Diffractive magic cube network with super-high capacity enabled by mechanical reconfiguration.

Area of Science:

• Optics and Photonics
• Information Technology
• Materials Science

Background:

• Free-space wavefront manipulation is crucial for advanced optical information systems.
• Optical multiplexing and dynamic reconfigurable devices are needed to handle increasing data demands.
• Mechanically reconfigurable systems offer a cost-effective approach but face limitations in channel capacity.

Purpose of the Study:

• To propose a novel diffractive magic cube network (DMCN) for enhancing the multiplexing capacity of mechanically reconfigurable optical systems.
• To overcome limitations of current mechanical transformations and optimization models in achieving higher channel numbers.
• To demonstrate a new paradigm for improving system capacity with low crosstalk.

Main Methods:

• Utilized a diffractive deep neural network (D^2NN) model for joint optimization.
• Optimized channel subsets generated by mechanical operations: permutation, translation, and rotation.
• Formulated an equivalent connectivity law to enhance model scalability.

Main Results:

• Experimentally demonstrated 144-channel holograms, 108-channel single/double focus, and 60-channel single/multi-mode OAM beam generation using diffractive optical elements (DOEs).
• Achieved super-high multiplexing capacity with low crosstalk.
• Validated the effectiveness of the D^2NN model and mechanical operation combinations.

Conclusions:

• The proposed DMCN strategy significantly advances multiplexing capacity in mechanically reconfigurable systems.
• This work provides a novel paradigm for optical information processing, storage, and communication.
• The developed methods pave the way for future advancements in optical technologies like computing and photolithography.