A CUHK-led team has developed an integrated all-optical signal processor (OSP) that corrects signal distortion directly in light form, reducing latency and energy use compared with conventional digital signal processors (DSP). Built on a silicon photonic chip, it can perform real-time signal equalization in the optical domain and achieve 1.6 Tb/s throughput with under 60 picoseconds latency and extremely low power consumption, while flexibly compensating various transmission impairments, including fiber dispersion, system bandwidth limitation, and nonlinearities. Published in Science, the breakthrough offers a high-bandwidth, low-latency, and energy-efficient solution for high-speed AI data centre interconnects and marks progress toward using light for both data transmission and processing.

A method to interpret artificial intelligence (AI) models used in materials discovery by analyzing their learned features has been developed by researchers from Japan. The method extracts key features from an AI model trained on atomic structural data and optical absorption spectra, and then groups materials with similar structural and spectral characteristics. This approach can be extended to reveal how atomic arrangements influence other material properties, paving the way for more efficient materials design.

Topological phase transition is attracting great attention in condensed matter physics, and its gate-voltage-reversible control is essential for electrically driven topological devices. Recently, Professor Jinxiong Wu’s team at Nankai University achieved the first gate-voltage-reversible topological phase transition in β-Ag₂Te nanodevices, evidenced by a π Berry phase transition. Theoretical attributes this to the Stark effect modulating band structure. Based on this, they built a prototype topological phase transition transistor with on-off ratio exceeding 10⁴, opening a new route for electrically controlled topological devices.