arXiv subjects

Linfeng Jiang

Publications and source records attributed to Linfeng Jiang.

At least 19 recordsLinked to original sources

arXiv2026

Closed-Loop Hybrid Digital Twin Platform for Connected and Automated Vehicle Validation

Comprehensive and efficient validation of connected and automated vehicles (CAVs) is critical prior to real-world deployment. While simulation-based testing offers scalability, existing approaches often lack seamless integration with real vehicles and field data, limiting their fidelity in capturing dynamic, real-world interactions. To bridge this gap, this paper proposes a novel real-time hybrid digital twin platform. Its core innovation lies in the tight coupling of a high-fidelity CARLA-SUMO co-simulation with a physical test site and vehicle via a low-latency Vehicle-to-Everything (V2X) communication link. A custom-developed middleware serves as the critical bridge, synchronizing a real CAV's kinematic state as a shadow vehicle in the simulation and translating virtual control commands into chassis-actuating Controller Area Network (CAN) messages for closed-loop control. Detailed implementation includes using photogrammetry for full-scale asset reconstruction and a cloud-edge collaborative architecture for scalable, multi-user operation. Experimental results demonstrate stable synchronization and effective closed-loop control with low latency, confirming the platform's practicality for multi-scenario CAV verification.

cs.RO↗

arXiv2026

Simulations of Particle-Laden Flows with Large Dispersed-Phase Size Disparities Using Highly Scalable Parallel Adaptive Methods

The numerical simulation of multiphase flows involving dispersed components with large scale disparities, such as the collisions between millimeter-sized bubbles and micron-sized mineral particles in flotation, poses a significant computational challenge. Accurately resolving the thin boundary layers of finite-size objects while tracking massive numbers of small particles within a large turbulent domain is often prohibitively expensive on uniform grids. To address this, we present a parallel scalable computational framework that couples the lattice Boltzmann method with the immersed boundary method on a dynamically adaptive octree grid. A key algorithm is developed for the efficient parallel host-cell searching, which significantly accelerates the tracking of Lagrangian points on distributed unstructured grids. The accuracy and robustness of the code are rigorously validated against canonical benchmarks, including the flow induced by an oscillating cylinder and the sedimentation of a sphere. The framework is applied to the multiscale problem of bubble-particle collisions. In quiescent flow, the simulations accurately capture the hydrodynamic interception mechanism, reproducing the theoretical collision efficiency scaling law proportional to the square of the particle-to-bubble size ratio. Furthermore, the framework is applied to the simulation of fully resolved bubbles interacting with inertial point particles in homogeneous isotropic turbulence.

physics.flu-dyn↗

arXiv2026

The First Challenge on Mobile Real-World Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview

This paper provides a review of the NTIRE 2026 challenge on mobile real-world image super-resolution, highlighting the proposed solutions and the resulting outcomes. The challenge aims to recover high-resolution (HR) images from low-resolution (LR) counterparts generated through unknown degradations with a x4 scaling factor while ensuring the models remain executable on mobile devices. The objective is to develop effective and efficient network designs or solutions that achieve state-of-the-art real-world image super-resolution performance. The track of the challenge evaluates performance using a weighted combination of image quality assessment (IQA) score and speedup ratios. The competition attracted 108 registrants, with 16 teams achieving a valid score in the final ranking. This collaborative effort advances the performance of mobile real-world image super-resolution while offering an in-depth overview of the latest trends in the field.

cs.CV↗

arXiv2026

TCL: Enabling Fast and Efficient Cross-Hardware Tensor Program Optimization via Continual Learning

Deep learning (DL) compilers rely on cost models and auto-tuning to optimize tensor programs for target hardware. However, existing approaches depend on large offline datasets, incurring high collection costs and offering suboptimal transferability across platforms. In this paper, we introduce TCL, a novel efficient and transferable compiler framework for fast tensor program optimization across diverse hardware platforms to address these challenges. Specifically, TCL is built on three core enablers: (1) the RDU Sampler, a data-efficient active learning strategy that selects only 10% of tensor programs by jointly optimizing Representativeness, Diversity, and Uncertainty, substantially reducing data collection costs while maintaining near-original model accuracy; (2) a new Mamba-based cost model that efficiently captures long-range schedule dependencies while achieving a favorable trade-off between prediction accuracy and computational cost through reduced parameterization and lightweight sequence modeling; and (3) a continuous knowledge distillation framework that effectively and progressively transfers knowledge across multiple hardware platforms while avoiding the parameter explosion and data dependency issues typically caused by traditional multi-task learning. Extensive experiments validate the effectiveness of each individual enabler and the holistic TCL framework. When optimizing a range of mainstream DL models on both CPU and GPU platforms, TCL achieves, on average, 16.8x and 12.48x faster tuning time, and 1.20x and 1.13x lower inference latency, respectively, compared to Tenset-MLP.

cs.LG↗

arXiv2025

PocketSR: The Super-Resolution Expert in Your Pocket Mobiles

Real-world image super-resolution (RealSR) aims to enhance the visual quality of in-the-wild images, such as those captured by mobile phones. While existing methods leveraging large generative models demonstrate impressive results, the high computational cost and latency make them impractical for edge deployment. In this paper, we introduce PocketSR, an ultra-lightweight, single-step model that brings generative modeling capabilities to RealSR while maintaining high fidelity. To achieve this, we design LiteED, a highly efficient alternative to the original computationally intensive VAE in SD, reducing parameters by 97.5% while preserving high-quality encoding and decoding. Additionally, we propose online annealing pruning for the U-Net, which progressively shifts generative priors from heavy modules to lightweight counterparts, ensuring effective knowledge transfer and further optimizing efficiency. To mitigate the loss of prior knowledge during pruning, we incorporate a multi-layer feature distillation loss. Through an in-depth analysis of each design component, we provide valuable insights for future research. PocketSR, with a model size of 146M parameters, processes 4K images in just 0.8 seconds, achieving a remarkable speedup over previous methods. Notably, it delivers performance on par with state-of-the-art single-step and even multi-step RealSR models, making it a highly practical solution for edge-device applications.

cs.CV↗

arXiv2025

A predictive model for bubble-particle collisions in turbulence

The modelling of bubble-particle collisions is crucial to improving the efficiency of industrial processes such as froth flotation. Although such systems usually have turbulent flows and the bubbles are typically much larger than the particles, there currently exist no predictive models for this case which consistently include finite-size effects in the interaction with the bubbles as well as inertial effects for the particles simultaneously. As a first step, Jiang and Krug (J. Fluid Mech., vol. 1006, 2025, A19) proposed a frozen turbulence approach which captures the collision rate between finite-size bubbles and inertial particles in homogeneous isotropic turbulence using the bubble slip velocity probability density function measured from simulations as an input. In this study, we further develop this approach into a model where the bubble-particle collision rate can be predicted a priori based on the bubble, particle, and turbulence properties. By comparing the predicted collision rate with simulations of bubbles with Stokes numbers of 2.8 and 6.3, and particles with Stokes numbers ranging from 0.01 to 2 in turbulence with a Taylor Reynolds number of 64, good agreement is found between model and simulations for Froude number $Fr \leq 0.25$. Beyond this range of bubble Stokes number, we propose a criterion for using our model and discuss the model's validity. Evaluating our model at typical flotation parameters indicates that particle inertia and settling effects are usually important. Generally, smaller bubbles, larger particles, and stronger turbulence increase the overall collision rate.

physics.flu-dyn↗

arXiv2024

YOLO-CCA: A Context-Based Approach for Traffic Sign Detection

Traffic sign detection is crucial for improving road safety and advancing autonomous driving technologies. Due to the complexity of driving environments, traffic sign detection frequently encounters a range of challenges, including low resolution, limited feature information, and small object sizes. These challenges significantly hinder the effective extraction of features from traffic signs, resulting in false positives and false negatives in object detection. To address these challenges, it is essential to explore more efficient and accurate approaches for traffic sign detection. This paper proposes a context-based algorithm for traffic sign detection, which utilizes YOLOv7 as the baseline model. Firstly, we propose an adaptive local context feature enhancement (LCFE) module using multi-scale dilation convolution to capture potential relationships between the object and surrounding areas. This module supplements the network with additional local context information. Secondly, we propose a global context feature collection (GCFC) module to extract key location features from the entire image scene as global context information. Finally, we build a Transformer-based context collection augmentation (CCA) module to process the collected local context and global context, which achieves superior multi-level feature fusion results for YOLOv7 without bringing in additional complexity. Extensive experimental studies performed on the Tsinghua-Tencent 100K dataset show that the mAP of our method is 92.1\%. Compared with YOLOv7, our approach improves 3.9\% in mAP, while the amount of parameters is reduced by 2.7M. On the CCTSDB2021 dataset the mAP is improved by 0.9\%. These results show that our approach achieves higher detection accuracy with fewer parameters. The source code is available at \url{https://github.com/zippiest/yolo-cca}.

eess.IV↗

arXiv2024

The effect of gravity on bubble-particle collisions in turbulence

Bubble-particle collisions in turbulence are key to the froth flotation process that is widely employed industrially to separate hydrophobic from hydrophilic materials. In our previous study (Chan et al., J. Fluid Mech., vol. 959, 2023, A6), we elucidated the collision mechanisms and critically reviewed the collision models in the no-gravity limit. In reality, gravity may play a role since ultimately separation is achieved through buoyancy-induced rising of the bubbles. This effect has been included in several collision models, which have remained without a proper validation thus far due to a scarcity of available data. We therefore conduct direct numerical simulations of bubbles and particles in homogeneous isotropic turbulence with various Stokes, Froude, Reynolds numbers, and particle density ratios using the point-particle approximation. Generally, turbulence enhances the collision rate compared to the pure relative settling case by increasing the collision velocity. Surprisingly, however, for certain parameters the collision rate is lower with turbulence compared to without, independent of the history force. This is due to turbulence-induced bubble-particle spatial segregation, which is most prevalent at weak relative gravity and decreases as gravitational effects become more dominant, and reduced bubble slip velocity in turbulence. The existing bubble-particle collision models only qualitatively capture the trends in our numerical data. To improve on this, we extend the model by Dodin & Elperin (Phys. Fluids, vol. 14, no. 8, 2002, pp.2921-2924) to the bubble-particle case and found excellent quantitative agreement for small Stokes numbers when the history force is negligible and segregation is accounted for.

physics.flu-dyn↗

arXiv2024

Optimization of noncollinear magnetic ordering temperature in Y-type hexaferrite by machine learning

Searching the optimal doping compositions of the Y-type hexaferrite Ba2Mg2Fe12O22 remains a long-standing challenge for enhanced non-collinear magnetic transition temperature (TNC). Instead of the conventional trial-and-error approach, the composition-property descriptor is established via a data driven machine learning method named SISSO (sure independence screening and sparsifying operator). Based on the chosen efficient and physically interpretable descriptor, a series of Y-type hexaferrite compositions are predicted to hold high TNC, among which the BaSrMg0.28Co1.72Fe10Al2O22 is then experimentally validated. Test results indicate that, under appropriate external magnetic field conditions, the TNC of this composition reaches up to reaches up to 568 K, and its magnetic transition temperature is also elevated to 735 K. This work offers a machine learning-based route to develop room temperature single phase multiferroics for device applications.

cond-mat.mtrl-sci↗

arXiv2024

How turbulence increases the bubble-particle collision rate

We study the effect of turbulence on collisions between a finite-size bubble and small inertial particles based on interface-resolved simulations. Our results show that the interaction with the flow field around the bubble remains the dominant effect. Nonlinear dependencies in this process can enhance the turbulent collision rate by up to 100\% compared to quiescent flow. Fluctuations in the bubble slip velocity during the interaction with the particle additionally increase the collision rate. We present a frozen-turbulence model that captures the relevant effects providing a physically consistent framework to model collisions of small inertial particles with finite-sized objects in turbulence.

physics.flu-dyn↗

arXiv2024

Data Smoothing Filling Method based on ScRNA-Seq Data Zero-Value Identification

Single-cell RNA sequencing (scRNA-seq) determines RNA expression at single-cell resolution. It provides a powerful tool for studying immunity, regulation, and other life activities of cells. However, due to the limitations of the sequencing technique, the scRNA-seq data are represented with sparsity, whichcontains missing gene values, i.e., zero values, called dropout. Therefore, it is necessary to impute missing values before analyzing scRNA-seq data. However, existing imputation computation methods often only focus on the identification of technical zeros or imputing all zeros based on cell similarity. This study proposes a new method (SFAG) to reconstruct the gene expression relationship matrix by usinggraph regularization technology to preserve the high-dimensional manifold information of the data, andto mine the relationship between genes and cells in the data, and then uses a method of averaging the clustering results to fill in the identified technical zeros. Experimental results show that SFAGcan helpimprove downstream analysis and reconstruct cell trajectory

q-bio.GN↗

arXiv2024

Accelerations of large inertial particles in turbulence

Understanding the dynamics of material objects advected by turbulent flows is a long standing question in fluid dynamics. In this perspective article we focus on the characterization of the statistical properties of non-interacting finite-sized massive spherical particles advected by a vigorous turbulent flow. We study the fluctuations and temporal correlations of particle accelerations and explore their behaviours with respect to the particle size and the particle mass density by means of fully-resolved numerical simulations. We observe that the measured trends can not be interpreted as the simple multiplicative combination of the two dominant effects: the spatial filtering of fluid accelerations and the added-mass-adjusted fluid-to-particle density ratio. We argue that other hydrodynamical forces or effects, e.g. preferential flow sampling, have still a significant role even at the largest particle sizes, which rich here the integral scale of turbulence.

physics.flu-dyn↗

arXiv2023

Mobile phone data reveal spatiotemporal dynamics of Omicron infections in Beijing after relaxing zero-COVID policy

The swift relaxation of the zero-COVID policy in December 2022 led to an unprecedented surge in Omicron variant infections in China. With the suspension of mandatory testing, tracking this epidemic outbreak was challenging because infections were often underrepresented in survey and testing results, which only involved partial populations. We used large-scale mobile phone data to estimate daily infections in Beijing from November 2022 to January 2023. We demonstrated that an individual's location records of mobile phone could be used to infer his or her infectious status. Then, the derived status of millions of individuals could be summed to reconstruct the citywide spatiotemporal dynamics of infections. We found that the infection incidence peaked on 21 December, and 80.1% of populations had been infected by 14 January 2023 in Beijing. Furthermore, infection dynamics exhibited significant demographic and spatiotemporal disparities. Our work provides a ubiquitous and high-coverage data source for monitoring epidemic outbreaks.

physics.soc-ph↗

arXiv2023

Numerical study on turbulence modulation of finite-size particles in plane-Couette flow

Turbulent plane-Couette flow suspended with finite-size spheroidal particles is studied using fully particle-resolved direct numerical simulations. The effects of particle aspect ratio on turbulent arguments and particle statistics are explored, leading to the same conclusions as the previous experimental findings \citep{wang2022finite}. By performing stress analysis, we find that the presence of particles introduces extra stresses to the system and accounts for the global drag increases. The particle-laden flow cases exhibit spectra that are consistent with the scaling of $k^{-5/3}$ and $k^{-3}$ in the large and small scales, respectively. While the $k^{-3}$ scaling observed in the particle-laden flow is reminiscent of bubbly flow, an examination of the particle Reynolds number suggests that the mechanism responsible may not be attributable to the pseudo-turbulence induced by particles as in the case of bubbles. In the view of particle statistics, we observe that spherical and non-spherical particles preferentially cluster in the near-wall and the bulk region, respectively, and that the orientations of non-spherical particles are affected by their aspect ratios, especially in the near-wall region. The present numerical results, combined with previous experimental findings in \cite{wang2022finite}, provide in-depth information on both the fluid and the particle phase, contributing to a better understanding of particle suspension in shear flows.

physics.flu-dyn↗

arXiv2023

Sea Water Freezing Modes in a Natural Convection System

Sea ice is crucial in many natural processes and human activities. Understanding the dynamical couplings between the inception, growth and equilibrium of sea ice and the rich fluid mechanical processes occurring at its interface and interior is of relevance in many domains ranging from geophysics to marine engineering. Here we experimentally investigate the complete freezing process of water with dissolved salt in a standard natural convection system, i.e., the prototypical Rayleigh-Bénard cell. Due to the presence of a mushy phase, the studied system is considerably more complex than the freezing of freshwater in the same conditions (Wang et al. 2021c). We measure the ice thickness and porosity at the dynamical equilibrium state for different initial salinities of the solution and temperature gaps across the cell. These observables are non-trivially related to the controlling parameters of the system as they depend on the heat transport mode across the cell. We identify in the experiments 5 out of the 6 possible modes of heat transport. We highlight the occurrence of brine convection through the mushy ice and of penetrative convection in stably stratified liquid underlying the ice. A one-dimensional multi-layer heat flux model built on the known scaling relations of global heat transport in natural convection systems in liquids and porous media is proposed. It allows, given the measured porosity of the ice, to predict the corresponding ice thickness, in a unified framework.

physics.flu-dyn↗

arXiv2022

Turbulence drag modulation by dispersed droplets in Taylor-Couette flow: the effects of the dispersed phase viscosity

The dispersed phase in turbulence can vary from almost inviscid fluid to highly viscous fluid. By changing the viscosity of the dispersed droplet phase, we experimentally investigate how the deformability of dispersed droplets affects the global transport quantity of the turbulent emulsion. Different kinds of silicone oil are employed to result in the viscosity ratio, $ζ$, ranging from $0.53$ to $8.02$. The droplet volume fraction, $ϕ$, is varied from 0\% to 10\% with a spacing of 2\%. The global transport quantity, quantified by the normalized friction coefficient $c_{f,ϕ}/c_{f,ϕ=0}$, shows a weak dependence on the turbulent intensity due to the vanishing finite-size effect of the droplets. The interesting fact is that, with increasing $ζ$, the $c_{f,ϕ}/c_{f,ϕ=0}$ first increases and then saturates to a plateau value which is similar to that of the rigid particle suspension. By performing image analysis, this drag modification is interpreted from the aspect of droplet deformability, which is responsible for the breakup and coalescence effect of the droplets. The statistics of the droplet size distribution show that, with increasing $ζ$, the stabilizing effect induced by interfacial tension comes to be substantial and the pure inertial breakup process becomes dominant. The measurement of the droplet distribution along the radial direction of the system shows a bulk-clustering effect, which can be attributed to the non-negligible coalescence effect of the droplet. It is found that the droplet coalescence effect could be suppressed as the $ζ$ increases, thereby affecting the contribution of interfacial tension to the total stress, and accounting for the observed emulsion rheology.

physics.flu-dyn↗

arXiv2022

Accumulation and alignment of elongated gyrotactic swimmers in turbulence

We study the dynamics of gyrotactic swimmers in turbulence, whose orientation is governed by gravitational torque and local fluid velocity gradient. The gyrotaxis strength is measured by the ratio of the Kolmogorov time scale to the reorientation time scale due to gravity, and a large value of this ratio means the gyrotaxis is strong. By means of direct numerical simulations, we investigate the effects of swimming velocity and gyrotactic stability on spatial accumulation and alignment. Three-dimensional Vorono{\"ı} analysis is used to study the spatial distribution and time evolution of the particle concentration. We study spatial distribution by examing the overall preferential sampling and where clusters and voids (subsets of particles that have small and large Vorono{\"ı} volumes respectively) form. Compared with the ensemble particles, the preferential sampling of clusters and voids is found to be more pronounced. The clustering of fast swimmers lasts much longer than slower swimmers when the gyrotaxis is strong and intermediate, but an opposite trend emerges when the gyrotaxis is weak. In addition, we study the preferential alignment with the Lagrangian stretching direction, with which passive slender rods have been known to align. We show that the Lagrangian alignment is reduced by the swimming velocity when the gyrotaxis is weak, while the Lagrangian alignment is enhanced for the regime in which gyrotaxis is strong.

physics.flu-dyn↗

arXiv2022

Dynamics of finite-size spheroids in turbulent flow: the roles of flow structures and particle boundary layers

We study the translational and rotational dynamics of neutrally-buoyant finite-size spheroids in hydrodynamic turbulence by means of fully resolved numerical simulations. We examine axisymmetric shapes, from oblate to prolate, and the particle volume dependences. We show that the accelerations and rotations experienced by non-spherical inertial-scale particles result from volume filtered fluid forces and torques, similar to spherical particles. However, the particle orientations carry signatures of preferential alignments with the surrounding flow structures, which is reflected in distinct axial and lateral fluctuations for accelerations and rotation rates. The randomization of orientations does not occur even for particles with volume equivalent diameter size in the inertial range, here up to 60 $η$ at $Re_λ=120$. Additionally, we demonstrate that the role of fluid boundary layers around the particles cannot be neglected to reach a quantitative understanding of particle statistical dynamics, as they affect the intensities of angular velocities, and the relative importance of tumbling with respect to spinning rotations. This study brings to the fore the importance of inertial-scale flow structures in homogeneous and isotropic turbulence and their impacts on the transport of neutrally-buoyant bodies with size in the inertial range.

physics.flu-dyn↗