: A Foundational Scaling Layer for AI-Era Computing — Photonics is becoming the foundational layer for scaling AI computing by addressing system-level data movement and memory bottlenecks.
• Scalable, Energy-Efficient Optical-Neural Architecture for Multiplexed Deepfake Video Detection — A hybrid optical-neural architecture significantly enhances deepfake detection throughput, energy efficiency, and robustness.
• Optical Implementation of Equilibrium Propagation Using Spatial Photonic Ising Machines — A Spatial Photonic Ising Machine enables energy-efficient optical-digital Equilibrium Propagation for training energy-based networks.
• Penn physicists use light-matter particles to boost AI chip speeds — Exciton-polaritons enable all-light signal switching in semiconductors, potentially enhancing AI chip speed and efficiency.
• Will Photonics Loosen the Bottlenecks in Neural Systems? — Optical neural networks using SOAs offer ultra-low-latency, all-optical computation with inherent noise compression and signal regeneration.
• A tiny light trap could unlock million qubit quantum computers — A new optical cavity design enables efficient, parallel qubit readout, paving the way for million-qubit quantum computers.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.NE updates on arXiv.org.