A research team led by Prof. Youngik Sohn has developed a dataset-based photonic simulation framework that enables the design of photonic integrated circuits (PICs) up to 1,000 times faster than conventional methods. This breakthrough allows ultrafast and high-precision design of complex multimode optical waveguides, greatly improving the productivity and efficiency of photonic device research in fields such as quantum information processing, optical computing, and high-speed communications.
Photonic integrated circuits are essential platforms widely used in data centers, quantum computing systems, and LiDAR sensors. However, accurately designing multimode structures typically requires thousands of simulations, demanding significant computational resources and time—long considered a major limitation in photonics research and development.
To overcome this challenge, the team implemented a simulation engine that stores pre-computed electromagnetic mode information—such as eigenmodes and inter-modal overlap matrices—in dataset form and utilizes it to rapidly analyze new structures. This approach maintains accuracy comparable to finite-difference time-domain (FDTD) and eigenmode expansion (EME) simulations while achieving speed enhancements of more than 1000×.
Using the proposed method, the team reduced the simulation time for a silicon multimode waveguide bend with a 30-µm effective radius from over 10 hours to under one second. They also demonstrated that thousands of optimization iterations can be completed in just 68 minutes on a standard desktop CPU. These advances lay the technical groundwork for automating the design of complex photonic circuits and performing large-scale optimization using everyday computing environments.
“This research fundamentally improves the slow and inefficient design workflow that has hindered photonic device development,” the team noted. “It offers an accessible environment in which anyone can design and optimize large-scale photonic circuits with ease. We anticipate broad applicability in quantum photonic devices, multimode signal-processing circuits, integrated optical computing, and more.”
The team has released the technology as open-source software (https://github.com/thdwotjd/dataset-based-eme) for researchers in the PIC design community to freely use.
This work was conducted with PhD candidate Jaesung Song as the first author and has been published in Optics Express, a leading international journal in the field of optics by Optica Publishing Group. (Title: Ultra-fast and accurate multimode waveguide design based on a dataset-based eigenmode expansion method, https://doi.org/10.1364/OE.567425)
This research was supported by the National Research Foundation of Korea (NRF), the Institute of Information & Communications Technology Planning & Evaluation (IITP), and Samsung Advanced Institute of Technology, Samsung Electronics(SAIT).
