Machine learning with knowledge constraints for design optimization of microring resonators as a quantum light source

Area of Science:

• Integrated photonics
• Quantum information science
• Machine learning applications

Background:

• Microring resonators are key for compact non-classical light sources in quantum information processing.
• Current design methods face computational challenges due to complex, high-dimensional parameter spaces.

Purpose of the Study:

• To present a knowledge-integrated machine learning framework for designing squeezed light sources using microring resonators.
• To overcome computational bottlenecks in the design flow of integrated photonic devices.

Main Methods:

• Utilized Bayesian Optimization (BO) for adaptive design exploration.
• Developed a knowledge-integrated machine learning framework tailored for microring resonators.
• Focused on optimizing escape efficiency and on-chip squeezing levels.

Main Results:

• Identified two optimal microring resonator structures within 5 optimization rounds.
• Achieved escape efficiencies exceeding 90% and on-chip squeezing levels of 7.48 dB and 9.86 dB.
• Demonstrated BO's effectiveness in finding optimal designs in over-coupled regions.

Conclusions:

• The developed framework streamlines the design of microring resonator-based squeezed light sources.
• The approach is applicable to silicon nitride resonators and extensible to other materials and structures.
• This method facilitates the design of various integrated photonic components for quantum circuits and optical neural networks.