arXiv subjects

Bin Jia

Publications and source records attributed to Bin Jia.

At least 19 recordsLinked to original sources

arXiv2026

As Cities Grow, They Spread Out Rather Than Rise

Cities and settlements are a global phenomenon, and their continued expansion is fundamentally transforming patterns of resource demand. This transformation is reflected in growing pressures on land, material use, and infrastructure. Thus, understanding how cities grow in space is essential for planning future urban development and determining all its consequences. Here, we characterize four geometrically related dimensions of urban growth: two horizontal dimensions describing the urban area, one vertical dimension represented by average building height, which together with the urban area defines building volume, and the population dimension. Analyzing a global dataset of growing cities, for the period 1975-2025 we identify three main patterns. First, urban growth is fundamentally anisotropic -- as cities gain population, they expand much more rapidly in the horizontal than in the vertical direction. Second, average building height decreases in the vast majority of cities as they grow, indicating that vertical development is not only slower than horizontal expansion but on average, can act in an effectively negative manner by giving greater weight to the horizontal dimensions. Third, despite their diverse growth trajectories, cities converge toward characteristic population densities, both with respect to urban area and building volume. We further explore alternative scenarios for future urban growth. These projections show that different assumptions about horizontal and vertical scaling can lead to substantially different demands for urban land and building volume. These findings reveal constraints on the long-term evolution of urban form and provide a quantitative assessment for understanding the spatial dynamics and development of cities.

physics.soc-ph↗

arXiv2026

Spatially resolved molecular gas conditions in the circumnuclear disc of 3C 84

The brightest cluster galaxy NGC 1275, at the centre of the Perseus cluster, hosts the radio-loud AGN 3C 84 and a circumnuclear disc (CND) of cold molecular gas onto which large-scale CO filaments accrete, yet the physical and chemical conditions of the gas in the CND remain poorly constrained. We present a spatially resolved analysis of ALMA CO(2-1), HCN(3-2), and HCO$^+$(3-2) observations at 72 pc. The data are partitioned into beam-sized hexagonal regions and modelled with a Bayesian framework that couples a neural network emulator of time-dependent chemistry (UCLCHEM) with non-LTE radiative transfer (SpectralRadex) to infer the gas density, kinetic temperature, and cosmic ray ionisation rate in each region. All three line ratios peak in the inner disc and decline with radius. The inference shows radial gradients in density ($\log_{10} n({\rm H_2}) \approx 6.3$ to $\sim 5$), kinetic temperature (~200 K to ~160 K), and cosmic ray ionisation rate ($\log_{10}(ζ/ζ_0) \approx 4.9$ to $\sim 3$). Despite the powerful radio AGN, the observed HCN(3-2)/HCO$^+$(3-2) ratio remains $\lesssim 1$ across the disc. Our modelling attributes this to optical depth saturation of HCN(3-2) ($τ\sim 1$-3), which suppresses the intensity ratio even when the HCN abundance exceeds that of HCO$^+$ by a factor of three or more. The HCN/HCO$^+$ intensity ratio therefore cannot be used as an abundance diagnostic without accounting for optical depth, and a low ratio does not necessarily imply weak AGN influence on the chemistry. Azimuthally resolved profiles suggest a localised HCO$^+$/CO enhancement at the western disc boundary, coinciding with the filament-disc accretion interface and consistent with shock processing by velocity shear between the infalling filaments and the rotating disc. These results indicate that the CND is shaped by accretion from its filamentary environment.

astro-ph.GA↗

arXiv2026

The curious case of HCO$^+$: Extreme abundances under extreme conditions

Context. HCO$^+$ is widely observed in both Galactic and extragalactic environments and typically exhibits abundances of $10^{-9}-10^{-8}$. However, recent modeling studies suggest that in environments exposed to elevated cosmic-ray ionization rates and strong thermal or mechanical processing its abundance may increase by several orders of magnitude. Aims. To interpret these predictions, we need to understand the physical conditions that produce extreme HCO$^+$ abundances and the chemical pathways that drive these enhancements. Methods. We used UCLCHEM, a gas-grain chemical code, to model the chemistry of HCO$^+$ in dense molecular, protostellar, and shocked gas under elevated cosmic-ray ionization rates ($ζ\ge 10^{-15}\,\mathrm{s^{-1}}$). Results. Extreme HCO$^+$ enhancements leading to $X$(HCO$^+$) $\gtrsim 10^{-4}$ occur only under specific combinations of temperature, density, and cosmic-ray ionization rate, primarily in protostellar and shocked gas. Increasing density generally suppresses the peak HCO$^+$ abundance, requiring higher ionization rates to produce comparable enhancements. More importantly, the extreme enhancements seem to be very dependent on the chemical network used (in our case UMIST12 versus UMIST22, with the latter leading to extreme abundances). These differences among networks arise from the removal of the destruction pathway of HCO$^+$: C + HCO$^+$ $\rightarrow$ CO + CH$^+$, and propagate to several other species including N$_2$H$^+$, H$_2$O, and H$_3$O$^{+}$.

astro-ph.GA↗

arXiv2026

Economic Distance Structures Urban Mobility in 109 U.S. Cities

Urban mobility promises social integration, yet daily movement is systematically constrained by socioeconomic hierarchies. Introducing "economic distance"--the continuous income gap between origin and destination--as a unified lens, we analyze large-scale mobility records across 109 U.S. cities to reveal how urban flows are structured. We identify a universal structural boundary: flows concentrate intensely within a narrow economic distance of 0.25 quantiles, defining the effective "economic radius" of routine mobility. This boundary exhibits profound asymmetry; upward mobility faces a uniform structural ceiling across cities, whereas downward mobility drives cross-city heterogeneity. Mechanistically, the boundary is physically anchored by meso-scale residential clustering but is further tightened by an independent economic-distance friction, validated via gravity modeling. These interactions yield four distinct mobility regimes, with "affluent-confined" systems exhibiting the strongest stratification. These findings establish economic distance as a fundamental, asymmetric, and multi-scale filter shaping urban inequality, offering new theoretical grounds for interventions targeting structural barriers to cross-class interaction.

physics.soc-ph↗

arXiv2026

Self-similarity of mobility networks

Mobility systems of people and goods are inherently multi-scale, spanning levels of organization from individual cities to regions and nations. Understanding whether mobility networks exhibit similar patterns across these scales is important. Such similarity would point to common organizing principles, enabling insights gained at one scale to inform planning and management at others. Despite growing efforts to analyze mobility at multiple scales, such cross-scale similarity remains poorly understood, and renormalization provides a natural framework for addressing this question. Here, we propose a Neighbor-Limited Box Covering method to renormalize undirected weighted mobility networks. This method iteratively selects box centers in descending order of node strength, merges each center with a fixed number of its highest-weight neighbors to form a renormalized node, and aggregates edge weights between renormalized nodes to generate the network at the next scale. We apply this technique to uncover multi-scale structures of real-world inter-city human mobility and freight trip networks in China and find that the topological structures, weighted structural features, and dynamic processes all exhibit self-similarity across these multi-scale mobility networks. Moreover, we find that the constituent nodes in most renormalized nodes show a strong spatial cohesion, and the boundaries of them closely follow existing political and socio-economic borders, even though the method does not explicitly incorporate any spatial information. Our study not only reveals the consistency of multi-scale inter-city mobility patterns, but also provides important insights into their spatial organization. Furthermore, our method is applicable to mobility networks of different sizes and has potential as a powerful tool for the multi-scale analysis of various other real-world complex systems.

physics.soc-ph↗

arXiv2026

Physical and Chemical Conditions of Molecular Gas in NGC 1068: The nuclear feedback in the circumnuclear disk and starburst ring

Molecular gas in galaxies is shaped by both star formation and active galactic nuclei. In NGC 1068, the circumnuclear disk and the starburst ring offer a nearby case to study these effects with many molecular tracers. Earlier work has shown strong outflow activity and complex chemistry, which motivates the use of methods that combine radiative transfer with time-dependent chemistry. Our aim is to map the physical conditions across the circumnuclear disk and the starburst ring of NGC 1068 and to test whether the nuclear outflow influences the molecular gas in the ring. We also examine whether the heating or the quiescent cloud scenario better matches the observations. We use archival ALMA observations obtained in Bands 3, 4, and 5, covering molecular species including HCN, HCO+, HNC, CS, CN and C2H. All data cubes are convolved to a common resolution of 0.8" and are sampled into 56 pc hexagons with a signal-to-noise threshold of three. We perform hierarchical Bayesian inference that links a non-LTE radiative transfer module SpectralRadex with chemical modelling. To make the analysis efficient, we replace direct UCLCHEM calculations with a neural network emulator trained on a large model grid. Sampling is done with Nautilus. We also compare our results with previous studies that used RADEX and UCLCHEM for selected regions. The emulator reproduces the UCLCHEM abundances with low error and allows inference at modest computational cost. We find clear radial and azimuthal variations in gas density, temperature, column density, and cosmic-ray ionization rate.

astro-ph.GA↗

arXiv2025

Observational evidence for a possible link between PAH emission and dust trap locations in protoplanetary disks

Polycyclic Aromatic Hydrocarbons (PAHs) are commonly detected in protoplanetary disks, but it is unclear what causes the wide range of intensities across the samples. In this work, the measured PAH intensities of a range of disks are compared with ALMA dust continuum images, in order to test whether there is evidence that PAHs are frozen out on pebbles in dust traps and only sublimate under certain conditions. A sample is constructed from 26 T Tauri and Herbig disks located within 300 pc, with constraints on the 3.3 $μ$m PAH intensity and with high-resolution ALMA continuum data. The midplane temperature is derived using a power-law or with radiative transfer modeling. The warm dust mass is computed by integrating the flux within the 30 K radius and convert to a dust mass. A strong correlation with a Pearson coefficient of 0.88+/-0.07 between the 3.3 micron PAH intensity and the warm dust mass was found. The correlation is driven by the combination of deep upper limits and strong detections corresponding to a range of warm dust masses. Possible correlations with other disk properties like FUV radiation field or total dust mass are much weaker. Correlations with PAH features at 6.2, 8.6 and 11.3 micron are potentially weaker, but this could be explained by the smaller sample for which these data were available. The correlation is consistent with the hypothesis that PAHs are generally frozen out on pebbles in disks, and are only revealed in the gas phase if those pebbles have drifted towards warm dust traps inside the 30 K radius and vertically transported upwards to the disk atmosphere with sufficiently high temperature to sublimate PAHs into the gas phase. This is similar to previous findings on complex organic molecules in protoplanetary disks and provides further evidence that the chemical composition of the disk is governed by pebble transport.

astro-ph.EP↗

arXiv2025

VeOmni: Scaling Any Modality Model Training with Model-Centric Distributed Recipe Zoo

Recent advances in large language models (LLMs) have driven impressive progress in omni-modal understanding and generation. However, training omni-modal LLMs remains a significant challenge due to the heterogeneous model architectures required to process diverse modalities, necessitating sophisticated system design for efficient large-scale training. Existing frameworks typically entangle model definition with parallel logic, incurring limited scalability and substantial engineering overhead for end-to-end omni-modal training. We present VeOmni, a modular and efficient training framework to accelerate the development of omni-modal LLMs. VeOmni introduces model-centric distributed recipes that decouples communication from computation, enabling efficient 3D parallelism on omni-modal LLMs. VeOmni also features a flexible configuration interface supporting seamless integration of new modalities with minimal code change. Using VeOmni, a omni-modal mixture-of-experts (MoE) model with 30B parameters can be trained with over 2,800 tokens/sec/GPU throughput and scale to 160K context lengths via 3D parallelism on 128 GPUs, showcasing its superior efficiency and scalability for training large omni-modal LLMs.

cs.CL↗

arXiv2025

Urban transport systems shape experiences of social segregation

Mobility is a fundamental feature of human life, and through it our interactions with the world and people around us generate complex and consequential social phenomena. Social segregation, one such process, is increasingly acknowledged as a product of one's entire lived experience rather than mere residential location. Increasingly granular sources of data on human mobility have evidenced how segregation persists outside the home, in workplaces, cafes, and on the street. Yet there remains only a weak evidential link between the production of social segregation and urban policy. This study addresses this gap through an assessment of the role of the urban transportation systems in shaping social segregation. Using city-scale GPS mobility data and a novel probabilistic mobility framework, we establish social interactions at the scale of transportation infrastructure, by rail and bus service segment, individual roads, and city blocks. The outcomes show how social segregation is more than a single process in space, but varying by time of day, urban design and structure, and service design. These findings reconceptualize segregation as a product of likely encounters during one's daily mobility practice. We then extend these findings through exploratory simulations, highlighting how transportation policy to promote sustainable transport may have potentially unforeseen impacts on segregation. The study underscores that to understand social segregation and achieve positive social change urban policymakers must consider the broadest impacts of their interventions and seek to understand their impact on the daily lived experience of their citizens.

physics.soc-ph↗

arXiv2025

Seed1.5-Thinking: Advancing Superb Reasoning Models with Reinforcement Learning

We introduce Seed1.5-Thinking, capable of reasoning through thinking before responding, resulting in improved performance on a wide range of benchmarks. Seed1.5-Thinking achieves 86.7 on AIME 2024, 55.0 on Codeforces and 77.3 on GPQA, demonstrating excellent reasoning abilities in STEM and coding. Beyond reasoning tasks, the method demonstrates notable generalization across diverse domains. For instance, it surpasses DeepSeek R1 by 8% in win rate on non-reasoning tasks, indicating its broader applicability. Compared to other state-of-the-art reasoning models, Seed1.5-Thinking is a Mixture-of-Experts (MoE) model with a relatively small size, featuring 20B activated and 200B total parameters. As part of our effort to assess generalized reasoning, we develop two internal benchmarks, BeyondAIME and Codeforces, both of which will be publicly released to support future research. Model trial link: https://www.volcengine.com/experience/ark.

cs.CL↗

arXiv2025

A Monitoring Method for the Ice Shape and the Freeze-Thaw Process of Ice Accretion on Transmission Lines Based on Circular FBG Plane Principal Strain Sensor

As a key infrastructure for China's "West-to-East Power Transmission" project, transmission lines (TL) face the threat of ice accretion under complex microclimatic conditions. This study proposes a plane principal strain sensing method based on a fiber Bragg grating circular array, achieving synchronous monitoring of 6 strains (ranging from -2000 to 2000 με) across the TL cross-section. Through finite element simulation experiments, a mapping relationship between the bending of TL and the plane principal strain has been established. After completing the sensor calibration, an experimental platform for the freeze-thaw process of ice accretion on the TL was built. The relationships between ice mass and bending strain, as well as the ice shape on the TL cross-section (C-shaped and circular ice) and plane principal strain, were studied. Furthermore, a BP neural network model was developed to determine the 4 states of the icing process (no ice/freeze/stable/thaw), achieving an accuracy of 91.23%. This study provides effective monitoring of the freeze-thaw process of ice accretion on the TL, offering important technical support for the prevention and control of ice accretion in power grid.

physics.ins-det↗

arXiv2024

Findings of the 2024 Mandarin Stuttering Event Detection and Automatic Speech Recognition Challenge

The StutteringSpeech Challenge focuses on advancing speech technologies for people who stutter, specifically targeting Stuttering Event Detection (SED) and Automatic Speech Recognition (ASR) in Mandarin. The challenge comprises three tracks: (1) SED, which aims to develop systems for detection of stuttering events; (2) ASR, which focuses on creating robust systems for recognizing stuttered speech; and (3) Research track for innovative approaches utilizing the provided dataset. We utilizes an open-source Mandarin stuttering dataset AS-70, which has been split into new training and test sets for the challenge. This paper presents the dataset, details the challenge tracks, and analyzes the performance of the top systems, highlighting improvements in detection accuracy and reductions in recognition error rates. Our findings underscore the potential of specialized models and augmentation strategies in developing stuttered speech technologies.

eess.AS↗

arXiv2024

AutoChunk: Automated Activation Chunk for Memory-Efficient Long Sequence Inference

Large deep learning models have achieved impressive performance across a range of applications. However, their large memory requirements, including parameter memory and activation memory, have become a significant challenge for their practical serving. While existing methods mainly address parameter memory, the importance of activation memory has been overlooked. Especially for long input sequences, activation memory is expected to experience a significant exponential growth as the length of sequences increases. In this approach, we propose AutoChunk, an automatic and adaptive compiler system that efficiently reduces activation memory for long sequence inference by chunk strategies. The proposed system generates chunk plans by optimizing through multiple stages. In each stage, the chunk search pass explores all possible chunk candidates and the chunk selection pass identifies the optimal one. At runtime, AutoChunk employs code generation to automatically apply chunk strategies. The experiments demonstrate that AutoChunk can reduce over 80\% of activation memory while maintaining speed loss within 10%, extend max sequence length by 3.2x to 11.7x, and outperform state-of-the-art methods by a large margin.

cs.PF↗

arXiv2024

HeteGen: Heterogeneous Parallel Inference for Large Language Models on Resource-Constrained Devices

In recent times, the emergence of Large Language Models (LLMs) has resulted in increasingly larger model size, posing challenges for inference on low-resource devices. Prior approaches have explored offloading to facilitate low-memory inference but often suffer from efficiency due to I/O bottlenecks. To achieve low-latency LLMs inference on resource-constrained devices, we introduce HeteGen, a novel approach that presents a principled framework for heterogeneous parallel computing using CPUs and GPUs. Based on this framework, HeteGen further employs heterogeneous parallel computing and asynchronous overlap for LLMs to mitigate I/O bottlenecks. Our experiments demonstrate a substantial improvement in inference speed, surpassing state-of-the-art methods by over 317% at most.

cs.PF↗

arXiv2023

Adaptive Kalman-based hybrid car following strategy using TD3 and CACC

In autonomous driving, the hybrid strategy of deep reinforcement learning and cooperative adaptive cruise control (CACC) can fully utilize the advantages of the two algorithms and significantly improve the performance of car following. However, it is challenging for the traditional hybrid strategy based on fixed coefficients to adapt to mixed traffic flow scenarios, which may decrease the performance and even lead to accidents. To address the above problems, a hybrid car following strategy based on an adaptive Kalman Filter is proposed by regarding CACC and Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithms. Different from traditional hybrid strategy based on fixed coefficients, the Kalman gain H, using as an adaptive coefficient, is derived from multi-timestep predictions and Monte Carlo Tree Search. At the end of study, simulation results with 4157745 timesteps indicate that, compared with the TD3 and HCFS algorithms, the proposed algorithm in this study can substantially enhance the safety of car following in mixed traffic flow without compromising the comfort and efficiency.

cs.AI↗

arXiv2022

Integrated optimization of train timetables rescheduling and response vehicles on a disrupted metro line

When an unexpected metro disruption occurs, metro managers need to reschedule timetables to avoid trains going into the disruption area, and transport passengers stranded at disruption stations as quickly as possible. This paper proposes a two-stage optimization model to jointly make decisions for two tasks. In the first stage, the timetable rescheduling problem with cancellation and short-turning strategies is formulated as a mixed integer linear programming (MILP). In particular, the instantaneous parameters and variables are used to describe the accumulation of time-varying passenger flow. In the second one, a system-optimal dynamic traffic assignment (SODTA) model is employed to dynamically schedule response vehicles, which is able to capture the dynamic traffic and congestion. Numerical cases of Beijing Metro Line 9 verify the efficiency and effectiveness of our proposed model, and results show that: (1) when occurring a disruption event during peak hours, the impact on the normal timetable is greater, and passengers in the direction with fewer train services are more affected; (2) if passengers stranded at the terminal stations of disruption area are not transported in time, they will rapidly increase at a speed of more than 300 passengers per minute; (3) compared with the fixed shortest path, using the response vehicles reduces the total travel time about 7%. However, it results in increased travel time for some passengers.

eess.SY↗

arXiv2022

Structure and evolution of urban heavy truck mobility networks

Revealing the structural properties and understanding the evolutionary mechanisms of the urban heavy truck mobility network (UHTMN) provide insights in assessment of freight policies to manage and regulate the urban freight system, and are of vital importance for improving the livability and sustainability of cities. Although massive urban heavy truck mobility data become available in recent years, in-depth studies on the structure and evolution of UHTMN are still lacking. Here we use massive urban heavy truck GPS data in China to construct the UHTMN and reveal its a wide range of structure properties. We further develop an evolving network model that simultaneously considers weight, space and system element duplication. Our model reproduces the observed structure properties of UHTMN and helps us understand its underlying evolutionary mechanisms. Our model also provides new perspectives for modeling the evolution of many other real-world networks, such as protein interaction networks, citation networks and air transportation networks.

physics.soc-ph↗

arXiv2022

Stochastic factors and string stability of traffic flow: Analytical investigation and numerical study based on car-following models

The emergence dynamics of traffic instability has always attracted particular attention. For several decades, researchers have studied the stability of traffic flow using deterministic traffic models, with less emphasis on the presence of stochastic factors. However, recent empirical and theoretical findings have demonstrated that the stochastic factors tend to destabilize traffic flow and stimulate the concave growth pattern of traffic oscillations. In this paper, we derive a string stability condition of a general stochastic continuous car-following model by the mean of the generalized Lyapunov equation. We have found, indeed, that the presence of stochasticity destabilizes the traffic flow. The impact of stochasticity depends on both the sensitivity to the gap and the sensitivity to the velocity difference. Numerical simulations of three typical car-following models have been carried out to validate our theoretical analysis. Finally, we have calibrated and validated the stochastic car-following models against empirical data. It is found that the stochastic car-following models reproduce the observed traffic instability and capture the concave growth pattern of traffic oscillations. Our results further highlight theoretically and numerically that the stochastic factors have a significant impact on traffic dynamics.

physics.soc-ph↗