To clarify the regulatory role and underlying mechanism of long-chain fatty acids (LCFAs) in the tumor immune microenvironment, this study focused on myeloid-derived suppressor cells (MDSCs) and various mouse tumor models. By employing in vitro and in vivo LCFA exposure and high-LCFA diet interventions, we analyzed the effects of LCFAs on the immunosuppressive function of MDSCs, CD8⁺ T cell-mediated anti-tumor immunity, and tumor progression. The results demonstrated that treating cells with LCFAs both in vitro and in vivo significantly decreased the expression levels of characteristic immunosuppressive genes in monocytic MDSCs (M-MDSCs) and polymorphonuclear MDSCs (PMN-MDSCs). Animal experiments showed that mice fed a high-LCFA diet exhibited delayed tumor progression and prolonged survival across different cancer models. Furthermore, this LCFA-mediated inhibition of M-MDSCs and PMN-MDSCs correlates with enhanced CD8⁺ T antitumor immunity, which is abolished in tumor-bearing nude mice. These results reveals a previously under-recognized role of LCFAs in the tumor immune microenvironment, implicating novel therapeutic strategies for cancer treatment. 

To address the problems that the maximum input charge of the application-specific integrated preamplifier RENA3 limits its measurement range and its optimal input capacitance (2 pF or 9 pF) is too low to match the large equivalent capacitance of particle detectors (tens to hundreds of picofarads), a source follower is proposed as the input stage of RENA3 to overcome the constraints on maximum input charge and optimal input capacitance, thereby expanding its application scope. The proposed method is verified via theoretical analysis, PSpice simulation, and experiments. The experimental results are consistent with the theoretical analysis and numerical simulation, which validates the feasibility of the proposed method.

To address the challenges of sample imbalance, insufficient learning of hard-to-classified samples, and difficulties in small target segmentation in rock fracture detection and segmentation tasks, this paper proposes YOLO11n-seg-RF, an improved lightweight algorithm based on YOLO11n-seg. The proposed method incorporates three key components: 1) a Multi-Receptive Field Joint Enhanced Convolutional Block Attention Module (JECBAM) to enhance feature representation, 2) a Grouped Channel Attention-based Feature Fusion Module (GCAConcat) for effective multi-scale feature integration, and 3) a Simplified Spatial Pyramid Pooling Fast module (SimSPPF) to optimize spatial information aggregation. Additionally, the Focaler-IoU loss function is adopted to improve segmentation accuracy for fine-grained and multi-branch fractures. Experimental results on a custom rock fracture dataset demonstrate superior performance. Detection metrics achieve 88.7% Precision (Box), 77.5% Recall (Box), 84.2% mAP0.5 (Box), and 67.3% mAP0.5:0.95 (Box). Segmentation metrics reach 78.5% Precision (Mask), 68.0% Recall (Mask), 68.0% mAP0.5 (Mask), and 27.0% mAP0.5:0.95 (Mask). The model achieves real-time inference at 144 FPS with only 2.47M parameters, outperforming baseline YOLO11n-seg and other mainstream instance segmentation models. Ablation studies confirm the effectiveness of each proposed module, showing significant improvements in detection/segmentation accuracy while reducing model complexity. Generalization experiments on public datasets (crack-seg and carparts-seg) demonstrate superior cross-domain performance, with mAP0.5 (Box) and mAP0.5 (Mask) exceeding comparative models. Practical validation in mining engineering applications reveals that the algorithm successfully identifies core fractures in borehole samples, enabling rapid estimation of uniaxial compressive strength through established porosity-compressive strength equations derived from fracture ratio analysis and uniaxial compression tests, thereby verifying the practical engineering value. To address the challenges of sample imbalance, insufficient learning of hard-to-classified samples, and difficulties in small target segmentation in rock fracture detection and segmentation tasks, this paper proposes YOLO11n-seg-RF, an improved lightweight algorithm based on YOLO11n-seg. The proposed method incorporates three key components: 1) a Multi-Receptive Field Joint Enhanced Convolutional Block Attention Module (JECBAM) to enhance feature representation, 2) a Grouped Channel Attention-based Feature Fusion Module (GCAConcat) for effective multi-scale feature integration, and 3) a Simplified Spatial Pyramid Pooling Fast module (SimSPPF) to optimize spatial information aggregation. Additionally, the Focaler-IoU loss function is adopted to improve segmentation accuracy for fine-grained and multi-branch fractures. Experimental results on a custom rock fracture dataset demonstrate superior performance. Detection metrics achieve 88.7% Precision (Box), 77.5% Recall (Box), 84.2% mAP0.5 (Box), and 67.3% mAP0.5:0.95 (Box). Segmentation metrics reach 78.5% Precision (Mask), 68.0% Recall (Mask), 68.0% mAP0.5 (Mask), and 27.0% mAP0.5:0.95 (Mask). The model achieves real-time inference at 144 FPS with only 2.47M parameters, outperforming baseline YOLO11n-seg and other mainstream instance segmentation models. Ablation studies confirm the effectiveness of each proposed module, showing significant improvements in detection/segmentation accuracy while reducing model complexity. Generalization experiments on public datasets (crack-seg and carparts-seg) demonstrate superior cross-domain performance, with mAP0.5 (Box) and mAP0.5 (Mask) exceeding comparative models. Practical validation in mining engineering applications reveals that the algorithm successfully identifies core fractures in borehole samples, enabling rapid estimation of uniaxial compressive strength through established porosity-compressive strength equations derived from fracture ratio analysis and uniaxial compression tests, thereby verifying the practical engineering value. 

Reinforcement learning (RL) for autonomous driving faces challenges such as low sample efficiency and convergence difficulties when directly trained in complex scenarios. To address this issue, we propose a cross-simulation agent construction method based on unified data representation and implement the DrivingGym training environment. This method abstracts the input state into three layers: sensor data, vehicle states, and road network information. The control interface unification is achieved across different simulation environments through action adapters. Experiments on common simulation platforms such as CARLA and Metadrive demonstrate that the proposed method can support training with mainstream reinforcement learning frameworks like RLlib and Stable-Baselines3, and enable cross-simulation application of autonomous driving policies from simple to complex scenarios.

To further improve the performance of Tibetan text automatic summarization, the paper proposes a Tibetan text summarization model TiDiffuSum, which integrates CINO-LoRA and Self-condition into the DiffuSum to address issues of insufficient sentence representation, large parameter scale limiting contextual modeling, and high training costs in the Tibetan task. TiDiffuSum model introduces CINO-LoRA mechanism into the sentence encoder to enhance Tibetan semantic representation and significantly reduce the number of training parameters. Additionally, it incorporates Self-condition strategy in the diffusion generation module to strengthen the comprehension and utilization of contextual semantics. Experimental results indicate that TiDiffuSum can effectively reduce the parameter count to 0.45% of the baseline model on the Tibetan summarization dataset (TSUM), and achieves improvements of 1.07, 0.78, and 1.08 in ROUGE-1, ROUGE-2, and ROUGE-L scores, significantly outperforming baseline models.

To assist in reviewing the standardization of sentencing in cases and determining whether a sentence is disproportionately light, reasonable, or disproportionately heavy, two datasets for predicting sentencing standardization are constructed, and a multi-source knowledge aggregation-based model for sentencing standardization prediction is proposed. The model employs a large language model to represent criminal cases as knowledge graphs, from which similar cases are generated. To integrate knowledge from criminal law and similar cases, a multi-source attention module is designed in the model. The multi-source features generated by this module, together with case representations, are used for sentencing standardization prediction. Comparative experimental results on sentencing standardization prediction show that the F1-score of the proposed model is higher than that of other comparative models. Ablation experiments and case analyses demonstrate that criminal law and similar cases play an important auxiliary role in sentencing standardization prediction. 

To extract the complex relationships between drugs and targets, this paper designs a dynamic representation learning algorithm based on deep heterogeneous graph gated convolutional networks (HGGCN) for biological graph modeling and representation learning. The algorithm combines the merits of the gated channels and feature channels to adaptively model interaction patterns of heterogeneous graphs, enhance the topological structure and semantic information of complex networks based on the fusion, and obtain the discriminative representation of drugs and targets for drug-target interaction mining. Experimental results show that the proposed model outperforms existing drug target interaction prediction methods, and is also an accurate drug target association prediction tool, which could provide the technical support for the precision treatment of complex diseases and network information mining.

A control method based on multi-step motion compensation is proposed to enhance the accuracy of path tracking control for car-like robots in the presence of signal time delay. By analyzing the response of existing path tracking control systems and those based on single-step motion compensation to signal time delay, we propose the multi-step motion compensation principle and develop a motion compensation algorithm based on multi-step trajectory prediction. Subsequently, integrating the multi-step motion compensation method with model predictive control (MPC) and proportional-integral-derivative (PID) control algorithms, we develop two distinct path tracking control systems for car-like robots. Joint simulations using MATLAB and Carsim under various working conditions show that the displacement error amplitude of the proposed systems remains below 0.0787 m, indicating high accuracy. When the average signal time delay is 0.2 s, the multi-step method reduces the displacement error amplitude by over 20.61% compared with the single-step method. When the delay is 0.4 s, the reduction exceeds 53.04%. Additionally, the path tracking control method based on multi-step motion compensation is robust against external disturbances like positioning errors, and shortening the control period can further improve its performance.

Two kinds of natural hydraulic fracturing were found in laminated argillaceous dolomite and homogeneous argillaceous dolomite of the Upper Xiaganchaigou Formation in the Shizigou Structure in the southwestern Qaidam Basin (SW Qaidam Basin). To explore the geometric characteristics of networks of natural hydraulic fracturing, the temperature-salinity conditions when hydraulic fracturing occured, the interaction between fluid in fractures and surrounding rocks, and controlling factors of natural hydraulic fracturing under the background of multiple overpressure factors, the two kinds of multilevel fracture networks and inclusions in veins were described at the core scale and thin section scale under single polarized light, orthogonal light and cathodoluminescence. Types, distribution, and contents of minerals in veins and surrounding rocks were further characterized by TESCAN integrated mineral analyzer (TIMA). In addition, the homogenization temperature and salinity of aqueous inclusions in veins were tested. Results show that natural hydraulic fracturing in laminated argillaceous dolomite and homogeneous argillaceous dolomite has similarities and differences. Regarding similarity, both fracture networks consist of three-level veins, including the main vein, the first-level hydraulic veins, and the second-level hydraulic veins, showing orthogonal nets in cores. In addition, anhydrite is the main filling mineral in veins. The differences are as follows. Firstly, the first-level hydraulic veins develop along the lamina directions that intersect the main vein with an angle in laminated argillaceous dolomite, while the first-level hydraulic veins are perpendicular to the main vein in homogeneous argillaceous dolomite. The thicker vein inhibits the development of thinner veins which are nearly parallel to the thicker one. Moreover, thinner first-level and second-level hydraulic veins in homogeneous argillaceous dolomite are mostly filled by calcite and celestite. Secondly, only low-mature heavy oil inclusions develop in the main vein in laminated argillaceous dolomite, while in homogeneous argillaceous dolomite, only aqueous inclusions develop in the main vein, and heavy oil traces exist in intergranular cleavage and secondary fissures. The average homogenization temperature of aqueous inclusions (215℃) is greater than the paleogeothermal range (35–150℃) controlled by the burial depth of the sample. Considering multiple overpressure backgrounds, the tectonic evolution history, thermal history, burial history, and hydrocarbon accumulation history, we propose that natural hydraulic fracturing in the laminated argillaceous dolomite may be attributed to the superposition of tectonic compression, gypsum-anhydrite transformation, and hydrocarbon generation. Hydraulic fracturing in homogeneous argillaceous dolomite is attributed to overpressure caused by upwelling of deep thermal fluid. During the reaction between hydrothermal solution and surrounding dolomite, celestite in thin veins is beneficial to enrichment by extracting Sr2+ from surrounding dolomite. Formation process and geometrical patterns of two kinds of hydraulic fracturing have enlightening significance for enrichment of celestite deposits and the current artificial hydraulic fracturing of shale oil reservoirs in the upper Xiaganchaigou Formation in the SW Qaidam Basin.

This paper reports the first documentation of ultramafic rocks exposed in the northern segment of Wuchuan-Sihui fault in Guangdong. These ultramafic rocks, classified as olivine pyroxenite, occur as a dike within early Late Jurassic (≈160 Ma) granites of the Dadongshan pluton. Zircon U-Pb LA-ICPMS analysis of the olivine pyroxenite yields an age of 160±2 Ma, suggesting that zircons were captured from the surrounding granite during dike intrusion. Additionally, the ⁴⁰Ar/³⁹Ar plateau age of biotite from the olivine pyroxenite is 135±1 Ma, indicating that the dike emplacement occurred at the Early Cretaceous, corresponding to a significant extensional event in South China. Whole-rock major element geochemical analysis reveals enrichment in K and Al, relatively high Ti and P contents, and moderate Mg. SiO₂ content (43.78%–44.80%) of the sample is less than 45%. These features classify the rock as an ultramafic rock, predominantly exhibiting characteristics of alkaline basalt. Whole-rock trace and rare earth element (REE) analyses show enrichment of Nb, Ta, Th, U, LILE (large ion lithophile elements), and LREE (light REE), along with pronounced fractionation between light and heavy REE. The rock displays signatures of high-Nb OIB-type mafic rocks (Ocean Island Basalts), suggesting a metasomatized asthenospheric mantle source modified by subduction components, primarily forming in an intraplate extensional tectonic setting. Combining fault activity, geochronological and geochemical characteristics, we propose that during the Early Cretaceous, the tearing and delamination of the subducted Paleo-Pacific Plate beneath South China created metasomatized asthenospheric mantle sources enriched for OIB-type mafic rocks. Subsequently, intracontinental extension caused crustal thinning, while the lithosphere-penetrating Wuchuan-Sihui Fault facilitated rapid ascent of OIB-type mafic magmas along the fracture system. These magmas ultimately crystallized to form the Jiangwan olivine pyroxenite, with intraplate geochemical affinities. 

To investigate the variations in water quality of the Phase I of the Eastern Route of the South-to-North Water Diversion Project before and after water diversion, based on monitoring data of 18 water quality indicators from 30 monitoring sections along the route, this study systematically analyzed the spatiotemporal distribution characteristics and influencing factors of water quality indicators before and after water diversion using significance analysis, correlation analysis, non-metric multidimensional scaling (NMDS), and positive matrix factorization (PMF) and other methods. The results indicated a significant improvement in water quality after the diversion, with concentrations of chemical oxygen demand (COD_Mn), ammonium nitrogen (NH⁴⁺-N), and total phosphorus (TP) decreasing by 24.7%, 87.9%, and 39.6%, respectively , and the comprehensive water quality index (WQI) improved by 1.3%. The hydrochemical types remained as Ca²⁺-Mg²⁺-Cl⁻-SO₄²⁻ and Ca²⁺-Mg²⁺-HCO₃⁻ without change after diversion. Spatial analysis showed ‘excellent’ WQI levels in both Jiangsu and Shandong sections, whereas the inter-provincial boundary section exhibited relatively high total nitrogen (TN) and total phosphorus (TP) concentrations. Correlation analysis revealed that the significant positive correlations between COD_Mn and nitrate nitrogen (NO₃⁻-N) and total dissolved phosphorus (TDP) shifted from a significant positive correlation to no correlation after the diversion, reflecting the difference in water sources before and after the diversion. Furthermore, PMF analysis indicated that the increase in nutrients concentrations in the inter-provincial boundary section was related to agricultural runoff input, and the rise in phosphorus content in some sections was associated with disturbance of regulation and storage lakes.

Most existing approaches for short-term landslide displacement prediction apply machine learning or deep learning models, which are unable to ensure both excellent generalization and memory capabilities and have limited interpretability. From the perspective of explicit and implicit feature interaction, this paper proposes an Interpretable Explicit Feature Interaction Network (IFIN) and constructs the Explicit and Implicit Feature Interaction-Model (EIFIM) for landslide displacement prediction. Based on the transfer learning method, EIFIM can be trained on deformation pattern dataset including many slopes and then applied to a new single slope to predict its displacement of the next three days based on both static and dynamic factors. Case studies show that prediction performance of EIFIM outperforms common baselines. Moreover, the explainable feature combinations output by the proposed model also indicate its good architectural rationality.

Based on the monitoring data of six representative lakes in Saihanba, this study analyzed the spatial distribution, sources and ecological risks of eight metal ions, followed by exploring the impacts of DOM and its spectral parameters on the distribution of heavy metals. The results show that the average contents of Cr, Fe, Co, As, Cu, Zn, Cd and Pb were 0.09, 82.89, 0.04, 0.88, 0.83, 8.70, 0.03 and 0.41 μg/L, respectively, which were all lower than Class III of the Environmental Quality Standard for Surface Water. The 8 heavy metals displayed high homology, suggesting the natural sources, with certain interference from the interference of human activities. To further investigate the relationship between DOM and heavy metals, three fluorescent components (C1, C2, C3) and spectral parameters of DOM were identified by excitation-emission matrix-parallel factor analysis (EEM‑PARAFAC). DOM can influence the spatial evolution of heavy metals. As, Cd and Cu were positively correlated with DOC. As was positively related to C3 components, Cd and Pb were positively correlated with BIX. Spatially, Taiyang Lake (TYH) showed high DOM content and BIX value, which significantly promoted the enrichment of Cu, As, Cd and Pb. The evaluation results of single factor and Nemero comprehensive pollution index show that the heavy metal pollution level in Saihanba is low, and the risks mainly come from Fe and As. Attention should be paid to the interactions between heavy metals and DOM, along with the potential ecological risks. 

Based on carbon emission data from 38 two-digit industrial sectors across 133 cities from 2009 to 2020, this study reappraises the carbon reduction effects of low-carbon city pilot policies using a difference-in-differences approach and explores potential mechanisms through resource allocation and technological innovation. The results indicate that low-carbon pilot policies significantly reduce carbon intensity at the city-sector level. Optimized land resource allocation serves as an important channel for the policy’s impact, while the level of technological innovation also plays a critical role in the emission reduction process. The carbon reduction effect is more pronounced in cities with greater fiscal pressure or lower dependence on land finance, as well as in medium-low technology, labor-intensive, and capital-intensive sectors. The policy exhibits a stronger emission reduction effect in high-carbon industries. This study helps to unveil the implementation mechanism of low-carbon policies at the city-sector level and provides theoretical insights for exploring localized low-carbon development pathways.

This study focused on Zhangjiakou City, a key water conservation area in the capital region and an ecological barrier in the Beijing-Tianjin-Hebei region. By integrating the Patch-generating Land Use Simulation (PLUS) model and the Geographically Weighted Regression (GWR) model, we quantified land-use changes and the trade-offs/synergies among four ecosystem services (windbreak and sand fixation, food production, soil conservation and water conservation), under different development scenarios from 2000 to 2020 and for the 2030 projection. Based on the trade-off/synergy perspective, we identified the optimal development scenario for Zhangjiakou City. The results indicated that: 1) From 2000 to 2020, the area of forestland in Zhangjiakou City expanded by 40.63%, while cultivated land, grassland, and water bodies declined significantly, intensifying spatial conflicts among ecosystem services, with trade-offs becoming dominant. 2) Four development scenarios, including natural development, cultivated land protection, ecological protection, and comprehensive development, were simulated. The multi-scenario simulations showed that ecosystem services generally exhibited an improving trend across all scenarios. However, conflicts between food production services and ecological regulation services remained prominent. The ecological protection scenario enhanced the synergy between windbreak and sand fixation and water conservation services. The comprehensive development scenario improved the synergy between food production and various ecological regulation services. 3) The comprehensive development scenario, which balanced the coordinated economic development, cultivated land protection and ecological protection, was identified as the optimal scenario for Zhangjiakou City. Future efforts should focus on strengthening comprehensive development planning in Zhangjiakou City to achieve sustainable regional development.

To explore the network structure characteristics and core variables of college students’ stress symptoms and coping strategies, we conducted a survey on stress symptoms and coping strategies among 248 college students using a questionnaire. Network analysis and visualization were performed with R software. The results revealed that meaning making had higher strength, betweenness centrality, and closeness centrality in the network than other nodes, indicating a significant expected influence, suggesting that meaning construction has an adaptive function, which can promote individual stress coping and reduce stress-related symptoms. These findings provide a basis for developing relevant intervention strategies for college students.

This paper systematically summarizes the history, technical principles, and current application status of photon counting imaging detectors based on microchannel plates (MCPs) for extreme ultraviolet (EUV) astronomical observations. The article first introduces the scientific significance and technical challenges of EUV astronomical observations, and explains the key role of photon counting imaging detectors in weak light detection. Secondly, it elaborates on the basic structure of the photocathode material, MCP detector, the electron multiplication mechanism, and the working principles and performance characteristics of optical readout and electronic readout types (such as resistive type, wedge strip type, cursor type, delay line type, cross strip type, etc.). It reviews the types, parameters, and in-orbit performance of MCP-based detectors used in important EUV and ultraviolet and soft X-ray astronomical missions worldwide since the 1970s, summarizes the differences in spatial resolution, count rate, dark count rate, and system complexity among various anode structures. Finally, in combination with international development trends and future mission requirements such as China’s Coronal Explorer for our Sun and nearly Stars (CESS) program, it analyzes the performance requirements of core parameters such as quantum efficiency, spatial resolution, maximum count rate, and dark count rate, and suggests the types of detectors to be focused on.

Based on the theoretical foundations of lava-flow numerical simulation, we first review five classes of models — stochastic, 1D channel flow, cellular automata, depth-averaged shallow water models, and 3D models — highlighting their underlying assumptions, computational efficiency, and applicable scenarios. Stochastic approaches deliver rapid run-times but supply no temporal evolution; 1D channel models quickly quantify heat loss; cellular automata models balance high efficiency with simultaneous temperature and morphological data; depth averaged models are best suited to low-relief terrains; while 3D models provide the most complete dynamical and thermal information at the price of high computational cost. We conclude by outlining future directions: tighter constraints on lava rheology (viscosity and yield strength), more accurate representations of internal velocity and temperature fields, and the adoption of cutting-edge techniques from broader computational fluid dynamics.

This paper provides a comprehensive review of the technological advancements in geospatial artificial intelligence (GeoAI) and its applications in traffic demand forecasting. It systematically analyzes the evolution of GeoAI technologies, with a particular focus on its role in addressing the challenges inherent in the four key stages of traffic demand forecasting: traffic generation, traffic distribution, traffic mode choice, and traffic flow assignment. Through the reconstruction of interdisciplinary frameworks, the decomposition of traffic demand forecasting problems into manageable phases, and the optimization of corresponding strategies, this review highlights how GeoAI integrates spatial representation learning, explicit and implicit spatial modeling, and advanced model evaluation techniques to improve prediction precision and reliability. The application of GeoAI has yielded substantial improvements in the accuracy of traffic forecasts, overcoming the limitations of traditional predictive models that often struggle with the complexity of high-dimensional, multimodal data. By enhancing spatiotemporal prediction capabilities and facilitating a more comprehensive understanding of traffic dynamics, GeoAI has been shown to enhance the robustness of predictive models, enabling more effective traffic management and policy formulation. Looking forward, the paper outlines key directions for future research in GeoAI for traffic demand forecasting. These include the optimization of multimodal traffic data governance, the development of large-scale generative models tailored to the transportation domain, and the establishment of cross-task adaptive learning frameworks. Addressing challenges such as data heterogeneity, traffic system coupling, and the dynamic evolution of spatiotemporal relationships will be crucial for advancing the field. Ultimately, these innovations will support China’s national strategy of building a strong transportation country, delivering key theoretical and practical insights for intelligent transportation systems and sustainable urban mobility.