Atomic-level precision meets strong oxidation: GOALL-epitaxy expands horizons in material growth
Peer-Reviewed Publication
Updates every hour. Last Updated: 28-Aug-2025 12:11 ET (28-Aug-2025 16:11 GMT/UTC)
The gigantic-oxidative atomic-layer-by-layer epitaxy (GOALL-Epitaxy) method substantially augments the oxidation power by orders of magnitude, enabling atomically precise construction of artificially designed metastable complex oxide structures.
Astronomers have developed a groundbreaking computer simulation to explore, in unprecedented detail, magnetism and turbulence in the interstellar medium (ISM) — the vast ocean of gas and charged particles that lies between stars in the Milky Way Galaxy. Described in a new study published today in Nature Astronomy, the model is the most powerful to date, requiring the computing capability of the SuperMUC-NG supercomputer at the Leibniz Supercomputing Centre in Germany. It directly challenges our understanding of how magnetized turbulence operates in astrophysical environments.
Researchers report on ionospheric sporadic E layer (Es) activity during the Mother’s Day geomagnetic storm. The team found that the Es layers were significantly enhanced over Southeast Asia, Australia and South Pacific, as well as the eastern Pacific regions during the recovery phase of the geomagnetic storm. They also observed a propagation characteristic in the Es enhancement region wherein the clouds were first detected in high latitudes and detected successively in lower latitudes as time progressed.
Focusing on the spatial cognitive capabilities of large language models (LLMs), researchers led by Prof. Danhuai Guo from Beijing University of Chemical Technology have introduced SRT4LLM, a standardized framework for testing the spatial cognition of LLMs. The framework systematically assesses LLMs across three key dimensions—spatial object types, spatial relations, and prompt engineering strategies—and establishes a unified testing process to support the development of native geographic LLMs.
Point-of-Interest (POI) recommendation is crucial in the recommendation system field. Graph neural networks are used for POI recommendations, but data sparsity affects GNN training. Existing GNN-based methods have two flaws. Firstly, they have coarse granularity for modelling heterogeneity, overlooking complex relationships due to time and space factors. Although some work constructs complex graphs, it may reduce performance by introducing noise. Secondly, they insufficiently consider interaction sparsity issues, with little attention in POI recommendations. To solve these problems, a novel method HestGCL is proposed. It builds a heterogeneous spatio-temporal graph with three node types and three relations to model heterogeneity at a finer granularity. Inspired by self-supervised learning, it uses a cross-view contrastive learning technique, splitting the graph into spatial and temporal views, designing specific graph neural networks, and using node representations for contrastive learning. Experiments on three datasets show that HestGCL outperforms state-of-the-art methods, with relative improvements in Recall@50, and ablation studies prove its effectiveness and robustness.