Self-configuring high-speed multi-plane light conversion

Area of Science:

• Optics and Photonics
• Diffractive Optics
• Neural Network Applications

Background:

• Multi-plane light converters (MPLCs), also known as diffractive neural networks, perform unitary transformations between optical field sets.
• Traditional MPLC design relies on digital models, but performance degrades significantly due to mismatches with physical implementations.
• Complexity in MPLC design leads to sensitivity to misalignments and aberrations.

Purpose of the Study:

• To create a self-configuring MPLC that automatically compensates for physical imperfections.
• To accelerate the convergence of MPLC design and improve performance fidelity.
• To demonstrate the capability of arbitrary optical transformations and universal mode sorting.

Main Methods:

• Introduction of 'multi-plane wavefront shaping' to simultaneously reshape multiple spatial light modes.
• Development of a high-speed MPLC platform utilizing a kHz-rate phase-only light modulator.
• Implementation of a self-configuring design approach that absorbs misalignments and aberrations.

Main Results:

• Achieved self-configuration of MPLCs within minutes, automatically correcting for unknown aberrations and misalignments.
• Demonstrated arbitrary optical transformations and universal mode sorters with ultra-high fidelity.
• Significantly reduced performance degradation caused by mismatches between digital models and physical devices.

Conclusions:

• The developed self-configuring MPLC technology overcomes critical limitations of previous designs.
• Multi-plane wavefront shaping combined with high-speed modulation enables rapid and robust MPLC convergence.
• This approach holds significant potential for advancing optical communications, photonic computing, and imaging technologies.