Related Experiment Video
Updated: Aug 6, 2026
06:42
Generation and Coherent Control of Pulsed Quantum Frequency Combs
Published on: June 8, 2018
All-optical computing towards 100-GHz clock rates
Gordon H Y Li1, Midya Parto2,3,4, Jinhao Ge5
1Department of Applied Physics, California Institute of Technology, Pasadena, 91125, CA, USA.
Light, Science & Applications
|July 17, 2026
Summary
Researchers developed an all-optical computer using recurrent neural networks to overcome electronic bottlenecks. This optical computing approach achieves high clock rates for ultrafast information processing and artificial intelligence applications.
Area of Science:
- Photonics
- Optical Computing
- Artificial Intelligence
Background:
- Conventional processor clock rates have stagnated at ~5 GHz for two decades, limiting real-time processing of ultrafast information.
- Moore's Law and parallel architectures have not overcome this electronic bottleneck for sequential operations.
Purpose of the Study:
- To propose and demonstrate an all-optical computer that circumvents electronic limitations.
- To leverage ultrafast optical operations for high-speed computation.
Main Methods:
- Developed an all-optical recurrent neural network architecture.
- Implemented linear operations, nonlinear functions, and memory entirely in the optical domain.
- Utilized integrated optical microresonators and quantum fluctuations.
Main Results:
- Achieved computational accuracy surpassing linear models at clock rates up to 80 GHz.
- Demonstrated noisy waveform classification and in-situ analysis of soliton states.
- Generated images using generative artificial intelligence based on quantum fluctuations.
Conclusions:
- All-optical computing offers potential beyond digital electronics by utilizing ultrafast optical functions.
- The demonstrated architecture enables high-speed processing and novel AI applications.
- Optical computing can harness quantum fluctuations for generative tasks.
