With the widespread adoption of 5G communication and smart devices, electromagnetic radiation and interference have become critical issues affecting electronic equipment operation and human health. A research team led by Professor Guanglei Wu and Associate Professor Zirui Jia from Qingdao University has developed a novel CoS₂/NiS₂@HCNFs composite material. By employing an innovative strategy that combines bimetallic sulfide heterointerfaces with sulfur vacancies, they have successfully resolved the core challenge of balancing impedance matching and loss capability in traditional electromagnetic wave (EMW) absorbing materials. This material achieves an ultra-wide absorption bandwidth of 10.08 GHz at a minimal thickness of 2.6 mm, representing a 61% performance improvement over single-metal systems. It offers a lightweight, high-efficiency solution for electromagnetic protection, stealth technology, and 5G device shielding.

Straintronics exploits the optical and electrical properties of layered materials through mechanical deformation, offering a facile and efficient approach for device optimization. Layered ZrSe₃, an in-plane anisotropic two-dimensional material, exhibits large strain response on the band structure, has emerged as an ideal candidate next-generation Straintronics.

Reliable defect detection is critical for the safe operation of heavy-haul railways. However, ultrasonic inspection signals are often distorted by complex non-Gaussian noise in small-radius curve sections. A research team developed a physics-informed denoising framework that integrates multi-feature fusion within an ideal binary mask paradigm. The proposed method improves defect echo extraction accuracy and reduces the risk of missed detections.

In a new comprehensive review, researchers from Tsinghua University, Xiaomi EV, Beijing Institute of Technology, and University of Electronic Science and Technology of China examine how these two AI paradigms are being applied across the development, validation, and operation of autonomous driving systems, discuss their strengths and limitations, and further review their emerging synergy for more robust, interpretable, and trustworthy autonomous driving systems. 

Using demographic profiles from surveys in Sydney, Hong Kong, and Nanjing, the authors construct role-playing prompts to test three LLMs on lottery-choice tasks. Three findings emerge. First, GPT models are more risk-averse than human benchmarks, whereas Gemini is more risk-seeking. Second, prompt language affects simulated risk attitudes, with non-English prompts often inducing more conservative choices. Third, LLMs fail to reproduce empirical heterogeneity, producing distributions that are either overly concentrated or unrealistically dispersed.