[论文解读] Self-configuring high-speed multi-plane light conversion

J. C. A. Rocha, Unė G. Būtaitė+2 more|arXiv (Cornell University)|Jan 23, 2025

Semiconductor Lasers and Optical Devices被引用 4

一句话总结

该论文展示了一种自配置的自由空间多平面光学变换器（MPLC），通过在每一平面测量传输矩阵实现就地优化，得益于高速 MEMS 相位调制器，能够实现高保真度的模式变换和快速切换。

ABSTRACT

Multi-plane light converters (MPLCs) - also known as linear diffractive neural networks - are an emerging optical technology, capable of converting an orthogonal set of optical fields into any other orthogonal set via a unitary transformation. MPLC design is a non-linear problem typically solved by optimising a digital model of the optical system. However, inherently high levels of design complexity mean that even a minor mismatch between this digital model and the physically realised MPLC leads to a severe reduction in real-world performance. Here we address this challenge by creating a self-configuring free-space MPLC. Despite the large number of parameters to be optimised (typically tens of thousands or more), our proof-of-principle device converges in minutes using a method in which light only needs to be transmitted in one direction through the MPLC. Two innovations make this possible. Firstly, we devise an in-situ optimisation algorithm combining wavefront shaping with the principles of wavefront matching that would conventionally be used to inverse-design MPLCs offline in simulation. Secondly, we introduce a new MPLC platform incorporating a microelectromechanical system (MEMS) phase-only light modulator - allowing rapid MPLC switching at up to kiloHertz rates. Our scheme automatically accounts for the physical characteristics of all system components and absorbs any unknown misalignments and aberrations into the final design. We demonstrate self-configured MPLCs capable of mapping random orthogonal speckle input fields to well-defined Laguerre-Gaussian and Hermite-Gaussian output modes, as well as universal mode sorters. Our work paves the way towards large-scale ultra-high-fidelity fast-switching MPLCs and diffractive neural networks, which promises to unlock new applications in areas ranging from optical communications to optical computing and imaging.

研究动机与目标

• Motivate high-dimensional spatial mode transformations beyond single-pass SLMs.
• Develop an in-situ optimization protocol to calibrate MPLCs despite misalignments and aberrations.
• Demonstrate high-fidelity single- and multi-mode transformations including mode sorters.
• Showcase a fast-switching MPLC platform enabling rapid convergence and switching.

提出的方法

• Treat MPLC as a complex optical medium with a transmission matrix (TM) between planes.
• Measure plane-to-output TMs for orthogonal input modes and compute complex spatial filters.
• Compute phase updates for each plane using a forward TM inverse (s_m = T'_m^{-1} v) and convert to phase masks.
• Iterate plane-by-plane updates (coordinate descent) until convergence of output fidelity.
• Extend to N input modes by summing mode-specific filters (φ_m = arg[R * sum_n s_m^n]).
• Implement with a MEMS-based Phase-only Light Modulator (PLM) to enable fast switching (up to kHz) and forward-only optimization to avoid bidirectional back-propagation.

实验结果

研究问题

• RQ1 Can a free-space MPLC be self-configured in-situ to achieve arbitrary mode mappings without a precise digital model?
• RQ2 What is the convergence behavior and fidelity achievable for single and multi-mode MPLCs using TM-based in-situ optimization?
• RQ3 How does a MEMS-based PLM enable high-speed MPLC reconfigurability and switching rates?
• RQ4 What are the practical fidelities and cross-talk levels when sorting or reshaping multiple orthogonal modes in real experiments?

主要发现

MPLC type	Figure no.	Inputs (N)	Samples (P)	Opt. params.	Tot. TMs	Opt. configs.	f (Hz)	d_TM (s)	d_mask (s)	t_opt (min)	Proj. t_opt (s)
Speckle to LG	1(b)	1	4096	16384	20	89,000	720	1.5	3	4	9
Speckle to HG	1(c)	3	4096	16384	80	354,000	720	1.5	7	13	27
HG sorter	4(a-e)	10	4096	16384	220	970,000	720	1.5	6	7	47	88
Speckle sorter	4(f-j)	7	8100	32400	160	1,225,000	720	10	15	64	122
Speckle sorter (Supp.)	Supp.	5	4096	16384	120	531,000	1440	1	7	10	44

• The self-configuring MPLC converges in minutes even with tens of thousands of parameters (e.g., up to 32,400 in experiments).
• Single-mode reshaping (speckle to LG) achieved fidelity up to 0.95 after 40 mask updates.
• Multi-mode reshaping to HG modes achieved fidelities of 0.87–0.92 for three input modes.
• Demonstrated universal mode sorters with average cross-talk around -21 dB per channel for HG mode sorter and -15 to -18 dB for speckle sorters depending on configuration.
• A 4-plane MPLC can sort 10 HG modes into a triangular-output channel layout with quantified cross-talk and convergence behavior.
• The platform supports fast switching up to 1.44 kHz (with current electronics) and projections to 10 kHz with next-generation MEMS PLMs.