The human gut microbiome is shaped by diverse selective forces that originate from host and environmental factors and it substantially influences health and disease. Whereasthe association of microbial lineages with various health conditions has been shown at different taxonomic levels1-5, the extent to which unifying adaptive mechanisms sort microbial lineages into ecologically differentiated populations remains poorly understood. Here we show that genome-wide selective sweeps are a pervasive mechanism that differentiates bacteria in the microbiome. This mechanism leads to population structures akin to global epidemics across geographically and ethnically diverse human populations. Such sweeps arise when an adaptation allows a clone to outcompete others in its niche followed by rediversification, and they manifest as clusters of closely related genomes on long branches in phylogenetic trees. This structure is revealed by excluding recombination events that mask the clonal descent of the genomes. Indeed, we show that genome-wide sweeps originate under a widerange of recombination rates in at least 66 taxa from 25 bacterial families. Estimated ages of divergence suggest that sweep clusters can spread globally within decades and that this process has occurred throughout human history. Sweep clusters are associated with different host conditions-such as age, colorectal cancer, inflammatory bowel diseases and type 2 diabetes-as an indication of their ecological differentiation. Our results reveal an evolutionary mechanism for the observation of stably inherited strains with differential associations and provide a theoretical foundation for analysing adaptation among microbial populations.

This article explores how simple reinforcement learning algorithms might be implemented by the anatomy of the cerebellum. In doing this, we highlight which anatomical and physiological details are most important for assessing algorithmic fit, and we discuss which algorithm components are easiest to accommodate in a neural system. We describe hypothetical cerebellar implementations of four reinforcement learning algorithms and discuss the anatomical plausibility of the various components required. We show how one of the algorithms can learn to generate short sequences of actions without continuous information on the resulting changes to the environment. We finish with simulations that illustrate the way that the algorithms learn to solve the problem of balancing an inverted pendulum, commonly known as the cart-pole problem. We highlight two physiological features: reward signals and combining information across time, that indicate that some sort of reinforcement learning adaptation may be taking place. We also describe why the commonly used algorithmic feature, an eligibility trace, presents particular problems to implement in known neural anatomy.

Leveraging preference disaggregation for context-dependent adaptive multi-criteria sorting with incomplete information

Three-way control combiner valves (TCCVs) are critical components used in nuclear power plants to regulate the concentration of boron acid for neutron absorption and reactor safety. However, current TCCV designs often suffer from suboptimal control performance and high flow resistance, leading to control deviations and reduced operational efficiency. In this paper, a numerical model based on the standard K–ω turbulence model is established and validated against experimental data to analyze the flow characteristics and local flow resistance of a TCCV. A parametric design method for the throttling windows is proposed, establishing relationships between shape parameters and performance indexes, including control performance and flow resistance. The adaptive non-dominated sorting genetic algorithm (ANSGA-II) is used to optimize the shape parameters of the throttling windows. The optimization results show an improvement in the performance indexes of the TCCV, with the adjustable operating range increasing by 31.0% and the maximum local resistance decreasing by 18.3%. We also introduce the concepts of effective and controllable domains to characterize the inlet backflow phenomena and regulation dead zones, which are crucial for ensuring the reliability and effectiveness of control valves. These findings provide insights for enhancing the design and performance of TCCVs in nuclear power plants.

Dynamic 3D Gaussian splatting (3DGS) extends static 3DGS to render dynamic scenes, enabling AR/VR applications with moving objects. However, implementing dynamic 3DGS on edge devices faces challenges: (1) Loading all Gaussian parameters from DRAM for frustum culling incurs high energy costs. (2) Increased parameters for dynamic scenes elevate sorting latency and energy consumption. (3) Limited on-chip buffer capacity with higher parameters reduces buffer reuse, causing frequent DRAM access. (4) Dynamic 3DGS operations are not readily compatible with digital compute-in-memory (DCIM). These challenges hinder real-time performance and power efficiency on edge devices, leading to reduced battery life or requiring bulky batteries. To tackle these challenges, we propose algorithm-hardware co-design techniques. At the algorithmic level, we introduce three optimizations: (1) DRAM-access reduction frustum culling to lower DRAM access overhead, (2) Adaptive tile grouping to enhance on-chip buffer reuse, and (3) Adaptive interval initialization Bucket-Bitonic sort to reduce sorting latency. At the hardware level, we present a DCIM-friendly computation flow that is evaluated using the measured data from a 16nm DCIM prototype chip. Our experimental results on Large-Scale Real-World Static/Dynamic Datasets demonstrate the ability to achieve high frame rate real-time rendering exceeding 200 frames per second (FPS) with minimal power consumption—merely 0.28 W for static Large-Scale Real-World scenes and 0.63 W for dynamic Large-Scale Real-World scenes. This work successfully addresses the significant challenges of implementing static/dynamic 3DGS technology on resource-constrained edge devices.

Sorting algorithms play a crucial role in computing, but most are designed with rigid structure that are only efficient under certain conditions. Although some sorting algorithms perform well in some circumstances, they do not perform well on some resistant platforms. This study introduces Wall-L Merge Sort, which combines quadratic sorting with a modifiable multi-layer merging approach. By setting a single parameter, L, which determines the number of merge layers, Wall-L Sort shows a transition in the time complexity from O(n2) to without any modification in the unique idea. This degree of freedom enables a broad variety of input sizes to be encompassed and expands to several constraint platforms. The results show that Wall-L Sort and K-way Merge Sort have the built-in ability to handle different situations where other algorithms fail without assistance functions. Wall-L Merge Sort is the only sorting algorithm that combines complexity tuning, cache efficiency, recursion depth control, parallelism, and broad adaptability into one framework. It may not be the best choice for every situation, but its flexibility makes it a good fit for many different platforms, from small embedded systems to big computing systems. The theoretical and empirical evidence in this paper substantiates these advantages.

Abstract The coordinate measuring machine (CMM) is commonly used to measure aero-engine blades with precision, and is an important tool for ensuring blade quality. Research on blade surface reconstruction methods has been highly regarded since the accuracy of surface reconstruction directly impacts manufacturing quality assessment. However, when addressing sparse CMM sampling points, mainstream point-cloud-based surface reconstruction methods frequently fail to achieve ideal reconstruction accuracy. This work proposes a framework for blade surface reconstruction based on geometric features, resulting in precise reconstruction in three steps. Firstly, adaptive sampling planning models are developed based on span theory, which consider local distortions and curvature variations. Secondly, to handle the unordered distribution of measurement points, a method of adaptive sorting combining angle and distance constraints is proposed. Finally, by matching measurement points across different cross-sections, a non-uniform rational B-spline reconstruction (NURBS) model is constructed. Using two blade examples, the proposed framework is demonstrated to be effective. Compared to the state-of-the-art method, the maximum error between the reconstructed surface and the design surface is 0.0235 mm.

As the number of feature dimensions increases, the decision-making space exhibits extensive and discrete characteristics, which poses a severe challenge to multi-objective (MO) evolutionary algorithms when searching for the optimal feature subset. Many existing algorithms face the difficulty of slow convergence speed and may fall into local optimal solutions. This study proposes AF-NSGA-II (an adaptive filtering-nondominated sorting genetic algorithm II), an improved MO evolutionary algorithm for high-dimensional feature selection, in which a novel sparse generation scheme for the solution set and an innovative adaptive crossover mechanism are introduced. This sparse initialization strategy, based on three distinct filter feature selection methods, produces initial solutions closer to the optimal Pareto solution set, which is beneficial for convergence. The adaptive crossover mechanism dynamically selects between geometric crossover operators (fostering convergence) and non-geometric crossover operators (enhancing diversity) based on parent similarity, effectively balancing both aspects and helping the algorithm to escape local optima. The algorithm is compared against six renowned multi-objective evolutionary algorithms across ten complex and publicly available datasets. The comparison results demonstrate the superiority of AF-NSGA-II over other algorithms, as well as its effectiveness in identifying the optimal feature subset.

We present EvoSort, a general-purpose adaptive parallel sorting framework accessible at the Python level. EvoSort employs a Genetic Algorithm (GA) to automatically discover and refine critical parameters, including insertion sort thresholds and algorithm selection (mergesort vs. LSD radix sort). By adapting continuously to input data and system architecture, EvoSort provides a drop-in replacement for standard Python routines like NumPy and Pandas. Experiments on up to 10 10 (10 billion) elements across nine data distributions and two hardware platforms demonstrate that EvoSort consistently outperforms competing methods. Results show speedups of up to 225 × , exemplifying a powerful auto-tuning solution for large-scale data processing.

Abstract This study addresses the planning requirements for submarine cables in regional power grid new energy integration systems, focusing on the path optimization of collector lines for offshore wind farms in complex sea areas. A multi-objective decision-based cable path planning method is proposed. By comprehensively considering multiple factors such as cable installation economics (total cost), total path length, average water depth, and obstacle avoidance, a parametric path model based on Bézier curves was constructed. An improved adaptive Non-Dominated Sorting Genetic Algorithm (NSGA-II) was employed to search for the Pareto optimal solution set. Case validation based on a typical regional offshore wind farm cluster demonstrates that the proposed method significantly reduces cable engineering costs while effectively decreasing path length in deep-water areas and construction risks. It provides systematic decision-making support and technical foundation for the reliable grid integration of wind power in sea areas characterized by multiple islands, reefs, and obstacles.

AS2: Adaptive sorting algorithm selection for heterogeneous workloads and systems

Sorting is a fundamental problem in computer science, critical for database indexing, numerical computations, and large-scale data processing. Traditional sorting algorithms, such as Quicksort and Merge Sort, achieve O(n log n) complexity but suffer from inefficiencies in pivot selection, recursion overhead, and memory usage. This paper introduces Adaptive Quantum-Inspired Sorting Algorithm with Entropy-Based Recursion Control and AI-Optimized Partitioning (AQSort), a novel hybrid algorithm that integrates quantum-inspired pivot selection, entropy-driven recursion control, and AI-assisted partitioning to optimize sorting efficiency. AQSort dynamically adjusts recursion depth based on data entropy, minimizing redundant operations and memory overhead. By leveraging probabilistic pivot selection inspired by Grover’s algorithm and parallel acceleration via SIMD and CUDA, AQSort achieves an average-case complexity of O(n log log n), outperforming traditional sorting techniques in large-scale applications. Benchmark results demonstrate significant improvements in execution time and memory efficiency, making AQSort highly suitable for high-performance computing environments. This research contributes to adaptive sorting methodologies by bridging quantum-inspired computing and classical sorting, paving the way for efficient large-scale data processing.

Abstract Planar single-degree-of-freedom (DOF) mechanisms are widely used in human rehabilitation devices, yet research on spatial single-DOF mechanisms remains limited. Given the complex spatial nature of human motion, this study proposes a multi-mode spatial RSCR mechanism for gait rehabilitation. Multi-mode mechanisms incorporate adjustable parameters, enabling structural adaptation to different task trajectories. A kinematic model is established, followed by a circuit analysis to ensure trajectory synthesis accuracy. Using Gaussian mixture model clustering, a dataset of human gait trajectories is divided into three clusters, with one representative trajectory regressed from each cluster. A two-stage optimization strategy is implemented: an adaptive reference point-based non-dominated sorting genetic algorithm performs multi-objective optimization to determine optimal adjustable parameters (yA and yFc), while GA-BFGS refines these and additional parameters via single-objective optimization. Simulation results show that the mechanism can accurately reproduce the three representative trajectories by adjusting yA and yFc, with fitting errors of 9.76×10−3 m, 3.35×10−2 m, and 1.09×10−2 m, respectively. These results confirm the feasibility of using distributed optimization for the multi-mode design of RSCR mechanisms. The proposed dual-parameter adjustment method significantly enhances trajectory adaptability, achieving subcentimeter precision in some cases. This work explores a viable path for the development of spatial rehabilitation mechanisms and highlights potential advancements in the multi-mode design of gait rehabilitation systems.

Simple SummaryStudying cold stress and adaptation provides theoretical insights for predicting and controlling pests. Temperature influences gut microbiota, which may in turn affect insect cold tolerance, though the specific mechanisms and bacteria involved remain unclear. Using multigenerational cold acclimation and 16S rDNA sequencing in Spodoptera frugiperda, we observed decreased larval mortality and increased pupation rate over generations, indicating cold adaptation. Acclimated adults also survived extreme cold better than controls but with reduced reproductive fitness, suggesting a survival–reproduction trade-off. Antibiotics impaired both fitness and cold tolerance in all lines and disrupted gut microbial balance. Notably, nine genera and eight species were more abundant in acclimated larvae but scarce in controls. These bacteria are potentially crucial for cold adaptation. Our findings elucidate microbiota’s role in insect environmental adaptation and support developing eco-friendly pest strategies.The study of cold stress and adaptability can provide a theoretical basis for predicting and controlling pests. Temperature shapes gut microbiota composition, and gut microbiota may affect insect temperature tolerance. However, the underlying mechanisms and bacteria species involved in insect temperature tolerance through gut microbiota are still poorly known. In this study, using a multigenerational cold-acclimation design and 16S rDNA sequencing, we investigated the possible pattern of cold acclimation and the contribution of gut bacteria in Spodoptera frugiperda. Results show that during cold acclimation, larval mortality decreased and pupation rate increased with the increase of treating generations, exhibiting some sort of cold adaptation. Cold tolerance tests showed that cold-acclimated adults exhibited significantly higher survival rates under extreme cold stress than those of controls. Cold acclimation also leads to the cost of reproductive fitness, indicating some trade-offs between survival and reproduction. Antibiotic treatment significantly decreased fitness and cold tolerance not only in the un-acclimated lines but also in cold-acclimated lines. Biodiversity studies through 16S RNA sequencing suggested that antibiotic ingestion disrupted the homeostasis of gut microbes. Differential analysis at the genus and species levels further showed that there were nine genera and eight species that had remarkably higher abundance in cold-acclimated lines compared with controls. References-based functional annotation revealed that most of these bacteria may play essential roles in the cold adaptation of hosts. These results provide valuable insights into the contribution of microbiota to insect fitness and will be valuable for guiding the future development of sustainable pest management approaches.

Waste management is a global challenge that demands effective solutions, especially in classification and recycling processes. This study presents the development of an Eco Cycle Classifier Deep Neural Network (ECCDN-Net) model based on deep learning for image-based waste classification. The model integrates the DenseNet201 and ResNet18 architectures to improve visual feature extraction and reduce the vanishing gradient problem. The dataset used is TrashNet, which contains 2,527 images across six waste categories. Training was conducted over 50 epochs, utilizing data augmentation and class balancing to address the imbalanced data. Results show that ECCDN-Net achieved a validation accuracy of 87.75% and an average F1-score of 0.88. The confusion matrix reveals that the model performs well in recognizing most classes, although it faces difficulty distinguishing categories with high visual similarity, such as plastic and glass. This research demonstrates that ECCDN-Net effectively provides accurate waste classification and could serve as a promising solution for more adaptive and sustainable automatic waste sorting.

The generation of self-organized phases drives the emergence of states of matter of higher complexity. Herein, we study in situ generated self-assembled systems based on the condensation between different aldehydes and hydrazides in water. The resulting acylhydrazones can self-organize into turbid hydrogels or bigger microcrystals depending on the component substituents. The generation of the organized phases was investigated by nuclear magnetic resonance (NMR) and ultraviolet-visible (UV-vis) spectroscopy as well as by microscopy, rheology and solid-state X-ray analyses. Polar substituents like imidazole rings, carboxylic acids and alcohols still lead to hydrogels due to the high propensity of the hydrophobic aromatic cores to self-assemble. The microcrystalline gels containing acidic and basic groups displayed pH-responsiveness. Such behavior allowed for adaptive scrambling-sorting transitions and sorting selectivity switching within 1 × 2 dynamic covalent libraries driven by self-organization in response to environmental conditions. Moreover, the generation of hydrophobic microenvironments in the self-assembled three-dimensional (3D)-network promoted selective imine formation made of apolar components as a result of the stabilization and protection of the reversible covalent bond from hydrolysis. Thus, the dynamic systems described here exhibit up to five levels of adaptive behaviors governed by self-organization (see conclusions).

Abstrak. Perkembangan teknologi otomasi di industri semakin pesat, mendorong peningkatan efisiensi dan produktivitas di berbagai sektor manufaktur. Penelitian ini menganalisis kinerja sensor solenoid dalam sistem penyortiran barang otomatis, yang memiliki peran krusial dalam algoritma penyortiran. Sensor solenoid dapat mengkonversi energi listrik menjadi energi mekanik, sehingga mengoptimalkan proses penyortiran. Melalui pendekatan eksperimental dan analisis desain, penelitian ini bertujuan mengembangkan sistem penyortiran yang adaptif dan berkelanjutan. Hasil menunjukkan bahwa sistem berbasis solenoid dapat mencapai akurasi hingga 97% dan meningkatkan throughput hingga 22%, membuka peluang untuk inovasi lebih lanjut dalam teknologi otomasi.Abstract. The rapid development of automation technology in the industry is driving increased efficiency and productivity across various manufacturing sectors. This study analyzes the performance of solenoid sensors in automatic sorting systems, which play a crucial role in sorting algorithms. Solenoid sensors can convert electrical energy into mechanical energy, thereby optimizing the sorting process. Through experimental approaches and design analysis, this research aims to develop an adaptive and sustainable sorting system. The results indicate that solenoid-based systems can achieve accuracy of up to 97% and improve throughput by 22%, opening opportunities for further innovation in automation technology.

Nowadays, visual object detection (VOD) is widely used in many AI applications, such as autonomous driving, intelligent robotics, and smart surveillance. As an essential postprocessing step in VOD, nonmaximum suppression (NMS) is employed to generate bounding boxes as detection results. However, NMS is difficult to parallelize and computationally intensive, resulting in high processing latency and energy consumption. To address this issue, this brief proposes an energy-efficient block-based NMS engine that incorporates both algorithm- and hardware-level design techniques to improve processing speed and energy efficiency. These techniques include a block-NMS scheme, an adaptive hybrid sorting architecture (AHSA), and a reconfigurable pipeline-based block-NMS computation architecture. The proposed engine is implemented in 28-nm CMOS technology. Compared with the state-of-the-art designs, it achieves the highest performance (237.97 GOPS) and energy efficiency (8.71 TOPS/W), while delivering results fully equivalent to those of the original NMS.

Environmental sustainability is receiving growing global attention, making the development of low-carbon and green transportation increasingly important. Low-carbon policies offer significant advantages in incentivizing energy conservation and reducing emissions in the transportation sector; however, it is vital to consider the impacts of regional differences on the implementation effect of low-carbon policies. This paper explores multimodal transportation route optimization under a carbon tax policy. First, a bi-objective route optimization model is constructed, with the goal of minimizing total transportation cost and time, while accounting for uncertain demand, fixed departure schedules, and regional differences. Trapezoidal fuzzy numbers are used to represent uncertain demand, and a fuzzy adaptive non-dominated sorting genetic algorithm is designed to solve the bi-objective optimization model. The algorithm is then tested on differently sized networks and on real-world transportation networks in eastern and western China to validate its effectiveness and to assess the impacts of regional differences. The experimental results show the following. (1) When considering transportation tasks at different network scales, the proposed fuzzy adaptive non-dominated sorting genetic algorithm outperforms the NSGA-II algorithm, achieving minimum differences in percentages of cost and time of 9.25% and 7.72%, respectively. (2) For transportation tasks assessed using real-world networks in eastern and western China, an increase in the carbon tax rate significantly affects carbon emissions, costs, and time. The degree of carbon emission reduction varies depending on the development of the regional transportation network. In the more developed eastern region, carbon emissions are reduced by up to 44.17% as the carbon tax rate increases. In the less developed western region, the maximum reduction in carbon emissions is 14.37%. The carbon tax policy has a more limited impact in the western region compared to the eastern one. Therefore, formulating differentiated carbon tax policies based on local conditions is an effective way to maximize the economic and environmental benefits of multimodal transportation.

With the development of digital technologies and the increase in the volume of processed data, the efficiency of sorting algorithms is becoming critical. The paper considers the evolution of sorting algorithms from classical to hybrid methods, in particular Timsort and Introsort, which demonstrate improved time characteristics and stability compared to traditional approaches. Special attention is paid to data preprocessing methods and their impact on performance. A constructive-synthesizing modeling approach is proposed to create adaptive sorting algorithms, which allows combining existing methods and forming new effective algorithms. The use of a genetic algorithm in the design process allows automating the selection of optimal sorting strategies according to the characteristics of the input data. The results obtained confirm the prospects of using a constructive-synthesizing approach to build adaptive sorting algorithms that provide high performance in various conditions.