Search Results (2,659)

Stable isotope analysis is very useful for studying animal nutritional ecology. Feces are the most accessible and non-invasive samples for short-term dietary reconstruction. The giant panda is a special Carnivora species with a highly specialized diet. However, no relevant research has yet explored the reliability of fecal isotopes in wild giant pandas, and the key parameter—fecal isotopic discrimination factors—remains unreported. Thus, we analyzed carbon and nitrogen isotopes of different bamboo species and parts with associated pandas’ feces collected from their foraging sites. The results showed carbon isotopes of shoots were more positive than those of leaves, and the isotopic composition of their feces can effectively reflect seasonal dietary shifts. The calculated fecal carbon discrimination factor was close to zero (Δ¹³C_diet-feces = 0.6 ± 0.8‰), while the nitrogen DFs were significantly positive (Δ¹⁵N_diet-feces = 2.1 ± 1.2‰). The typical metabolic pattern, physiological adaptations and distinctive microbiota of giant pandas contribute to the unique DFs different from those of other herbivores. These findings provide valuable short-term dietary records, key parameters for the application of fecal isotopes to interpret foraging strategies and nutritional status for an endangered species in the wild, expand the application of stable isotope methods in studies to specialized diet animals, and offer a reference for studies utilizing non-invasive materials in other mammals. Full article

The Gonzaga family promoted, in the early seventeenth century, a visual and devotional program aimed at positioning Blessed Luigi Gonzaga as both a spiritual standard-bearer and a political instrument of their dynasty. A comparative analysis of prints, paintings, and liturgical objects from this period has made it possible to reconstruct the iconographic model that shaped subsequent representations of the young religious. The consolidation of the prototype of his likeness was facilitated by his family circle and enabled the dissemination of his charisma and virtues among the nobility and the Society of Jesus across the territories of the Spanish monarchy and the states of the Italian peninsula. This strategy sought to secure the preeminence of the House of Gonzaga through the canonization of a “family saint,” emulating the practices of other Italian dynasties. The article highlights the pressures exerted by the beatus’s relatives on the Jesuits and the papal court in their efforts to accelerate his canonization. The manuscript and printed sources cited underscore that the principal promoters of Luigi’s sanctity were his brother and sister-in-law, Francesco Gonzaga and Bibiana von Pernstein, although their early deaths curtailed broader dissemination initiatives. The couple, together with other members of the Gonzaga–Tana family, relied on narrative, visual propaganda, and political ambition to hasten the canonization of Blessed Luigi—an event that, nonetheless, would be delayed until 1726. In parallel, the circulation of, and devotion to, the earliest images depicting the Jesuit novice’s likeness brings to light the significant role of female agency in the diffusion of his cult. Full article

In practical applications of deep learning-based Synthetic Aperture Radar (SAR) Automatic Target Recognition (ATR) systems, new target categories emerge continuously. This requires the systems to learn incrementally—acquiring new knowledge while retaining previously learned information. To mitigate catastrophic forgetting in Class-Incremental Learning (CIL), this paper proposes a CIL method for SAR ATR named Multi-center Prototype Feature Distribution Reconstruction (MPFR). It has two core components. First, a Multi-scale Hybrid Attention feature extractor is designed. Trained via a feature space optimization strategy, it fuses and extracts discriminative features from both SAR amplitude images and Attribute Scattering Center data, while preserving feature space capacity for new classes. Second, each class is represented by multiple prototypes to capture complex feature distributions. Old class knowledge is retained by modeling their feature distributions through parameterized Gaussian diffusion, alleviating feature confusion in incremental phases. Experiments on public SAR datasets show MPFR achieves superior performance compared to existing approaches, including recent SAR-specific CIL methods. Ablation studies validate each component’s contribution, confirming MPFR’s effectiveness in addressing CIL for SAR ATR without storing historical raw data. Full article

This paper investigates an artificial intelligence (AI)-enhanced visible light communication (VLC) system for body area networks (BANs) based on a hybrid modulation framework that jointly employs quadrature amplitude modulation (QAM) and pulse-position modulation (PPM). The dual-modulation strategy leverages the high spectral efficiency of QAM together with the robustness of PPM to light-emitting diode (LED) nonlinearity and timing distortions, enabling simultaneous high-rate and reliable communication, two essential requirements in BAN applications. To address the nonlinear response of light-emitting diodes and the variability in indoor optical channels, the system integrates classical predistortion techniques with a deep learning equalizer combining convolutional neural network (CNN)–transformer layers. This hybrid model captures both local and long-range distortion patterns, improving symbol reconstruction for both modulation branches. The study further examines pilot-assisted equalization and adaptive bit loading, showing that these strategies strengthen link robustness under diverse channel conditions while enhancing spectral efficiency. The proposed architecture demonstrates that combining dual modulation with AI-driven equalization and adaptive transmission strategies leads to a more resilient and efficient VLC system, well-suited for the dynamic constraints of wearable and body-centric communication environments. Full article

Direct methods provide a model-free approach to solving the crystallographic phase problem and deliver unbiased atomic structures. However, conventional iterative projection algorithms such as Hybrid Input–Output (HIO) face two critical challenges: discontinuous density modification at the protein-solvent boundary and inaccurate molecular envelope reconstruction that fails to account for trapped solvent, particularly in crystals with solvent content approaching the lower limits of direct phasing applicability. We introduced four continuous iterative projection algorithms, including our improved continuous version, which implements smooth density modification at protein-solvent interfaces. To address envelope inaccuracy, we developed a two-step refined reconstruction scheme using sequential large-radius and small-radius Gaussian filters to identify trapped solvent molecules within surface cavities and internal channels. This scheme enhances the performance of both continuous and classical algorithms, including HIO, the difference map, and our improved versions. Benchmarking on 28 protein structures (solvent contents 55–78%, resolutions 1.46–3.2 Å, reported R-factor less than 0.22) showed that the refined envelope scheme increased average success rates of continuous algorithms by 45.7% and classical algorithms by 60.5%. The performance of continuous algorithms and improved classical algorithms proved comparable to the well-established HIO algorithm, forming a top-tier group that exceeded other classical algorithms. Integrating a genetic algorithm co-evolution strategy further enhanced average success rates by approximately 2.5-fold and accelerated convergence through population-wide information sharing. Although the success rate correlates with solvent content, our strategy improved success probability at any given solvent level, extending the practical boundaries of direct methods. The high success rate enabled averaging of multiple independent solutions, which reduced mean phase error by approximately 6.83° and yielded atomic models with backbone root-mean-square deviation (RMSD) typically below 0.5 Å relative to structures reported in the Protein Data Bank (PDB). This work introduces novel algorithms, a refined envelope reconstruction methodology, and an effective optimization strategy with genetic algorithm evolution. The complete framework enhances the capability and reliability of direct methods for phasing protein crystals with limited solvent content and provides a toolkit for addressing challenging cases in structural biology. Full article

Background: Borderline left ventricle represents a heterogeneous spectrum of congenital heart disease for which accurate prediction of suitability for biventricular versus univentricular circulation is often difficult. Serial fetal echocardiography may provide dynamic information to support postnatal decision-making. Case Presentation: We report the case of a fetus diagnosed at 32 weeks’ gestation with a borderline left ventricle, ventricular disproportion, hypoplastic left-sided structures, ductal-dependent systemic circulation, and a non-restrictive ostium secundum atrial septal defect. Serial fetal echocardiographic evaluations demonstrated stable left ventricular dimensions, preserved systolic function, impaired diastolic relaxation, and absence of endomyocardial fibroelastosis. Postnatal echocardiography confirmed hypoplastic aortic arch and coarctation. Following multidisciplinary evaluation, a biventricular repair strategy was selected. At 14 days of life, the patient underwent aortic arch reconstruction and partial atrial septal defect closure with preservation of a small therapeutic interatrial communication. Postoperative evolution was favorable, with progressive left ventricular growth and preserved function. At 2-year follow-up, echocardiography showed normalized mitral and aortic valve z-scores, good left ventricular systolic performance, and no evidence of myocardial fibrosis. Conclusions: This case highlights the value of serial fetal echocardiography in guiding individualized management of borderline left ventricle. Careful assessment of ventricular function and atrial septal physiology may support selection of a biventricular strategy in selected patients and contribute to favorable mid-term outcomes. Full article

Cryo-electron microscopy (cryo-EM) has revolutionized structural virology, enabling routine structure determination at 2–4 Å resolution, with exceptional cases reaching 1.56 Å. The structural diversity of viruses across vertebrate, plant, and insect hosts provides fundamental insights into infection mechanisms, host–pathogen coevolution, and therapeutic target identification. However, analysis of Electron Microscopy Data Bank entries reveals notable disparities in structural coverage: among 11,717 eukaryotic virus structures (excluding bacteriophages), vertebrate viruses constitute 97.6% (n = 11,432) of deposited entries, while plant viruses (1.0%; n = 117) and insect viruses (1.4%; n = 168) remain significantly underrepresented. This bias stems from distinct technical barriers including size limitations for giant viruses exceeding 200 nm, the loss of asymmetric information during symmetry-imposed processing, and the morphological complexity of filamentous and pleomorphic viruses. Each barrier has driven the development of specialized methodological solutions: block-based local refinement overcomes through-focus variations in giant viruses, cryo-electron tomography (cryo-ET) validates and reveals asymmetric features lost in symmetrized reconstructions, and subtomogram averaging enables structural analysis of pleomorphic assemblies. This review synthesizes recent methodological advances, critically evaluates their capacity to address specific technical barriers, and proposes strategies for expanding structural investigations across underrepresented host systems to achieve comprehensive understanding of viral structural biology. Full article

Three-leg AC-DC Vienna rectifiers are susceptible to single open-switch faults, which make DC-link voltage ripple and make three-leg input AC currents distorted and unbalanced. Thus, this paper presents a quick identification method for single open-switch faults based on three-leg fault currents and output capacitors voltage difference. Fault-leg identification depended on zero-plateaus in the three-leg fault currents, whereas fault-side identification was dependent on reconstruction variables obtained through Clark transformation and phase shifting. In order to improve the reliability of the diagnosis system, the harmonic component of capacitor voltage difference is used to realize the missed diagnosis detection and adjust the time threshold automatically. This method requires no additional hardware and is easy to implement. Experimental results verify the effectiveness of this strategy. It is shown that the fault diagnosis method proposed in this paper has the advantages of fast diagnosis speed, high accuracy and good robustness. Full article

Two-dimensional (2D) materials open up exciting possibilities for the study of ion transport behavior for green energy. Here, a simple and effective strategy to fabricate high-conductivity nanofluidic channels based on exfoliated montmorillonite (MTM) nanosheets is proposed. The resource-rich and low-cost layered MTM was first exfoliated into monolayer nanosheets using Exolit OP 550. Subsequently, the MTM nanosheets with Exolit OP 550 were assembled into 2D nanofluidic devices by the layer-by-layer self-assembly method. The results show that Exolit OP 550 exfoliates different types of layered MTM into monolayer nanosheets with uniform contrast and integrity. The reconstructed Na-MTM nanofluidic device has the highest ionic conductance. The ionic conductivity of the Na-MTM 2D nanofluidic device was effectively improved after Li⁺ modification with a higher charge density. After further optimizing the content of Exolit OP 550, the ion conductivity of the MTM nanofluidic device reached 4.66 × 10⁻⁴ S cm⁻¹, which is 55.3% higher than the highest known value among the same nanofluidic devices. Interestingly, this nanofluidic device exhibited a very high sensitivity in detecting water evaporation, which can reach 10⁻¹² S s⁻¹ in resolution. This economically viable strategy may advance the study of low-dimensional ion transport properties in new energy coatings and the design of evaporation detectors. Full article

“Placoderm” and sarcopterygian fishes dominated Devonian waters. Following the end-Devonian crisis, actinopterygians rapidly became major contributors to vertebrate diversity. This transition constitutes the first major diversification event of actinopterygians. Here, we investigate the morphological diversification of Devonian and Carboniferous actinopterygians by quantifying disparity using two-dimensional (2D) geometric morphometrics, which estimates disparity from continuous data and brings geometric information related to the shape changes in several morphological features. In total, 13 landmarks and 203 semi-landmarks were digitized on the body shape reconstructions of 84 species, and 18 landmarks and 50 semi-landmarks were digitized on the reconstructions of the lateral view of the skulls of 86 species. When compared to variations in taxonomic diversity over time, the pattern of body shape variations is congruent, reaching a maximum during the Viséan, but the pattern of skull disparity is not entirely congruent, presenting a first increase during the Late Devonian. Changes in body shape are associated with locomotory properties, while changes in skull shape are associated with functional properties of the feeding apparatus. This pattern strongly suggests the diversification of actinopterygians to be driven by divergence in trophic strategies. This evolutionary radiation seems to be the result of an adaptive response to new ecological opportunities, triggered by big environmental changes in mid-Paleozoic oceans. Full article

Coal gangue sorting is crucial for improving coal quality and reducing environmental pollution; however, traditional methods suffer from resource wastage, high cost, and intensive labor demands. To address these challenges, this paper investigates an image recognition-based coal gangue sorting technique and proposes an improved YOLOv7-tiny detection model tailored for edge GPU devices with limited computational power and memory. YOLOv7-tiny is selected as the baseline due to its balanced performance in detection accuracy, architectural maturity, and deployment stability on edge GPUs. Compared to newer lightweight detectors such as YOLOv8-N and YOLOv6-N, YOLOv7-tiny adopts an ELAN-based modular design, which facilitates structural optimization without relying on anchor-free reconstruction or complex post-training strategies, making it particularly suitable for engineering enhancements in real-time industrial sorting under resource constraints. To tackle the limitations in computing and storage, we first introduce an ELAN-PC feature extraction module based on partial convolution and ELAN. Secondly, a GhostCSP module is proposed by integrating cross-stage aggregation and Ghost bottleneck concepts. These modules replace the original ELAN structures in the backbone and neck networks, significantly reducing floating-point operations (FLOPs) and the number of parameters. Furthermore, the SIoU loss function is adopted to replace the original bounding box loss, enhancing detection accuracy. Experimental results demonstrate that compared with the baseline YOLOv7-tiny, the improved model increases mAP0.5 from 86.9% to 88.7% (a gain of 1.8%), reduces FLOPs from 13.2 G to 9.2 G (a decrease of 30%), and cuts parameters from 6.0 M to 4.3 M (a reduction of 28%). In dynamic sorting tests, the model achieves a coal gangue sorting rate of 82.2% with a misclassification rate of 8.1%, indicating promising practical applicability. Full article

TwinArray Sort is a non-comparison integer sorting algorithm designed for non-negative integers with relatively dense key ranges, offering competitive runtime performance and reduced memory usage relative to other counting-based methods. The algorithm introduces a conditional distinct-array verification mechanism that adapts the reconstruction strategy based on data characteristics while maintaining worst-case time and space complexity of O(n + k). Comprehensive experimental evaluations were conducted on datasets containing up to 10⁸ elements across multiple data distributions, including random, reverse-sorted, nearly sorted, and their unique variants. The results demonstrate consistent performance improvements compared with established algorithms such as Counting Sort, Pigeonhole Sort, MSD Radix Sort, Spreadsort, Flash Sort, Bucket Sort, and Quicksort. TwinArray Sort achieved execution times up to 2.7 times faster and reduced memory usage by up to 50%, with particularly strong performance observed for unique and reverse-sorted datasets. The algorithm exhibits good scalability for large datasets and key ranges, with performance degradation occurring primarily in extreme cases where the key range significantly exceeds the input size due to auxiliary array requirements. These findings indicate that TwinArray Sort is a competitive solution for in-memory sorting in high-performance and distributed computing environments. Future work will focus on optimizing performance for wide key ranges and developing parallel implementations for multi-core and GPU architectures. Full article

Urban heat not only affects thermal comfort but also constrains the sustainable development of cities, underscoring the necessity of understanding the temporal response of land surface temperature (LST) to urban characteristics over time. Most existing studies rely on single-overpass satellite observations or daily averages, failing to capture continuous diurnal variability and the time-dependent influence of different drivers. In this study, we reconstructed seasonal hourly LST series for Beijing using an improved diurnal temperature cycle (DTC) model (${\mathrm{GEM}}_{\eta }$) based on MODIS data, and employed a random forest framework to quantify the relative contributions of natural, urban morphological, and anthropogenic factors throughout the diurnal cycle. Unlike previous studies that rely on traditional DTC models and machine learning for largely static or single-scale assessments, our research provides a unified, time-explicit comparison of LST driver dominance across seasons, hourly diurnal cycles, and urban–rural contexts. The results indicate that persistent urban heat island (UHI) effects occur in all seasons, with the maximum intensity reaching approximately 5.0 °C in summer. Generally, natural factors exert a cooling influence, whereas urban morphology and human activities contribute to warming. More importantly, the dominant drivers show strong temporal dependence: a nature-dominated regime prevails in summer, where vegetation exerts an overwhelming cooling effect. Conversely, during transition seasons and winter, LST variability is governed by a mixed-driven mechanism characterized by an hourly-resolved diurnal handoff, in which the dominant contributors shift hour by hour between surface physical properties and anthropogenic proxies. Our findings challenge the static view of urban heat drivers and provide quantitative evidence for developing time-sensitive and seasonally adaptive mitigation strategies, thereby supporting sustainable urban planning and enhancing climate resilience in megacities. Full article

Defect detection on Printed Circuit Boards (PCBs) constitutes a pivotal component of the quality control system in electronics manufacturing. However, owing to the intricate circuitry structures on PCB surfaces and the characteristics of defects—specifically their minute scale, irregular morphology, and susceptibility to background texture interference—existing generic deep learning models frequently fail to achieve an optimal equilibrium between detection accuracy and inference speed. To address these challenges, this study proposes MDEB-YOLO, a lightweight real-time detection network tailored for PCB micro-defects. First, to enhance the model’s perceptual capability regarding subtle geometric variations along conductive line edges, we designed the Efficient Multi-scale Deformable Attention (EMDA) module within the backbone network. By integrating parallel cross-spatial channel learning with deformable offset networks, this module achieves adaptive extraction of irregular concave–convex defect features while effectively suppressing background noise. Second, to mitigate feature loss of micro-defects during multi-scale transformations, a Bidirectional Residual Multi-scale Feature Pyramid Network (BRM-FPN) is proposed. Utilizing bidirectional weighted paths and residual attention mechanisms, this network facilitates the efficient fusion of multi-view features, significantly enhancing the representation of small targets. Finally, the detection head is reconstructed based on grouped convolution strategies to design the Lightweight Grouped Convolution Head (LGC-Head), which substantially reduces parameter volume and computational complexity while maintaining feature discriminability. The validation results on the PKU-Market-PCB dataset demonstrate that MDEB-YOLO achieves a mean Average Precision (mAP) of 95.9%, an inference speed of 80.6 FPS, and a parameter count of merely 7.11 M. Compared to baseline models, the mAP is improved by 1.5%, while inference speed and parameter efficiency are optimized by 26.5% and 24.5%, respectively; notably, detection accuracy for challenging mouse bite and spur defects increased by 3.7% and 4.0%, respectively. The experimental results confirm that the proposed method outperforms state-of-the-art approaches in both detection accuracy and real-time performance, possessing significant value for industrial applications. Full article

Floor plans are a central representational component of architectural design, operating in close relation to sections, elevations, and three-dimensional reasoning to support the production and understanding of architectural space. In this context, we address the bounded computational task of completing incomplete floor plan representations as a form of early-stage design assistance, rather than treating the floor plan as an isolated architectural object. Within this workflow, being able to automatically complete a floor plan from an unfinished draft is highly valuable because it allows architects to generate preliminary schemes more quickly, streamline early discussions, and reduce the repetitive workload involved in revisions. To meet this need, we present FP-MAE, a self-supervised learning framework designed for floor plan completion. This study proposes three core contributions: (1) We developed FloorplanNet, a dedicated dataset that includes 8000 floorplans consisting of both schematic line drawings and color-coded plans, providing diverse yet consistent examples of residential layouts. (2) On top of this dataset, FP-MAE applies the Masked Autoencoder (MAE) strategy. By deliberately masking sections of a plan and using a lightweight Vision Transformer (ViT) to reconstruct the missing regions, the model learns to capture the global structural patterns of floor plans from limited local information. (3) We evaluated FP-MAE across multiple masking scenarios and compared its performance with state-of-the-art baselines. Beyond controlled experiments, we also tested the model on real sketches produced during the early stages of design projects, which demonstrated its robustness under practical conditions. The results show that FP-MAE can produce complete plans that are both accurate and functionally coherent, even when starting from highly incomplete inputs. FP-MAE is a practical and scalable solution for automated floor plan generation. It can be integrated into design software as a supportive tool to speed up concept development and option exploration, and it also points toward broader opportunities for applying AI in architectural automation. While the current framework operates on two-dimensional plan representations, future extensions may integrate multi-view information such as sections or three-dimensional models to better reflect the relational nature of architectural design representations. Full article