Researchers have developed a coordinate‑aware Mamba2 framework (CA‑Mamba2) for predicting the remaining useful life of lithium‑ion batteries. By integrating a coordinate feature attention network and a swiGLU‑gated residual network with the state‑space Mamba2 backbone, the model effectively captures both long‑term degradation trends and local fluctuation patterns. On the NASA, Tongji University, and Xi’an Jiaotong University datasets under single‑variable, multivariate, and cross‑dataset generalization settings, the model achieved mean absolute error values below 0.0098, 0.0016, and 0.0083, respectively, with superior computational efficiency and robustness across different prediction starting points.

Researchers in China propose a physics-data co-driven deep neural network. By embedding coupled mode theory into a data-driven deep neural network, this framework achieves rapid and precise prediction of physical parameters in complex coupled resonant systems, deciphering their underlying physical properties in a unified way. As a proof-of-principle experiment, the precise measurement of nanometer-scale displacements is demonstrated. This study not only provides a new method for analyzing the underlying physical mechanisms of coupled resonant systems but also opens up a new avenue for high-precision sensing applications.

- A research team led by Professor Mooseok Jang of the Department of Bio and Brain Engineering has developed a technology that reconstructs a transparent object hidden between two dynamic scattering layers—layers that scatter light in multiple directions—from a single shot.

- The technology is expected to have applications in a wide range of fields, including precision inspection for semiconductors and displays and biomedical imaging.

Researchers have developed a simple visual method for tracking how volatile products move inside a fixed-bed coal pyrolysis reactor. Their findings show that carefully designed internal structures can redirect these vapors, limit undesirable secondary reactions, and improve the production of valuable tar and fuel gas.

Inspired by naturally-occurring mats and balls of seaweed, researchers have created highly-porous, superadhesive meshes that are capable of capturing both large and small microplastic particles – a longstanding challenge in the field. The mesh can clean microplastics from both saltwater and freshwater and is made from sustainable and widely available biopolymers.

A University of Kansas project to design and build two cutting-edge particle detectors for the Compact Muon Solenoid experiment at CERN’s Large Hadron Collider has been approved for funding through the U.S. Department of Energy’s Established Program to Stimulate Competitive Research, or DOE EPSCoR.

The analytical results reveal a clear transition between a regime governed by the snow slab that detaches and another governed by the weak layer, or its interface, along which the fracture propagates. In the presence of a stiffer weak layer or interface, the crack may 'feel' a local stiffness greater than the global one, and therefore appear to propagate at a globally supersonic speed while remaining locally subsonic. This finding at least partially reconciles the different interpretations reported in the avalanche literature and is based on a characteristic energy length previously introduced to explain the two propagation regimes that had already emerged from numerical studies.