Optical diffractive neural network-based orbital angular momentum mode fixed-base multiplication/division.

Area of Science:

• Optics and Photonics
• Computational Science
• Artificial Intelligence

Background:

• Optical digital computing offers high-speed, efficient, and precise information processing for AI and communication.
• Key challenges in optical computing include developing effective computational dimensions and precise control for multiplication/division operations.

Purpose of the Study:

• To propose and demonstrate a fixed-base multiplication and division scheme using orbital angular momentum (OAM) modes and optical diffractive neural networks (ODNNs).
• To overcome limitations in optical multiplication/division by utilizing OAM modes as the computational physical dimension.

Main Methods:

• Employed OAM modes as the computational physical dimension within an optical system.
• Utilized ODNNs to perform mode-parallel transformations for numerical shifts, enabling multiplication and division.
• Constructed a 3-layer ODNN to realize fixed-base multiplication and division for n=1, 2, and 3.

Main Results:

• Achieved fixed-base multiplication and division operations with OAM modes, reaching 99% mode purity in outputs.
• Demonstrated dynamic switching between multiplication and division within the same system via phase matrix rotation.
• Successfully implemented the scheme for n=1, 2, and 3.

Conclusions:

• The proposed OAM-mode-based ODNN scheme provides a feasible pathway for fixed-base optical multiplication and division.
• This research offers valuable insights for the development of future digital optical computing architectures.
• The method enhances the capabilities of optical computing for complex arithmetic operations.