Searcharxiv⌕ Search

arXiv subjects

Yong Zhang

Publications and source records attributed to Yong Zhang.

At least 73 records · Page 4Linked to original sources
arXiv2026

WarmServe: Enabling One-for-Many GPU Prewarming for Multi-LLM Serving

Deploying multiple models within shared GPU clusters is a key strategy to improve resource efficiency in large language model (LLM) serving. Existing multi-LLM serving systems improve GPU utilization at the cost of degraded inference performance, particularly time-to-first-token (TTFT). We attribute this degradation to the lack of awareness regarding future workload characteristics. In contrast, recent analyses have shown the strong periodicity and long-term predictability of real-world LLM serving workloads. In this paper, we propose one-for-many GPU prewarming, which proactively loads parameters from multiple models onto GPUs based on workload forecasts. These prewarmed weights enable the system to promptly instantiate serving instances upon encountering request bursts. We design and implement WarmServe, a multi-LLM serving system incorporating three key techniques: (1) a model placement algorithm that optimizes prewarming decisions to minimize cross-model prewarming interference, (2) a KV cache reservation strategy that repurposes idle KV cache space on running GPUs for prewarming new models, and (3) an efficient GPU memory switching mechanism for tensor management. Evaluation on real-world datasets shows that WarmServe reduces tail TTFT by up to 50.8$\times$ compared to the state-of-the-art autoscaling-based system, while supporting up to 2.5$\times$ higher request throughput than the GPU-sharing system.

cs.DC↗
arXiv2026

A Boundary-Aware Non-parametric Granular-Ball Classifier Based on Minimum Description Length

Existing granular-ball classification methods are often driven by handcrafted quality measures, neighborhood rules, or heuristic splitting and stopping criteria, which may reduce the transparency of local construction decisions and hinder explicit modeling of boundary-sensitive regions. To address this issue, this paper proposes a Minimum Description Length based Granular-Ball Classifier (MDL-GBC), a boundary-aware non-parametric and interpretable granular-ball classifier. MDL-GBC formulates class-conditional granular-ball construction as a local model selection problem under the Minimum Description Length principle. For each class, samples from the target class provide positive class evidence, while samples from the remaining classes provide negative boundary evidence. For each current granular ball, three candidate explanations are compared under a unified description-length criterion: a single-ball model, a two-ball model, and a core-boundary model. The selected model determines whether the ball is retained, geometrically split, or refined into core and boundary-sensitive child balls, thereby making local construction decisions consistent with the MDL-based classification mechanism. During prediction, a class-level mixture coding rule aggregates stable granular balls of the same class and assigns the test sample by comparing class-wise coding costs. Experiments on 18 benchmark datasets show that MDL-GBC achieves competitive classification performance against classical classifiers and representative granular-ball-based methods, obtaining the best average Accuracy, Macro-F1, and average rank. These results indicate that MDL-GBC provides an effective and interpretable alternative to conventional heuristic granular-ball classification strategies.

cs.LG↗
arXiv2026

Game-Theoretic Modeling of Heterogeneous Investor Interactions for Stock Price Forecasting

Accurate stock price forecasting has consistently remained a pivotal yet challenging FinTech task that underpins quantitative trading and investment decision making. Recent efforts have been dedicated to modeling various complex relationships among stocks in the stock market toward more reliable stock price forecasting.These methods depend heavily on strong static prior assumptions by modeling either temporal dependencies within individual stocks or spatial dependencies across different stocks based on predefined structures, while the complex market dynamics that drive stock price movements remain unexplored. To alleviate this issue, we propose a novel game-theoretic modeling method that captures heterogeneous investor interactions for stock price forecasting. The core idea is to embed game-theoretic mechanisms into the heterogeneous graph structure to finely model the dynamic strategic interactions among heterogeneous investors with respect to target stocks. Additionally, temporal positional encoding is adopted to reflect the differentiated influences of each game event at different time steps within the time window on future stock price movements. Leveraging heterogeneous graph networks, we proxy the intricate dynamics of the stock market through investor games and enable real-time information propagation and node updates among all nodes. Extensive experiments conducted on two real-world benchmark dataset demonstrate that our method effectively outperforms state-of-the-art stock price forecasting methods.

q-fin.TR↗
arXiv2026

The Influence of Aliphatic Components on the Aromatic Emission Characteristics of Polycyclic Aromatic Hydrocarbons

Intensity ratios of aromatic emission features are widely used to diagnose the size and ionization state of polycyclic aromatic hydrocarbons (PAHs) in astronomical environments. However, PAHs are known to typically carry aliphatic side chains, a structural feature that may compromise the reliability of traditional diagnostic methods. This study systematically investigates the effects of aliphatic components on the aromatic emission properties of PAHs. Based on theoretical data from the NASA Ames PAH IR Spectroscopic Database, we compare the emission behavior of purely aromatic PAHs with those containing aliphatic substituents, revealing that aliphatic functionalization may modify the intensity ratio of the 11.2 $μ$m band relative to the 7.7 $μ$m and 3.3 $μ$m bands. This potentially leads to misidentification of their ionization state if molecular structural effects are neglected. Further analysis indicates that the impact of aliphatic components on diagnostic band ratios strongly depends on PAH size: small PAHs exhibit significant emission ratio shifts, deviating from traditional size/ionization trends, while larger PAHs are minimally affected. Despite these shifts, the classic $(I_{11.2/7.7})$ versus $(I_{11.2/3.3})$ diagnostic grid remains largely applicable to mixed aromatic-aliphatic PAHs, although some systematic calibration may be needed. Our findings emphasize the necessity for caution when interpreting PAH band ratios in aliphatic-rich environments, as variations in PAH molecular composition may distort inferences about physical conditions.

astro-ph.GA↗
arXiv2026

MARLaaS: Multi-Tenant Asynchronous Reinforcement Learning as a Service

Reinforcement Learning from Verifiable Rewards (RLVR) has significantly improved the reasoning capabilities of large language models (LLMs), particularly in multi-turn agentic settings involving environment interaction like tool use. However, fine-tuning such models remains prohibitively expensive due to high computational requirements, limiting accessibility. We propose MARLaaS (Multi-tenant Asynchronous RL as a Service), a system for concurrent RL fine-tuning across multiple users and tasks. Our approach is based on two key ideas: (1) sharing a base model across tenants using lightweight LoRA adapters, and (2) a disaggregated asynchronous architecture that decouples rollout generation, environment interaction, and policy training into independently scheduled stages. This design enables tasks to progress through the RL pipeline at their own pace in an event-driven manner, reducing cross-task interference, idle time, and end-to-end latency. In multi-task settings (we report up to 32 concurrent tasks), MARLaaS achieves single-task state-of-the-art performance while improving accelerator utilization by up to 4.3x and reducing end-to-end training time by 85%.

cs.DC↗
arXiv2026

Squeezed spectra and back-to-back correlations of protons and antiprotons at RHIC energies

This study constrains the range of in-medium mass modification through a comparison of theoretical calculations with experimental transverse momentum spectra and the yield ratio {\bar{p}}/p of protons and antiprotons. Based on the constrained range and a Gaussian source model with radial ow, the theoretical predictions for the fermion back-to-back correlation (fBBC) of p{\bar{p}} pairs at RHIC energies are presented. The results reveal a strong sensitivity of the fBBC signal to the assumed source time distribution: a Lorentzian form generates a pronounced high-momentum signal, whereas an α-stable Lévy form leads to a marked low-momentum signal. Moreover, the in-medium mass modification is shown to enhance the yield ratio {\bar{p}}/p. Therefore, events characterized by a larger {\bar{p}}/p ratio are predicted to have a significantly higher probability of exhibiting a detectable fBBC signal. This study may propose a promising new direction for the experimental observation of this phenomenon

hep-ph↗
arXiv2026

A Breast Vision Pathology Foundation Model for Real-world Clinical Utility

Pathology foundation models have shown strong retrospective performance, but whether such systems can support clinically relevant use remains unclear. This challenge is particularly important in breast cancer, where pathological assessment serves as the gold standard for diagnosis and guides treatment planning, surgical decision-making and risk stratification across pre-, intra- and post-operative stages. Here we present \textbf{BRAVE}, a breast-adaptive pathology foundation model developed and evaluated using a total resource of 101,638 breast whole-slide images from 32 sources across Asia, Europe and North America. We assessed BRAVE across 34 tasks in 82 cohorts spanning pre-operative biopsy, intra-operative frozen section and post-operative resection, using an evidence chain comprising retrospective benchmarking, clinically challenging scenarios, workflow-oriented clinical impact simulations, prospective observational validation with the thresholds locked in the retrospective cohorts and crossover pathologist-AI interaction studies. Across these settings, BRAVE supported practical roles in the clinical workflow, including safe exclusion of low-risk cases from routine review, AI-assisted second-review rescue of initially missed positives and prioritization of cases for further assessment. In prospective validation across three centres, BRAVE excluded 76.9% of negative biopsy cases (NPV 0.953) and 70.1% of negative frozen-section cases (NPV 0.973), and triaged 78.8% of post-operative subtyping cases as high-confidence clear-cut cases (NPV 1.000). In reader studies, AI assistance improved balanced accuracy from 88.5% to 95.1% (OR 3.14, P<0.001), with better efficiency, confidence and inter-rater agreement. BRAVE-derived scores also independently predicted disease-free survival (adjusted HR 4.79, P<0.001) and overall survival (adjusted HR 8.14, P<0.001).

cs.CV↗
arXiv2026

TurboTalk: Progressive Distillation for One-Step Audio-Driven Talking Avatar Generation

Existing audio-driven video digital human generation models rely on multi-step denoising, resulting in substantial computational overhead that severely limits their deployment in real-world settings. While one-step distillation approaches can significantly accelerate inference, they often suffer from training instability. To address this challenge, we propose TurboTalk, a two-stage progressive distillation framework that effectively compresses a multi-step audio-driven video diffusion model into a single-step generator. We first adopt Distribution Matching Distillation to obtain a strong and stable 4-step student, and then progressively reduce the denoising steps from 4 to 1 through adversarial distillation. To ensure stable training under extreme step reduction, we introduce a progressive timestep sampling strategy and a self-compare adversarial objective that provides an intermediate adversarial reference that stabilizes progressive distillation. Our method achieve single-step generation of video talking avatar, boosting inference speed by 120 times while maintaining high generation quality.

cs.CV↗
arXiv2026

Large-Amplitude Steady Electrohydrodynamic Solitary Waves with Constant Vorticity

This paper investigates solitary water waves propagating along the surface of a two-dimensional dielectric fluid with constant vorticity in the presence of an external electric field. We formulate the system as a nonlinear free boundary problem where the Euler equations and electric potential equations are strongly coupled at the interface. A major challenge in such setting is the loss of standard monotonicity arguments due to the interaction between the velocity and electric fields. We overcome this difficulty by establishing new nodal properties for the combined system, ensuring the wave remains a symmetric elevation profile along the global branch. Moreover, along the global bifurcation curve, one of the following case must occur: (i) the formation of an equilibrium stagnation point, (ii) the degeneration of the conformal mapping, (iii) the onset of flow stagnation, or (iv) an unbounded increase in the dimensionless wave speed.

math.AP↗
arXiv2026

GeoFunFlow-3D: A Physics-Guided Generative Flow Matching Framework for High-Fidelity 3D Aerodynamic Inference over Complex Geometries

Deep generative models and neural operators have demonstrated significant potential for 3D aerodynamic inference. However, they often face inherent challenges in maintaining physical consistency and preserving high-frequency features, primarily due to spectral bias and gradient conflicts within the governing equations. To address these issues, we propose GeoFunFlow-3D, a physics-guided generative flow matching framework. Temporally, we utilize optimal transport theory to build the generation path, ensuring stable training dynamics. Spectrally, we introduce a high-order discrete engine without automatic differentiation (No-AD) to reduce gradient stiffness. Spatially, a topology-aware super-resolution module (SATO) is employed to rigorously enforce physical laws in localized regions such as shock waves. We evaluated our framework on complex industrial datasets. On the BlendedNet dataset, the model successfully avoids mode collapse even under sparse data conditions. For the NASA Rotor37 test, it accurately captures 3D detached shock structures. Compared to conventional operators, GeoFunFlow-3D significantly improves accuracy, reducing the pressure field error (RRMSE) to 0.0215 while maintaining competitive inference efficiency. Ultimately, this work provides a reliable, geometry-driven approach for generating high-dimensional fluid fields.

math.NA↗
arXiv2026

HFX: Joint Design of Algorithms and Systems for Multi-SLO Serving and Fast Scaling

Large language model (LLM) serving faces the dual challenge of meeting strict user-specific service-level objectives (SLOs) while minimizing computational cost under dynamic, multi-task workloads. Existing approaches either rely on static scheduling policies or focus on single-task settings, limiting their applicability in real-world deployments with heterogeneous requests, variable prompt lengths, and elastic scaling requirements. We present HFX, a production LLM serving system that jointly optimizes request scheduling and elastic scaling across model replicas to satisfy diverse SLOs. HFX introduces a \textbf{scheduler} that performs proactive budget estimation and prioritization to ensure SLO compliance for both new and in-flight requests. HFX also integrates a \textbf{scaler} that supports fast device-to-device (D2D) weight transfer, reducing cold-start latency. Additionally, the system supports both colocated and disaggregated prefill/decode deployments, enabling adaptation to diverse workload patterns and cloud environments. Through extensive experiments on multi-task workloads, we demonstrate consistently higher SLO attainment, lower end-to-end latency, and lower NPU usage cost by up to 4.44$\times$, 65.82\%, and 49.81\%, respectively, compared to state-of-the-art systems. Our results highlight the effectiveness of SLO-aware scheduling and scaling in practical LLM serving, providing a robust framework for cost-efficient and SLO-compliant deployments.

cs.DC↗
arXiv2026

AnalogMaster: Large Language Model-based Automated Analog IC Design Framework from Image to Layout

Design automation has the potential to substantially improve the efficiency of analog integrated circuit (IC) design. However, existing algorithms and tools typically focus on individual stages, such as device sizing, placement, or routing, and still require significant manual intervention to complete the full design flow. While large language models (LLMs) have recently demonstrated remarkable success in automating digital IC design workflows, these advances cannot be directly transferred to analog IC design. Key challenges include strongly coupled performance metrics, the predominance of unstructured circuit schematic images, and the fact that most prior approaches address only isolated stages of the analog design process, limiting their ability to capture end-to-end performance impact. To address these challenges, we propose AnalogMaster, an extensible, LLM-based framework that enables end-to-end automation of analog IC design through a unified pipeline spanning circuit image-to-netlist generation, parameter optimization, placement, and routing. AnalogMaster integrates a joint reasoning mechanism that leverages in-context learning and intent reasoning to achieve accurate and robust image-to-netlist conversion. A parameter search agent integrating self-enhanced prompt engineering and context truncation is developed for effective device sizing and downstream physical design. Experimental evaluations on 15 representative circuits with varying levels of complexity demonstrate strong and consistent performance across multiple models. In particular, GPT-5 achieves success rates of 92.9% and 99.9% on Pass@1 and Pass@5, respectively. These results validate the effectiveness and robustness of the proposed framework and establish a practical paradigm for applying LLMs to full-stack analog IC design automation.

cs.AR↗
arXiv2026

Quantum-Classical Hybrid Computation of Electron Transfer in a Cryptochrome Protein via VQE-PDFT and Multiscale Modeling

Accurate calculation of strongly correlated electronic systems requires proper treatment of both static and dynamic correlations, which remains challenging for conventional methods. To address this, we present VQE-PDFT,aquantum-classical hybrid framework that integrates variational quantum eigensolver with multiconfiguration pair-density functional theory (MC-PDFT). This framework strategically employs quantum circuits for multiconfigurational wavefunction representation while utilizing density functionals for correlation energy evaluation. The hybrid strategy maintains accurate treatment of static and dynamic correlations while reducing quantum resource requirements compared to highly expressive quantum algorithms. Benchmark validation, performed via noiseless quantum circuit simulator, on the Charge-Transfer dataset confirmed that VQE-PDFT achieved results comparable to conventional MC-PDFT. Building upon this, we developed shallow-depth hardware-efficient ansatz circuits and integrated them into a QM/MM multiscale architecture to enable applications in complex biological systems. This extended framework, when applied to electron transfer in the European robin cryptochrome protein ErCRY4 with noiseless simulations, yielded transfer rates that aligned well with experimental measurements. Finally, as a proof-of-concept hardware demonstration, we executed the reduced-density-matrix measurements for a single protein conformation on a 13-qubit superconducting device and showed the impact of noise through a comprehensive error analysis.

quant-ph↗
arXiv2026

High-power-handling ultra-compact acousto-optic modulators using one-dimensional topological interface states on thin-film lithium tantalate

Recent advances in integrated photonics have enabled on-chip signal modulation and processing through localized photon-phonon interactions. For acousto-optic devices, compact footprint and high efficiency are essential for dense integration, while strong power handling is critical for stable operation in demanding applications. However, it remains challenging to achieve these features simultaneously on existing integrated platforms. Here, we propose and experimentally demonstrate, for the first time on a thin-film lithium tantalate platform, an ultra-compact acousto-optic modulator based on topological interface states. Benefiting from the strong optical confinement of the topological boundary state, the device achieves a footprint of 0.13 by 0.12 mm2 and a half-wave voltage-length product of 0.491 Vcm. We further demonstrate stable acousto-optic modulation at an on-chip optical power of up to 28 dBm (630.9 mW), highlighting the strong power-handling capability of the thin-film lithium tantalate topological structure. This work provides a compact and high-power solution for microwave-to-photonic transduction and shows the potential of the thin-film lithium tantalate for robust integrated photonic systems.

physics.optics↗
arXiv2026

Analysis of Eruptive Prominence Plasma Parameters' Effects on the \ion{He}{2} 304~Å Line with Solar Orbiter EUI Observations

An observation of a large prominence on the solar limb took place on February 15, 2022, by the Extreme Ultraviolet Imager (EUI) on board Solar Orbiter. We aim to determine the range of physical parameters of this prominence, such as temperature, radial velocity, and altitude, and examine how these parameters affect the formation of the 304~Å line of \ion{He}{2}, especially how collisional excitation and radiative processes contribute to line formation. After constraining these parameters, we generate 200 random models and compute the \ion{He}{2} 304~Å line profile. We present these results using parallel coordinate plots to explore how these parameters affect the results. This allows us to infer the key physical parameters that impact the formation of the \ion{He}{2} 304~Å line. This study demonstrates that column mass and the steepness of the temperature profile are key factors in the formation of the \ion{He}{2} 304~Å line during the solar prominence eruption on February 15, 2022. Radiative processes remain dominant in the formation of the \ion{He}{2} 304~Å line. These insights provide a foundation for future research and comparative studies.

astro-ph.SR↗
arXiv2026

Analysis of Lyman-beta and Lyman-gamma Lines in a Pre-Eruptive and Eruptive Prominence with Solar Orbiter SPICE Observations

The first dedicated observation of an off-limb prominence by the Spectral Imaging of the Coronal Environment (SPICE) instrument on board Solar Orbiter took place on April 15, 2023. Our aim is to provide an overview of the potentiality of the diagnostics using these data. We show that we can derive the changes in the physical parameters of the pre-eruptive and eruptive prominence using the Lyman lines. We investigate the integrated intensity and line widths of the Lyman $β$ and Lyman $γ$ lines, finding variations between the prominence, disk, and coronal regions. The results reflect dynamic changes in density, temperature, and optical thickness. We analyze the spatial and temporal evolution of the Lyman $β$ and Lyman $γ$ line profiles. Using a simple geometric model, we obtain the radial velocity of this prominence at the early phase of its eruption with GONG H$α$ images. This offers a way of calculating the radial velocity of an eruptive filament from a pair of 2D images. The result helps us understand the potential Doppler effect in line profiles. Overall, the spectral profiles indicate that the eruption enhances spatial and temporal variations in line intensity, reflecting dynamic changes in plasma conditions within the prominence. These findings highlight the diagnostic potential of SPICE observations, and future Non-LTE radiative transfer modeling will help to further constrain prominence plasma parameters.

astro-ph.SR↗
arXiv2026

Non-LTE Analysis of Pre-eruptive Prominence Plasma Parameters Effects on the Lyman-beta and Lyman-gamma Lines with Solar Orbiter SPICE Observations

The first dedicated observation of an off-limb prominence by Solar Orbiter took place on April 15, 2023. Our aim is to determine the range of different physical parameters of this prominence and to examine how these parameters affect the formation of the Lyman $β$ and Lyman $γ$ lines of hydrogen. We have found a way to refine key physical parameters by observational data. We will test the method by this prominence observation. We generate 200 random non-LTE models using these observational constraints and compute the Lyman $β$ line and the Lyman $γ$ line profiles. We use the Spectral Imaging of the Coronal Environment (SPICE) full-disk mosaic from November 13, 2023 to constrain the incident radiation. We present the parameters and results of 200 random models using parallel coordinate plots to explore how different parameters affect the results. This allows us to infer the key physical parameters (e.g., central pressure, column mass and temperature gradient) that impact the formation of the Lyman $β$ line and the Lyman $γ$ line in this observation.

astro-ph.SR↗
arXiv2026

TAMISeg: Text-Aligned Multi-scale Medical Image Segmentation with Semantic Encoder Distillation

Medical image segmentation remains challenging due to limited fine-grained annotations, complex anatomical structures, and image degradation from noise, low contrast, or illumination variation. We propose TAMISeg, a text-guided segmentation framework that incorporates clinical language prompts and semantic distillation as auxiliary semantic cues to enhance visual understanding and reduce reliance on pixel-level fine-grained annotations. TAMISeg integrates three core components: a consistency-aware encoder pretrained with strong perturbations for robust feature extraction, a semantic encoder distillation module with supervision from a frozen DINOv3 teacher to enhance semantic discriminability, and a scale-adaptive decoder that segments anatomical structures across different spatial scales. Experiments on the Kvasir-SEG, MosMedData+, and QaTa-COV19 datasets demonstrate that TAMISeg consistently outperforms existing uni-modal and multi-modal methods in both qualitative and quantitative evaluations. Code will be made publicly available at https://github.com/qczggaoqiang/TAMISeg.

cs.CV↗