The evolution of the flow structure around an impulsively stopped sphere is investigated in an incompressible viscous fluid under a transverse magnetic field. The study focuses on the wake structure and drag force over the range of Reynolds numbers $60 \leqslant {\textit{Re}}_{\!D} \leqslant 300$ and $ {\textit{Re}}_{\!D}=1000$, with the interaction parameters $0 \leqslant N \leqslant 10$, where $N$ characterises the strength of the magnetic field. The wake is fully developed before the impulsive stop, after which it moves downstream and interacts with the sphere under the influence of a transverse magnetic field. The complex flow structures are characterised by skin friction lines on the downstream side of the sphere and categorised into five regimes in the $\{N, {\textit{Re}}_{\!D}\}$ phase diagram based on nearly 200 cases. The drag force generally decays over time following the impulsive stop. A drag decomposition model based on the vorticity diffusion scale is proposed, attributing the drag decay to three components: the original Stokes contribution, an inertia correction at high Reynolds numbers and a magnetohydrodynamic (MHD) correction, where the inertia and MHD effects both contribute a temporal power-law decay with an exponent of $-1/6$. Temporal scaling laws of the drag decay are derived by coupling these three different effects, considering flow structures at short and long time scales, as well as the dependence on ${\textit{Re}}_{\!D}$ and $N$. The prediction results are consistent with present simulations. Furthermore, the proposed drag decomposition model is successfully extended to complex vortex flow past a sphere at ${\textit{Re}}_{\!D}=1000$, to an anisotropic ellipsoidal particle and to different magnetic field orientations.

Accurately modelling wind turbine wakes is essential for optimising wind farm performance but remains a persistent challenge. While the dynamic wake meandering (DWM) model captures unsteady wake behaviour, it suffers from near-wake inaccuracies due to empirical closures. We propose a symbolic regression-enhanced DWM (SRDWM) framework that achieves equation-level closure by embedding symbolic expressions for volumetric forcing and boundary terms explicitly into governing equations. These physically consistent expressions are discovered from large-eddy simulations (LES) data using symbolic regression guided by a hierarchical, domain-informed decomposition strategy. A revised wake-added turbulence formulation is further introduced to enhance turbulence intensity predictions. Extensive verification across varying inflows shows that SRDWM accurately reproduces both mean wake characteristics and turbulent dynamics, achieving full spatiotemporal resolution with over three orders of magnitude speed-up compared to LES. The results highlight symbolic regression as a bridge between data and physics, enabling interpretable and generalisable modelling.

African swine fever (ASF) is a highly contagious animal disease caused by African swine fever virus (ASFV). It is listed by the World Organization for Animal Health (WOAH) as an animal disease subject to statutory reporting. ASFV, a large, enveloped double-stranded DNA virus with high genomic complexity, exhibits a case fatality rate of up to 100%, posing a significant threat to the global pig industry and food safety. To date, the absence of a safe commercial ASFV vaccine primarily stems from challenges in identifying immunogenic viral antigens, insufficient characterization of ASFV pathogenesis, and limited understanding of the virus’s immune evasion mechanisms. Here, we review the pathogenic characteristics (morphological structure, clinical symptoms, and epidemiological characteristics), molecular biological characteristics, and infection mechanism of ASFV, as well as the immune response mechanism, vaccine research, and the latest information on ASFV in other areas. This review will be in favour of understanding the current state of knowledge of ASF and developing effective vaccines to control this disease.

With the eastward expansion of the Western Zhou c. 1050 BC, the Jiaodong Peninsula on the north-east coast of modern-day China became part of a large polity. Excavations at Qianzhongzitou, located on this peninsula, are revealing how political control over local populations took place. Here, the authors focus on a sequence of Zhou-period, non-residential platforms, the construction of which signifies new forms of ritual spaces. These types of spaces, also found elsewhere in the region, arguably aided in the state assimilation of local deities, illustrating the critical role that ritual played in political unification of early Chinese states and dynasties.

Fiber-coupled laser pumps with low size, weight and power consumption (SWaP) have become more and more compelling for applications in both industrial and defense applications. This study presents an innovative approach employing the spectral beam combining technique and double-junction laser diode chips to create efficient, high-power, high-brightness fiber-coupled packages. We successfully demonstrated a wavelength-stabilized pump module capable of delivering over 560 W of ex-fiber power with an electro-optical conversion efficiency of 55% from a 135 μm diameter, 0.22 numerical aperture fiber. The specific mass and volume metrics achieved are 0.34 $\mathrm{kg}/\mathrm{kW}$ and 0.23 ${\mathrm{cm}}^3/\mathrm{W}$, respectively. The module exhibits a stabilized spectrum with a 3.6 nm consistent interval of two spectral peaks and a 4.2 nm full width at half maximum across a wide range of operating currents.

GenAI has significant potential to transform the design process, driving efficiency and innovation from ideation to testing. However, its integration into professional design workflows faces a gap: designers often lack control over outcomes due to inconsistent results, limited transparency, and unpredictability. This paper introduces a framework to foster human ownership in GenAI-assisted design. Developed through a mixed- methods approach—including a survey of 21 designers and a workshop with 12 experts from product design and architecture—the framework identifies strategies to enhance ownership. It organizes these strategies into source, interaction, and outcome, and maps them across four design phases: define, ideate, deliver, and test. This framework offers actionable insights for responsibly integrating GenAI tools in design practices.

Shock interactions on a V-shaped blunt leading edge (VBLE) that are commonly encountered at the cowl lip of an inward-turning inlet are investigated at freestream Mach numbers ($ M_\infty$) 3–6. The swept blunt leading edges of the VBLE generate a pair of detached shocks with varying shapes due to the changes in $ M_\infty$ and $L/r$ (i.e. the ratio of the leading-edge length $L$ to the leading-edge blunt radius $r$), which causes intriguing shock interactions at the crotch of the VBLE. Three subtypes of regular reflection (RR) and a Mach reflection (MR) are produced successively with increasing $ M_\infty$ for a given $L/r$, which appear in the opposite order to those with increasing $L/r$ for a given $ M_\infty$. These shock interactions identified in numerical simulations are verified in supersonic and hypersonic wind tunnel experiments. It is demonstrated that the relative position of the shocks is crucial in determining the transitions of shock interactions by varying either $L/r$ or $ M_\infty$. Transition criteria between subtypes of RR and from RR to MR are theoretically established in the parameter space $(M_\infty,L/r)$ by analysing the shock structures, showing good agreement with the numerical and experimental results. Interactions between either immature or fully developed detached shocks are embedded in these criteria. Specifically, the transition criteria asymptotically approach the corresponding critical $ M_\infty$ when $L/r$ is sufficiently large. These transition criteria provide guidelines for improving the design of the cowl lip of an inward-turning inlet in supersonic and hypersonic regimes.

Coherent beam combining (CBC) of laser arrays is increasingly attracting attention for generating free-space structured light, unlocking greater potential in aspects such as power scaling, editing flexibility and high-quality light field creation. However, achieving stable phase locking in a CBC system with massive laser channels still remains a great challenge, especially in the presence of heavy phase noise. Here, we propose an efficient phase-locking method for a laser array with more than 1000 channels by leveraging a deep convolutional neural network for the first time. The key insight is that, by elegantly designing the generation strategy of training samples, the learning burden can be dramatically relieved from the structured data, which enables accurate prediction of the phase distribution. We demonstrate our method in a simulated tiled aperture CBC system with dynamic phase noise and extend it to simultaneously generate orbital angular momentum (OAM) beams with a substantial number of OAM modes.

Screw theory serves as an influential mathematical tool, significantly contributing to mechanical engineering, with particular relevance to mechanism science and robotics. The instantaneous screw and the finite displacement screw have been used to analyse the degree of freedom and perform kinematic analysis of linkage mechanisms with only lower pairs. However, they are not suitable for higher pair mechanisms, which can achieve complex motions with a more concise structure by reasonably designing contact contours, and they possess advantages in some particular areas. Therefore, to improve the adaptability of screw theory, this paper aims to analyse higher kinematic pair (HKP) mechanisms and proposes a method to extend instantaneous screw and finite displacement screw theory. This method can not only analyse the instantaneous degree of freedom of HKP mechanisms but also determine the relationships between the motion variables of HKP mechanisms. Furthermore, this method is applied to calculate the degree of freedom and the relationships between the motion angles in both planar and spatial cam mechanisms, thereby demonstrating its efficiency and advantages.

We prove a genus zero Givental-style mirror theorem for all complete intersections in toric Deligne-Mumford stacks, which provides an explicit slice called big I-function on Givental’s Lagrangian cone for such targets. In particular, we remove a technical assumption called convexity needed in the previous mirror theorem for such complete intersections. In the realm of quasimap theory, our mirror theorem can be viewed as solving the quasimap wall-crossing conjecture for big I-function [13] for these targets. In the proof, we discover a new recursive characterization of the slice on Givental’s Lagrangian cone, which may be of self-independent interests.

Spatial intensity modulation in amplified laser beams, particularly hot spots, critically constrains attainable pulse peak power due to the damage threshold limitations of four-grating compressors. This study demonstrates that the double-smoothing grating compressor (DSGC) configuration effectively suppresses modulation through directional beam smoothing. Our systematic investigation validated the double-smoothing effect through numerical simulations and experimental measurements, with comprehensive spatiotemporal analysis revealing excellent agreement between numerical and practical pulse characteristics. Crucially, the DSGC enables a 1.74 times energy output boost compared to conventional compressors. These findings establish the DSGC as a pivotal advancement for next-generation ultrahigh-power laser systems, providing a viable pathway toward hundreds of PW output through optimized spatial energy redistribution.

Systematic reviews play important roles but manual efforts can be time-consuming given a growing literature. There is a need to use and evaluate automated strategies to accelerate systematic reviews. Here, we comprehensively tested machine learning (ML) models from classical and deep learning model families. We also assessed the performance of prompt engineering via few-shot learning of GPT-3.5 and GPT-4 large language models (LLMs). We further attempted to understand when ML models can help automate screening. These ML models were applied to actual datasets of systematic reviews in education. Results showed that the performance of classical and deep ML models varied widely across datasets, ranging from 1.2 to 75.6% of work saved at 95% recall. LLM prompt engineering produced similarly wide performance variation. We searched for various indicators of whether and how ML screening can help. We discovered that the separability of clusters of relevant versus irrelevant articles in high-dimensional embedding space can strongly predict whether ML screening can help (overall R = 0.81). This simple and generalizable heuristic applied well across datasets and different ML model families. In conclusion, ML screening performance varies tremendously, but researchers and software developers can consider using our cluster separability heuristic in various ways in an ML-assisted screening pipeline.

The safety of human-collaborative operations with robots depends on monitoring the external torque of the robot, in which there are toque sensor-based and torque sensor-free methods. Economically, the classic method for estimating joint external torque is the first-order momentum observer (MOB) based on a physic model without torque sensors. However, uncertainties in the dynamic model, which encompasses parameters identification error and joint friction, affect the torque estimation accuracy. To address this issue, this paper proposes using the backpropagation neural network (BPNN) method to estimate joint external torque without the delicate physical model by utilizing the powerful machine learning ability to handle the uncertainties of the MOB method and improve the accuracy of torque estimation. Using data obtained from the torque sensor to train the BPNN to build up a digital torque model, the trained BPNN can perceive force in practical applications without relying on the torque sensor. In the end, by contrast to the classic first-order MOB, the result demonstrates that BPNN achieves higher estimation accuracy compared to the MOB.

Let E be an elliptic curve defined over $\mathbb {Q}$ with good ordinary reduction at a prime $p\geq 5$ and let F be an imaginary quadratic field. Under appropriate assumptions, we show that the Pontryagin dual of the fine Mordell–Weil group of E over the $\mathbb {Z}_{p}^2$-extension of F is pseudo-null as a module over the Iwasawa algebra of the group $\mathbb {Z}_{p}^2$.

Despite the important role of state-owned enterprises (SOEs) in government policy implementation, there is a lack of research on how SOEs owned by different government entities differ. We draw on an attention-based view (ABV) to understand how central government-owned (called central SOEs) and local government-owned enterprises (called local SOEs) differ in their response to digitalization, a major state objective in China in recent years. The two types of SOEs differ in the foundational feature of attention structure – the rules of the game (as embodied in their different goals, identities, and evaluation of top executives) – as well as important features such as governance structures and resources. These features can trigger more attention in central SOEs to digitalization. Given the interdependence of these features in shaping the structural distribution of attention, we further propose how governance structures and resources can influence strategic attention differently in SOEs with different rules of the game. The arguments are tested using data from all Chinese-listed manufacturing SOEs between 2009 and 2020. The study reveals different responses to national strategy between central and local SOEs due to their distinct attention structures designed by the state. It also extends the ABV and research on corporate digital transformation.