arXiv subjects

Jun Lin

Publications and source records attributed to Jun Lin.

At least 73 records · Page 4Linked to original sources

arXiv2023

Three-dimensional simulation of thermodynamics on confined turbulence in a large-scale CME-flare current sheet

Turbulence plays a key role for forming the complex geometry of the large-scale current sheet (CS) and fast energy release in a solar eruption. In this paper, we present full 3D high-resolution simulations for the process of a moderate Coronal Mass Ejection (CME) and the thermodynamical evolution of the highly confined CS. Copious elongated blobs are generated due to tearing and plasmoid instabilities giving rise to a higher reconnection rate and undergo the splitting, merging and kinking processes in a more complex way in 3D. A detailed thermodynamical analysis shows that the CS is mainly heated by adiabatic and numerical viscous terms, and thermal conduction is the dominant factor that balances the energy inside the CS. Accordingly, the temperature of the CS reaches to a maximum of about 20 MK and the range of temperatures is relatively narrow. From the face-on view in the synthetic Atmospheric Imaging Assembly 131 $\mathring{A}$, the downflowing structures with similar morphology to supra-arcade downflows are mainly located between the post-flare loops and loop-top, while moving blobs can extend spikes higher above the loop-top. The downward-moving plasmoids can keep the twisted magnetic field configuration until the annihilation at the flare loop-top, indicating that plasmoid reconnection dominates in the lower CS. Meanwhile, the upward-moving ones turn into turbulent structures before arriving at the bottom of the CME, implying that turbulent reconnection dominates in the upper CS. The spatial distributions of the turbulent energy and anisotropy are addressed, which show a significant variation in the spectra with height.

astro-ph.SR↗

arXiv2023

High $β$ magnetic reconnection at different altitudes in the cool low solar atmosphere

We numerically studied magnetic reconnection in a high $β$ hydrogen-helium plasma at different altitudes from the photosphere to the upper chromosphere. The time dependent ionization degrees were included to get more realistic diffusivities and viscosity, and appropriate radiative cooling models were applied. Our numerical results indicate that the plasmoid instability always plays a vital role in speeding up magnetic reconnection at different atmospheric layers. In addition, both the strong radiative cooling and the magnetic diffusion caused by the electron-neutral collision ($η_{en}$) can significantly accelerate magnetic reconnection below the middle chromosphere. On the other hand, both the ambipolar diffusion and the viscosity result in higher temperature and plasma pressure in the reconnection region in the upper chromosphere, which then hinder the fast reconnection process from developing. The local compression heating triggered by turbulent reconnection mediated with plasmoids is the dominant heating mechanism in the unstable reconnection stage at different atmospheric layers, but the viscous heating and the ambipolar diffusion heating are equally important in the upper chromosphere. The Joule heating contributed by $η_{en}$ dominates during the early quasi-steady reconnection stage below the middle chromosphere, the strong radiative cooling also leads to much stronger compression heating and more generation of thermal energy in this region. Though the plasma $β$ is the same in all the simulation cases at different altitudes, the temperature increase is more significant in the upper chromosphere with much lower density and weaker radiative cooling.

astro-ph.SR↗

arXiv2022

Fast and Accurate FSA System Using ELBERT: An Efficient and Lightweight BERT

With the development of deep learning and Transformer-based pre-trained models like BERT, the accuracy of many NLP tasks has been dramatically improved. However, the large number of parameters and computations also pose challenges for their deployment. For instance, using BERT can improve the predictions in the financial sentiment analysis (FSA) task but slow it down, where speed and accuracy are equally important in terms of profits. To address these issues, we first propose an efficient and lightweight BERT (ELBERT) along with a novel confidence-window-based (CWB) early exit mechanism. Based on ELBERT, an innovative method to accelerate text processing on the GPU platform is developed, solving the difficult problem of making the early exit mechanism work more effectively with a large input batch size. Afterward, a fast and high-accuracy FSA system is built. Experimental results show that the proposed CWB early exit mechanism achieves significantly higher accuracy than existing early exit methods on BERT under the same computation cost. By using this acceleration method, our FSA system can boost the processing speed by nearly 40 times to over 1000 texts per second with sufficient accuracy, which is nearly twice as fast as FastBERT, thus providing a more powerful text processing capability for modern trading systems.

cs.CL↗

arXiv2022

Modelling the magnetic structure of a large-scale horse-shoe-like filament in a decaying and diffuse active region

A large-scale, horse-shoe-like filament was investigated and the magnetic field around it was reconstructed. This is an intermediate filament (IF) that appeared on the solar disk for the first time at 02:00 UT on 2015 November 7, and took 8 days to move to the central median on the solar disk. The active region AR 12452 around which the filament occurred was diffuse so that the magnetic field nearby was weak, the average field strength is 106 G. Therefore, the existing approaches to extrapolating the coronal magnetic field and to constructing the filament configuration in the region with strong background field do not work well here. On the basis of the regularized Biot-Savart laws method, we successfully constructed a data-constrained, non-linear force-free field configuration for this IF observed on 2015 November 14. The overall IF configuration obtained in this way matches well the morphology suggested by a 304~Å\ image taken by the Atmospheric Imaging Assembly on board Solar Dynamics Observatory. Magnetic dips in the configuration were coincident in space with the H$α$ features of the filament, which is lower in altitude than the features seen in 304~Å. This suggests that the cold plasma fills the lower part of the filament, and hot plasma is situated in the higher region. A quasi-separatrix layer wraps the filament, and both the magnetic field and the electric current are stronger near the inner edge of the filament.

astro-ph.SR↗

arXiv2022

Accelerate Three-Dimensional Generative Adversarial Networks Using Fast Algorithm

Three-dimensional generative adversarial networks (3D-GAN) have attracted widespread attention in three-dimension (3D) visual tasks. 3D deconvolution (DeConv), as an important computation of 3D-GAN, significantly increases computational complexity compared with 2D DeConv. 3D DeConv has become a bottleneck for the acceleration of 3D-GAN. Previous accelerators suffer from several problems, such as large memory requirements and resource underutilization. To handle the above issues, a fast algorithm for 3D DeConv (F3DC) is proposed in this paper. F3DC applies a fast algorithm to reduce the number of multiplications and achieves a significant algorithmic strength reduction. Besides, F3DC removes the extra memory requirement for overlapped partial sums and avoids computational imbalance to fully utilize resources. Moreover, we design an F3DC-based hardware architecture, which consists of four fast processing units (FPUs). Each FPU includes a pre-process module, a EWMM module and a post-process module for F3DC transformation. By implementing our design on the Xilinx VC709 platform for 3D-GAN, we achieve a throughput up to 1700 GOPS and 4$\times$ computational efficiency improvement compared with prior works.

eess.SP↗

arXiv2022

An Efficient FPGA Accelerator for Point Cloud

Deep learning-based point cloud processing plays an important role in various vision tasks, such as autonomous driving, virtual reality (VR), and augmented reality (AR). The submanifold sparse convolutional network (SSCN) has been widely used for the point cloud due to its unique advantages in terms of visual results. However, existing convolutional neural network accelerators suffer from non-trivial performance degradation when employed to accelerate SSCN because of the extreme and unstructured sparsity, and the complex computational dependency between the sparsity of the central activation and the neighborhood ones. In this paper, we propose a high performance FPGA-based accelerator for SSCN. Firstly, we develop a zero removing strategy to remove the coarse-grained redundant regions, thus significantly improving computational efficiency. Secondly, we propose a concise encoding scheme to obtain the matching information for efficient point-wise multiplications. Thirdly, we develop a sparse data matching unit and a computing core based on the proposed encoding scheme, which can convert the irregular sparse operations into regular multiply-accumulate operations. Finally, an efficient hardware architecture for the submanifold sparse convolutional layer is developed and implemented on the Xilinx ZCU102 field-programmable gate array board, where the 3D submanifold sparse U-Net is taken as the benchmark. The experimental results demonstrate that our design drastically improves computational efficiency, and can dramatically improve the power efficiency by 51 times compared to GPU.

eess.SP↗

arXiv2022

Research on Stable Obstacle Avoidance Control Strategy for Tracked Intelligent Transportation Vehicles in Non-structural Environment Based on Deep Learning

Existing intelligent driving technology often has a problem in balancing smooth driving and fast obstacle avoidance, especially when the vehicle is in a non-structural environment, and is prone to instability in emergency situations. Therefore, this study proposed an autonomous obstacle avoidance control strategy that can effectively guarantee vehicle stability based on Attention-long short-term memory (Attention-LSTM) deep learning model with the idea of humanoid driving. First, we designed the autonomous obstacle avoidance control rules to guarantee the safety of unmanned vehicles. Second, we improved the autonomous obstacle avoidance control strategy combined with the stability analysis of special vehicles. Third, we constructed a deep learning obstacle avoidance control through experiments, and the average relative error of this system was 15%. Finally, the stability and accuracy of this control strategy were verified numerically and experimentally. The method proposed in this study can ensure that the unmanned vehicle can successfully avoid the obstacles while driving smoothly.

cs.RO↗

arXiv2022

Plausibility of ultraviolet burst generation in the low solar chromosphere

Ultraviolet (UV) bursts and Ellerman bombs (EBs) are small-scale magnetic reconnection events taking place in the highly stratified, low solar atmosphere. It is still not clear whether UV bursts have to be generated at a higher atmospheric layer than EBs or whether both UV bursts and EBs can occur in the low chromosphere. We numerically studied the low $β$ magnetic reconnection process around the solar temperature minimum region (TMR). The time-dependent ionization degrees of hydrogen and helium are included in the MHD code, which lead to a more realistic magnetic diffusion caused by electron-neutral collision and ambipolar diffusion. A more realistic radiative cooling model from Carlsson & Leenaarts 2012 is included in the simulations. Our results in high resolution indicate that the plasmas in the reconnection region are heated up to more than $20,000$ K if the reconnecting magnetic field is as strong as $500$ G, which suggests that UV bursts can be generated in the dense low chromosphere. The dominant mechanism for producing the UV burst in the low chromosphere is heating, as a result of the local compression in the reconnection process. The thermal energy occurring in the reconnection region rapidly increases after the turbulent reconnection mediated by plasmoids is invoked. The average power density of the generated thermal energy in the reconnection region can reach over $1000$ erg cm$^{-3}$ s$^{-1}$, which is comparable to the average power density accounting for a UV burst. With the strength of the reconnecting magnetic field exceeding $900$ G, the width of the synthesized Si IV 1394 A line profile with multiple peaks can reach up to $100$ km s$^{-1}$, which is consistent with observations.

astro-ph.SR↗

arXiv2022

Data-Driven Thermal Modelling for Anomaly Detection in Electric Vehicle Charging Stations

The rapid growth of the electric vehicle (EV) sector is giving rise to many infrastructural challenges. One such challenge is its requirement for the widespread development of EV charging stations which must be able to provide large amounts of power in an on-demand basis. This can cause large stresses on the electrical and electronic components of the charging infrastructure - negatively affecting its reliability as well as leading to increased maintenance and operation costs. This paper proposes a human-interpretable data-driven method for anomaly detection in EV charging stations, aiming to provide information for the condition monitoring and predictive maintenance of power converters within such a station. To this end, a model of a high-efficiency EV charging station is used to simulate the thermal behaviour of EV charger power converter modules, creating a data set for the training of neural network models. These machine learning models are then employed for the identification of anomalous performance.

eess.SP↗

arXiv2022

Two-dimensional modeling of the tearing-mode-governed magnetic reconnection in the large-scale current sheet above the two-ribbon flare

We attempt to model magnetic reconnection during the two-ribbon flare in the gravitationally stratified solar atmosphere with the Lundquist number of $S=10^6$ using 2D simulations. We found that the tearing mode instability leads to the inhomogeneous turbulence inside the reconnecting current sheet (CS) and invokes the fast phase of reconnection. Fast reconnection brings an extra dissipation of magnetic field which enhances the reconnection rate in an apparent way. The energy spectrum in the CS shows the power-law pattern and the dynamics of plasmoids governs the associated spectral index. We noticed that the energy dissipation occurs at a scale $l_{ko}$ of 100-200~km, and the associated CS thickness ranges from 1500 to 2500~km, which follows the Taylor scale $l_T=l_{ko} S^{1/6}$. The termination shock(TS) appears in the turbulent region above flare loops, which is an important contributor to heating flare loops. Substantial magnetic energy is converted into both kinetic and thermal energies via TS, and the cumulative heating rate is greater than the rate of the kinetic energy transfer. In addition, the turbulence is somehow amplified by TS, of which the amplitude is related to the local geometry of the TS.

astro-ph.SR↗

arXiv2022

The Chinese Hα Solar Explorer (CHASE) mission: An overview

The Chinese Hα Solar Explorer (CHASE), dubbed "Xihe" - Goddess of the Sun, was launched on October 14, 2021 as the first solar space mission of China National Space Administration (CNSA). The CHASE mission is designed to test a newly developed satellite platform and to acquire the spectroscopic observations in the Hα waveband. The Hα Imaging Spectrograph (HIS) is the scientific payload of the CHASE satellite. It consists of two observational modes: raster scanning mode and continuum imaging mode. The raster scanning mode obtains full-Sun or region-of-interest spectral images from 6559.7 to 6565.9 Å and from 6567.8 to 6570.6 Å with 0.024 Å pixel spectral resolution and 1 minute temporal resolution. The continuum imaging mode obtains photospheric images in continuum around 6689 Å with the full width at half maximum of 13.4 Å. The CHASE mission will advance our understanding of the dynamics of solar activity in the photosphere and chromosphere. In this paper, we present an overview of the CHASE mission including the scientific objectives, HIS instrument overview, data calibration flow, and first results of on-orbit observations.

astro-ph.SR↗

arXiv2022

Leveraging Online Shopping Behaviors as a Proxy for Personal Lifestyle Choices: New Insights into Chronic Disease Prevention Literacy

Objective: Ubiquitous internet access is reshaping the way we live, but it is accompanied by unprecedented challenges in preventing chronic diseases that are usually planted by long exposure to unhealthy lifestyles. This paper proposes leveraging online shopping behaviors as a proxy for personal lifestyle choices to improve chronic disease prevention literacy, targeted for times when e-commerce user experience has been assimilated into most people's everyday lives. Methods: Longitudinal query logs and purchase records from 15 million online shoppers were accessed, constructing a broad spectrum of lifestyle features covering various product categories and buyer personas. Using the lifestyle-related information preceding online shoppers' first purchases of specific prescription drugs, we could determine associations between their past lifestyle choices and whether they suffered from a particular chronic disease. Results: Novel lifestyle risk factors were discovered in two exemplars--depression and type 2 diabetes, most of which showed reasonable consistency with existing healthcare knowledge. Further, such empirical findings could be adopted to locate online shoppers at higher risk of these chronic diseases with decent accuracy [i.e., (area under the receiver operating characteristic curve) AUC=0.68 for depression and AUC=0.70 for type 2 diabetes], closely matching the performance of screening surveys benchmarked against medical diagnosis. Conclusions: Mining online shopping behaviors can point medical experts to a series of lifestyle issues associated with chronic diseases that are less explored to date. Hopefully, unobtrusive chronic disease surveillance via e-commerce sites can grant consenting individuals a privilege to be connected more readily with the medical profession and sophistication.

cs.CY↗

arXiv2021

Chinese E-Romance: Analyzing and Visualizing 7.92 Million Alibaba Valentine's Day Purchases

The days that precede Valentine's Day are characterized by extensive gift shopping activities all across the globe. In China, where much shopping takes place online, there has been an explosive growth in e-commerce sales during Valentine's Day over the recent years. This exploratory study investigates the extent to which each product category and each shopper group can exhibit romantic love within China's e-market throughout the 2 weeks leading up to 2019 Valentine's Day. Massive data from Alibaba, the biggest e-commerce retailer worldwide, are utilized to formulate an innovative romance index (RI) to quantitatively measure e-romantic values for products and shoppers. On this basis, millions of shoppers, along with their millions of products purchased around Valentine's Day, are analyzed as a case study to demonstrate their love consumption and romantic gift-giving. The results of the analysis are then illustrated to help understand Chinese e-romance based on the perspectives of different product categories and shopper groups. This empirical information visualization also contributes to improving the segmentation, targeting, and positioning of China's e-market for Valentine's Day.

cs.CY↗

arXiv2021

Unsupervised Summarization for Chat Logs with Topic-Oriented Ranking and Context-Aware Auto-Encoders

Automatic chat summarization can help people quickly grasp important information from numerous chat messages. Unlike conventional documents, chat logs usually have fragmented and evolving topics. In addition, these logs contain a quantity of elliptical and interrogative sentences, which make the chat summarization highly context dependent. In this work, we propose a novel unsupervised framework called RankAE to perform chat summarization without employing manually labeled data. RankAE consists of a topic-oriented ranking strategy that selects topic utterances according to centrality and diversity simultaneously, as well as a denoising auto-encoder that is carefully designed to generate succinct but context-informative summaries based on the selected utterances. To evaluate the proposed method, we collect a large-scale dataset of chat logs from a customer service environment and build an annotated set only for model evaluation. Experimental results show that RankAE significantly outperforms other unsupervised methods and is able to generate high-quality summaries in terms of relevance and topic coverage.

cs.CL↗

arXiv2021

Topic-Oriented Spoken Dialogue Summarization for Customer Service with Saliency-Aware Topic Modeling

In a customer service system, dialogue summarization can boost service efficiency by automatically creating summaries for long spoken dialogues in which customers and agents try to address issues about specific topics. In this work, we focus on topic-oriented dialogue summarization, which generates highly abstractive summaries that preserve the main ideas from dialogues. In spoken dialogues, abundant dialogue noise and common semantics could obscure the underlying informative content, making the general topic modeling approaches difficult to apply. In addition, for customer service, role-specific information matters and is an indispensable part of a summary. To effectively perform topic modeling on dialogues and capture multi-role information, in this work we propose a novel topic-augmented two-stage dialogue summarizer (TDS) jointly with a saliency-aware neural topic model (SATM) for topic-oriented summarization of customer service dialogues. Comprehensive studies on a real-world Chinese customer service dataset demonstrated the superiority of our method against several strong baselines.

cs.CL↗

arXiv2021

The Ellerman bomb and Ultraviolet burst triggered successively by an emerging magnetic flux rope

Ellerman bombs (EBs) and Ultraviolet (UV) bursts are common brightening phenomena which are usually generated in the low solar atmosphere of emerging flux regions. In this paper, we have investigated the emergence of an initial un-twisted magnetic flux rope based on three-dimensional (3D) magneto-hydrodynamic (MHD) simulations. The EB-like and UV burst-like activities successively appear in the U-shaped part of the undulating magnetic fields triggered by Parker Instability. The EB-like activity starts to appear earlier and lasts for about 80 seconds. Six minutes later, a much hotter UV burst-like event starts to appear and lasts for about 60 seconds. Along the direction vertical to the solar surface, both the EB and UV burst start in the low chromosphere, but the UV burst extends to a higher altitude in the up chromosphere. The regions with apparent temperature increase in the EB and UV burst are both located inside the small twisted flux ropes generated in magnetic reconnection processes, which are consistent with the previous 2D simulations that most hot regions are usually located inside the magnetic islands. However, the twisted flux rope corresponding to the EB is only strongly heated after it floats up to an altitude much higher than the reconnection site during that period. Our analyses show that the EB is heated by the shocks driven by the strong horizontal flows at two sides of the U-shaped magnetic fields. The twisted flux rope corresponding to the UV burst is heated by the driven magnetic reconnection process.

astro-ph.SR↗

arXiv2020

A Magnetic Reconnection model for Hot Explosions in the Cool Atmosphere of the Sun

UV bursts and Ellerman bombs are transient brightenings observed in the low solar atmospheres of emerging flux regions. Observations have discovered the cospatial and cotemporal EBs and UV bursts, and their formation mechanisms are still not clear. The multi-thermal components with a large temperature span in these events challenge our understanding of magnetic reconnection and heating mechanisms in the low solar atmosphere. We have studied magnetic reconnection between the emerging and background magnetic fields. The initial plasma parameters are based on the C7 atmosphere model. After the current sheet with dense photosphere plasma is emerged to $0.5$ Mm above the solar surface, plasmoid instability appears. The plasmoids collide and coalesce with each other, which makes the plasmas with different densities and temperatures mixed up in the turbulent reconnection region. Therefore, the hot plasmas corresponding to the UV emissions and colder plasmas corresponding to the emissions from other wavelenghts can move together and occur at about the same height. In the meantime, the hot turbulent structures basically concentrate above $0.4$ Mm, whereas the cool plasmas extend to much lower heights to the bottom of the current sheet. These phenomena are consistent with the observations of Chen et al. 2019, ApJL. The synthesized Si IV line profiles are similar to the observed one in UV bursts, the enhanced wing of the line profiles can extend to about $100$ km s$^{-1}$. The differences are significant among the numerical results with different resolutions, which indicate that the realistic magnetic diffusivity is crucial to reveal the fine structures and realistic plasmas heating in these reconnection events. Our results also show that the reconnection heating contributed by ambipolar diffusion in the low chromosphere around the temperature minimum region is not efficient.

astro-ph.SR↗

arXiv2020

Wide-field Decodable Orthogonal Fingerprints of Single Nanoparticles Unlock Multiplexed Digital Assays

The control in optical uniformity of single nanoparticles and tuning their diversity in orthogonal dimensions, dot to dot, holds the key to unlock nanoscience and applications. Here we report that the time-domain emissive profile from single upconversion nanoparticle, including the rising, decay and peak moment of the excited state population (T2 profile), can be arbitrarily tuned by upconversion schemes, including interfacial energy migration, concentration dependency, energy transfer, and isolation of surface quenchers. This allows us to significantly increase the coding capacity at the nanoscale. We further implement both time-resolved wide-field imaging and deep-learning techniques to decode these fingerprints, showing high accuracies at high throughput. These high-dimensional optical fingerprints provide a new horizon for applications spanning from sub-diffraction-limit data storage, security inks, to high-throughput single-molecule digital assays and super-resolution imaging.

physics.app-ph↗