en.Wedoany.com Reported - Chinese researchers have developed the Digital Twin Optical Computing System (DT-OCS), aimed at addressing the research bottleneck of limited access to physical hardware in traditional optical computing systems. The relevant findings were published in Opto-Electronic Advances.
Against the backdrop of growing large-scale artificial intelligence and deep learning workloads, optical computing leverages the interference and diffraction properties of light, demonstrating superior potential in speed, energy efficiency, and parallel processing capabilities compared to traditional electronic systems. However, task development in traditional optical computing systems heavily relies on direct access to physical hardware. Researchers often need to queue for equipment, repeatedly adjust parameters and calibrate errors, and reset the system state after a single user's session, leading to inefficient parallel research and high trial-and-error costs.
To address this bottleneck, the research team constructed a digital twin model that can fully replicate the input-output response of a physical optical computing system under different configuration parameters in software. The team describes DT-OCS as a high-fidelity simulator, allowing researchers to conduct development and validation without directly accessing the physical hardware.
The research team paired DT-OCS with a high-speed optical computing system and a silicon photonic feature computing chip, testing it on image classification and sequential decision-making tasks. Results showed that configuration parameters trained and optimized within the digital twin could be directly transferred to the physical system without additional adjustments. Task performance on the physical hardware closely matched the predictions of the digital model, validating the fidelity and transferability of the approach. Since training and optimization primarily occur in the digital domain, multiple researchers can simultaneously develop different tasks without queuing for shared hardware.
The research team has made the DT-OCS framework and related datasets publicly available, aiming to establish it as "a reproducible, accessible, and scalable software resource for broader sharing and validation." This open approach transforms optical computing from a dedicated resource limited by equipment availability into a more shareable and reproducible research platform. The researchers believe that future optical computing systems should pair physical hardware with open digital models that provide equivalent computational behavior, drawing an analogy to modern transportation relying on both physical roads and digital maps, and noting that mature optical computing platforms will similarly require such a dual structure in the future.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
