用物理启发的分块策略，让光子神经网络训练快1825倍且可实现。
SP2RINT: Spatially-Decoupled Physics-Inspired Progressive Inverse Optimization for Scalable, PDE-Constrained Meta-Optical Neural Network Training
- 将超表面设计拆解为可并行求解的局部块，避免逐次求解偏微分方程。
- 训练速度比传统方法快1825倍，性能接近数字模型。
- 适合需要快速部署可制造光子神经网络的研究与工程应用。
DONNs 利用光传播实现高效的模拟人工智能与信号处理。纳米光子制造和基于超表面的波前调控技术，使跨光谱范围的高容量 DONNs 成为可能。但训练这些系统以确定超表面结构仍具挑战：启发式方法虽快，却过度简化调制，常生成物理不可实现的设计，导致性能下降；仿真-闭环优化虽能生成可实现结构，但计算开销大且无法扩展。为此，我们提出 SP2RINT，一种空间解耦、渐进式的训练框架，将 DONN 训练建模为偏微分方程约束学习问题。先将超表面响应松弛为具有带状结构的自由可训练传递矩阵，再通过交替训练传递矩阵与伴随法逆设计，逐步施加物理约束，避免每轮迭代求解偏微分方程，同时确保最终设计的物理可实现性。为进一步降低运行时间，引入基于场相互作用自然局域性的物理启发式空间解耦逆设计策略，将超表面划分为独立可解的区块，支持可扩展的并行逆设计与系统级校准。在多种 DONN 训练任务中，SP2RINT 实现了与数字方法相当的精度，同时比仿真-闭环方法快 1825 倍。该方法弥合了抽象的 DONN 模型与可制造光子硬件之间的差距，实现了物理可实现的元光学神经系统的高效可扩展训练。代码已开源：https://github.com/ScopeX-ASU/SP2RINT。
原文摘要 · Abstract (English)
DONNs leverage light propagation for efficient analog AI and signal processing. Advances in nanophotonic fabrication and metasurface-based wavefront engineering have opened new pathways to realize high-capacity DONNs across various spectral regimes. Training such DONN systems to determine the metasurface structures remains challenging. Heuristic methods are fast but oversimplify metasurfaces modulation, often resulting in physically unrealizable designs and significant performance degradation. Simulation-in-the-loop optimizes implementable metasurfaces via adjoint methods, but is computationally prohibitive and unscalable. To address these limitations, we propose SP2RINT, a spatially decoupled, progressive training framework that formulates DONN training as a PDE-constrained learning problem. Metasurface responses are first relaxed into freely trainable transfer matrices with a banded structure. We then progressively enforce physical constraints by alternating between transfer matrix training and adjoint-based inverse design, avoiding per-iteration PDE solves while ensuring final physical realizability. To further reduce runtime, we introduce a physics-inspired, spatially decoupled inverse design strategy based on the natural locality of field interactions. This approach partitions the metasurface into independently solvable patches, enabling scalable and parallel inverse design with system-level calibration. Evaluated across diverse DONN training tasks, SP2RINT achieves digital-comparable accuracy while being 1825 times faster than simulation-in-the-loop approaches. By bridging the gap between abstract DONN models and implementable photonic hardware, SP2RINT enables scalable, high-performance training of physically realizable meta-optical neural systems. Our code is available at https://github.com/ScopeX-ASU/SP2RINT
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。