Generalizing to unseen graph tasks without task-specific supervision remains challenging.Graph Neural Networks (GNNs) are limited by fixed label spaces, while Large Language Models (LLMs) lack structural inductive biases.Recent advances in Large Reasoning Models (LRMs) provide a zero-shot alternative via explicit, long chain-of-thought reasoning.Inspired by this, we propose a GNN-free approach that reformulates graph tasks-node classification, link prediction, and graph classification-as textual reasoning problems solved by LRMs.We introduce the first datasets with detailed reasoning traces for these tasks and develop GRAPH-R1, a reinforcement learning framework that leverages task-specific rethink templates to guide reasoning over linearized graphs.Experiments demonstrate that GRAPH-R1 outperforms state-of-the-art baselines in zero-shot settings, producing interpretable and effective predictions.Our work highlights the promise of explicit reasoning for graph learning and provides new resources for future research.Codes are available at https://github.com/lgybuaa/Graph-R1.

This paper explores the nature of the Go game. Go has been played on different size of the board in its long history where it was born in 2356 BC and played on 9×9 board, and later played on 13×13, 15×15, 17×17, 19×19, and 21×21 board. Self-play experiments using an open-source implementation of AlphaZero, namely LeelaZero, as the artificial intelligence (AI) player, are conducted for data collection, whereas game refinement (GR) measures had been employed for the assessment. An informational analysis indicates that Go game have been searching not only for a harmonic balance between deterministic and stochastic aspects found in the similar board games (Chess and Mahjong) but also the expected fairness when playing Go. Three aspects of the game have been analyzed: the perspective of game designer (using the GR measure), the perspective of Go evolution relative to Chess and Mahjong evolutionary histories, and other perspectives that closely related to the cultural and strategic development of Go game. Based on such aspects, the impact of komi values and the reasons for 19×19 board size to the contemporary Go popularity had been identified.

BACKGROUND: Imaging and image analysis advances are yielding increasingly complete and complicated records of cellular events in tissues and whole embryos. The ability to follow hundreds to thousands of cells at the individual level demands a spatio-temporal data infrastructure: tools to assemble and collate knowledge about development spatially in a manner analogous to geographic information systems (GIS). Just as GIS indexes items or events based on their spatio-temporal or 4D location on the Earth these tools would organize knowledge based on location within the tissues or embryos. Developmental processes are highly context-specific, but the complexity of the 4D environment in which they unfold is a barrier to assembling an understanding of any particular process from diverse sources of information. In the same way that GIS aids the understanding and use of geo-located large data sets, software can, with a proper frame of reference, allow large biological data sets to be understood spatially. Intuitive tools are needed to navigate the spatial structure of complex tissue, collate large data sets and existing knowledge with this spatial structure and help users derive hypotheses about developmental mechanisms. RESULTS: Toward this goal we have developed WormGUIDES, a mobile application that presents a 4D developmental atlas for Caenorhabditis elegans. The WormGUIDES mobile app enables users to navigate a 3D model depicting the nuclear positions of all cells in the developing embryo. The identity of each cell can be queried with a tap, and community databases searched for available information about that cell. Information about ancestry, fate and gene expression can be used to label cells and craft customized visualizations that highlight cells as potential players in an event of interest. Scenes are easily saved, shared and published to other WormGUIDES users. The mobile app is available for Android and iOS platforms. CONCLUSION: WormGUIDES provides an important tool for examining developmental processes and developing mechanistic hypotheses about their control. Critically, it provides the typical end user with an intuitive interface for developing and sharing custom visualizations of developmental processes. Equally important, because users can select cells based on their position and search for information about them, the app also serves as a spatially organized index into the large body of knowledge available to the C. elegans community online. Moreover, the app can be used to create and publish the result of exploration: interactive content that brings other researchers and students directly to the spatio-temporal point of insight. Ultimately the app will incorporate a detailed time lapse record of cell shape, beginning with neurons. This will add the key ability to navigate and understand the developmental events that result in the coordinated and precise emergence of anatomy, particularly the wiring of the nervous system.