A team of Chinese researchers has developed an AI-based modeling approach that revolutionizes the prediction of complex nonlinear dynamics in Kerr resonators. By leveraging recurrent neural networks (RNNs)—specifically gated recurrent units (GRUs)—and a hybrid convolutional neural network (CNN)-GRU model for complex scenarios, the team achieved nearly 20x faster simulations than traditional methods, while maintaining high accuracy. The work paves the way for faster design of next-generation optical systems, from optical memories to all-optical computers.

Pressure sensors are essential for a wide range of applications, including health monitoring, industrial diagnostics, etc. However, achieving both high sensitivity and mechanical ability to withstand high pressure in a single material remains a significant challenge. This study introduces a high-performance cellulose hydrogel inspired by the biomimetic layered porous structure of human skin. The hydrogel features a novel design composed of a soft layer with large macropores and a hard layer with small micropores, each of which contribute uniquely to its pressure-sensing capabilities. The macropores in the soft part facilitate significant deformation and charge accumulation, providing exceptional sensitivity to low pressures. In contrast, the microporous structure in the hard part enhances pressure range, ensuring support under high pressures and preventing structural failure. The performance of hydrogel is further optimized through ion introduction, which improves its conductivity, and as well the sensitivity. The sensor demonstrated a high sensitivity of 1622 kPa⁻¹, a detection range up to 160 kPa, excellent conductivity of 4.01 S m⁻¹, rapid response time of 33 ms, and a low detection limit of 1.6 Pa, outperforming most existing cellulose-based sensors. This innovative hierarchically porous architecture not only enhances the pressure-sensing performance but also offers a simple and effective approach for utilizing natural polymers in sensing technologies. The cellulose hydrogel demonstrates significant potential in both health monitoring and industrial applications, providing a sensitive, durable, and versatile solution for pressure sensing.

Until now, AI services based on Large Language Models (LLMs) have mostly relied on expensive data center GPUs. This has resulted in high operational costs and created a significant barrier to entry for utilizing AI technology. A research team at KAIST has developed a technology that reduces reliance on expensive data center GPUs by utilizing affordable, everyday GPUs to provide AI services at a much lower cost.

Researchers at the Dalian Institute of Chemical Physics have designed a rhodium catalyst whose microenvironment is tuned by both sulfur and phosphine ligands, based on an industrial single-site Rh₁/POPs catalyst. The new “single-site” catalyst hydroformylates propylene and higher olefins up to twice as fast as the current benchmark, while maintaining high selectivity and stability. The study explains how a carefully controlled amount of sulfur can switch from poisoning the catalyst to promoting its performance.

The PtRu(111)-alkali interface demonstrates superior hydrogen oxidation activity owing to reduced solvent reorganization energy, enabled by the synergistic matching of reaction coordinates, favorable water orientation, and an optimized hydrogen bond network.

A new study published in Engineering reveals that NSUN2, a key protein involved in RNA modification, significantly contributes to cardiac hypertrophy and heart failure by activating the LARP1–GATA4 axis. This research not only deepens our understanding of the molecular mechanisms underlying heart failure but also identifies NSUN2 as a potential therapeutic target for preventing and treating cardiac diseases.