arXiv subjects

Jian Cao

Publications and source records attributed to Jian Cao.

At least 37 records · Page 2Linked to original sources

arXiv2026

3D Bivariate Spatial Modelling of Argo Ocean Temperature and Salinity

Variables contained within the global oceans can detect and reveal the effects of the warming climate as the oceans absorb huge amounts of solar energy. Hence, information regarding the joint spatial distribution of ocean variables is critical for climate monitoring. In this paper, we investigate the spatial correlation structure between ocean temperature and salinity using data harvested from the Argo program and construct a model to capture their bivariate spatial dependence from the surface to the ocean's interior. We develop a flexible class of multivariate nonstationary covariance models defined in 3-dimensional (3D) space (longitude $\times$ latitude $\times$ depth) that allows for the variances and correlation to change along the vertical pressure dimension. These models are able to describe the joint spatial distribution of the two variables while incorporating the underlying vertical structure of the ocean. We demonstrate that the proposed cross-covariance models describe the complex vertical cross-covariance structure well, while existing cross-covariance models including bivariate Matérn models poorly fit empirical cross-covariance structure. Furthermore, the results show that using one more variable significantly enhances the prediction of the other variable and that the estimated spatial dependence structures are consistent with the ocean stratification.

stat.AP↗

arXiv2026

Hestia: Hyperthread-Level Scheduling for Cloud Microservices with Interference-Aware Attention

Modern cloud servers routinely co-locate multiple latency-sensitive microservice instances to improve resource efficiency. However, the diversity of microservice behaviors, coupled with mutual performance interference under simultaneous multithreading (SMT), makes large-scale placement increasingly complex. Existing interference aware schedulers and isolation techniques rely on coarse core-level profiling or static resource partitioning, leaving asymmetric hyperthread-level heterogeneity and SMT contention dynamics largely unmodeled. We present Hestia, a hyperthread-level, interference-aware scheduling framework powered by self-attention. Through an extensive analysis of production traces encompassing 32,408 instances across 3,132 servers, we identify two dominant contention patterns -- sharing-core (SC) and sharing-socket (SS) -- and reveal strong asymmetry in their impact. Guided by these insights, Hestia incorporates (1) a self-attention-based CPU usage predictor that models SC/SS contention and hardware heterogeneity, and (2) an interference scoring model that estimates pairwise contention risks to guide scheduling decisions. We evaluate Hestia through large-scale simulation and a real production deployment. Hestia reduces the 95th-percentile service latency by up to 80\%, lowers overall CPU consumption by 2.3\% under the same workload, and surpasses five state-of-the-art schedulers by up to 30.65\% across diverse contention scenarios.

cs.DC↗

arXiv2026

Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture

Market making (MM) through Reinforcement Learning (RL) has attracted significant attention in financial trading. With the development of Large Language Models (LLMs), more and more attempts are being made to apply LLMs to financial areas. A simple, direct application of LLM as an agent shows significant performance. Such methods are hindered by their slow inference speed, while most of the current research has not studied LLM distillation for this specific task. To address this, we first propose the normalized fluorescent probe to study the mechanism of the LLM's feature. Based on the observation found by our investigation, we propose Cooperative Market Making (CMM), a novel framework that decouples LLM features across three orthogonal dimensions: layer, task, and data. Various student models collaboratively learn simple LLM features along with different dimensions, with each model responsible for a distinct feature to achieve knowledge distillation. Furthermore, CMM introduces an Hájek-MoE to integrate the output of the student models by investigating the contribution of different models in a kernel function-generated common feature space. Extensive experimental results on four real-world market datasets demonstrate the superiority of CMM over the current distillation method and RL-based market-making strategies.

cs.AI↗

arXiv2026

Linear-Cost Vecchia Approximation of Multivariate Normal Probabilities

Multivariate normal (MVN) probabilities arise in myriad applications, but they are analytically intractable and need to be evaluated via Monte-Carlo-based numerical integration. For the state-of-the-art minimax exponential tilting (MET) method, we show that the complexity of each of its components can be greatly reduced through an integrand parameterization that utilizes the sparse inverse Cholesky factor produced by the Vecchia approximation, whose approximation error is often negligible relative to the Monte-Carlo error. Based on this idea, we derive algorithms that can estimate MVN probabilities and sample from truncated MVN distributions in linear time (and that are easily parallelizable) at the same convergence or acceptance rate as MET, whose complexity is cubic in the dimension of the MVN probability. We showcase the advantages of our methods relative to existing approaches using several simulated examples. We also analyze a groundwater-contamination dataset with over twenty thousand censored measurements to demonstrate the scalability of our method for partially censored Gaussian-process models.

stat.CO↗

arXiv2026

Machine Learning-Driven Creep Law Discovery Across Alloy Compositional Space

Hihg-temperature creep characterization of structural alloys traditionally relies on serial uniaxial tests, which are highly inefficient for exploring the large search space of alloy compositions and for material discovery. Here, we introduce a machine-learning-assisted, high-throughput framework for creep law identification based on a dimple array bulge instrument (DABI) configuration, which enables parallel creep testing of 25 dimples, each fabricated from a different alloy, in a single experiment. Full-field surface displacements of dimples undergoing time-dependent creep-induced bulging under inert gas pressure are measured by 3D digital image correlation. We train a recurrent neural network (RNN) as a surrogate model, mapping creep parameters and loading conditions to the time-dependent deformation response of DABI. Coupling this surrogate with a particle swarm optimization scheme enables rapid and global inverse identification with sparsity regularization of creep parameters from experiment displacement-time histories. In addition, we propose a phenomenological creep law with a time-dependent stress exponent that captures the sigmoidal primary creep observed in wrought INCONEL 625 and extracts its temperature dependence from DABI test at multiple temperatures. Furthermore, we employ a general creep law combining several conventional forms together with regularized inversion to identify the creep laws for 47 additional Fe-, Ni-, and Co-rich alloys and to automatically select the dominant functional form for each alloy. This workflow combined with DABI experiment provides a quantitative, high-throughput creep characterization platform that is compatible with data mining, composition-property modeling, and nonlinear structural optimization with creep behavior across a large alloy design space.

cond-mat.mtrl-sci↗

arXiv2026

Deciphering the nature of $X(2300)$ with the PACIAE model

Inspired by the BESIII newest observation of an axial-vector particle $X(2300)$ in the $ψ(3686)\rightarrow ϕηη'$ process, we simulate its production in $e^+e^-$ collisions at $\sqrt{s}=4.95$ GeV using the parton and hadron cascade model PACIAE 4.0. In this model, the final partonic state (FPS) and hadronic state (FHS) are simulated and recorded sequentially. We propose, for the first time, that $X(2300)$ could be a $q\bar{q}s\bar{s}$ ($q=u/d$) state or a hadro-strangeonium state, i.e., a bound system of a strangeonium and a light hadron. The excited strangeonium candidate is formed by coalescing an $s\bar{s}$ quark pair in the FPS with the quantum statistical mechanics inspired dynamically constrained phase-space coalescence model. The tetraquark candidates of $q\bar{q}s\bar{s}$ and $ss\bar{s}\bar{s}$ are similarly produced by coalescing four constituent quarks in the FPS. In contrast, a hadro-strangeonium candidate emerges from the recombination of the constituent $ϕ$ and $η/η$ in the FHS. We then calculate the $X(2300)$'s orbital angular momentum quantum number in its rest frame and perform the spectral classification for each of the above candidates. Given its quantum numbers $J^{PC}=1^{+-}$, $X(2300)$ is identified as a $P$-wave $s\bar{s}$, an $S$-wave $q\bar{q}s\bar{s}/ss\bar{s}\bar{s}$ or $S$-wave $ϕη'/ϕη$ candidate. For the first time, we estimate the production rates for these configurations. The $P$-wave $s\bar{s}$ and $S$-wave $q\bar{q}s\bar{s}$ states are produced at rates on the order of $10^{-5}$, whereas the $S$-wave $ss\bar{s}\bar{s}$ and $ϕη'/ϕη$ states appear at rates on the order of $10^{-6}$. Moreover, significant discrepancies are observed in the rapidity and transverse momentum distributions among different candidates. These discrepancies could be served as valuable criteria for deciphering the nature of $X(2300)$.

hep-ph↗

arXiv2026

AP2O-Coder: Adaptively Progressive Preference Optimization for Reducing Compilation and Runtime Errors in LLM-Generated Code

LLMs' code generation capabilities have yielded substantial improvements in the effectiveness of programming tasks. However, LLM-generated code still suffers from compilation and runtime errors. Existing offline preference optimization methods primarily focus on enhancing LLMs' coding abilities using pass/fail signals in the preference data, overlooking the deep-level error types in the failed codes. To address this, we propose Adaptively Progressive Preference Optimization (AP2O) for coding (i.e., AP2O-Coder), a method that guides LLMs adaptively and methodically to reduce code errors for code generation. Specifically, we construct an error notebook from failed codes and progressively optimize the LLM to correct errors type by type. Furthermore, we adaptively replay error types to tailor to the LLM's changing weaknesses throughout the training process. Through extensive experiments on both code and general LLMs (Llama, Qwen, and DeepSeek series) with parameters ranging from 0.5B to 34B, our AP2O-Coder improves code generation performance by up to 3% in pass@k while using less preference data. Code: https://github.com/TsingZ0/AP2O

cs.SE↗

arXiv2026

GAPO: Robust Advantage Estimation for Real-World Code LLMs

Reinforcement learning (RL) is widely used for post-training large language models (LLMs) in code editing, where group-relative methods, such as GRPO, are popular due to their critic-free and normalized advantage estimation. However, in real-world code-editing scenarios, reward distributions are often skewed with unpredictable noise, leading to distorted advantage computation and increased rollout outliers. To address this issue, we propose Group Adaptive Policy Optimization (GAPO), which adaptively finds an interval with the highest SNR (Signal to Noise Ratio) per prompt and uses the median of that interval as an adaptive Q to replace the group mean in advantage calculation to reduce noise further. This adaptive Q robustly handles rollout noise while remaining plug-and-play and efficient. We evaluate GAPO on nine instruction-tuned LLMs (3B-14B) using a collected large dataset of 51,844 real-world, history-aware code-editing tasks spanning 10 programming languages. GAPO yields up to 4.35 in-domain (ID) and 5.30 out-of-domain (OOD) exact-match improvements over GRPO and its variant DAPO, while achieving lower clipping ratios and higher GPU throughput. Code: https://github.com/TsingZ0/verl-GAPO.

cs.LG↗

arXiv2025

Efficient GPU-computing simulation platform JAX-PF for differentiable phase field model

We present JAX-PF, an open-source, GPU-accelerated, and differentiable Phase Field (PF) software package, supporting both explicit and implicit time stepping schemes. Leveraging the modern computing architecture JAX, JAX-PF achieves high performance through array programming and GPU acceleration, delivering ~5x speedup over PRISMS-PF with MPI (24 CPU cores) for systems with ~4.19 million degrees of freedom using explicit schemes, and scaling efficiently with implicit schemes for large-size problems. Furthermore, a key feature of JAX-PF is automatic differentiation (AD), eliminating manual derivations of free-energy functionals and Jacobians. Beyond forward simulations, JAX-PF demonstrates its potential in inverse design by providing sensitivities for gradient-based optimization. We demonstrate, for the first time, the calibration of PF material parameters using AD-based sensitivities, highlighting its capability for high-dimensional inverse problems. By combining efficiency, flexibility, and full differentiability, JAX-PF offers a fast, practical, and integrated tool for forward simulation and inverse design, advancing co-designing of material and manufacturing processes and supporting the goals of the Materials Genome Initiative.

physics.comp-ph↗

arXiv2025

DABL: Detecting Semantic Anomalies in Business Processes Using Large Language Models

Detecting anomalies in business processes is crucial for ensuring operational success. While many existing methods rely on statistical frequency to detect anomalies, it's important to note that infrequent behavior doesn't necessarily imply undesirability. To address this challenge, detecting anomalies from a semantic viewpoint proves to be a more effective approach. However, current semantic anomaly detection methods treat a trace (i.e., process instance) as multiple event pairs, disrupting long-distance dependencies. In this paper, we introduce DABL, a novel approach for detecting semantic anomalies in business processes using large language models (LLMs). We collect 143,137 real-world process models from various domains. By generating normal traces through the playout of these process models and simulating both ordering and exclusion anomalies, we fine-tune Llama 2 using the resulting log. Through extensive experiments, we demonstrate that DABL surpasses existing state-of-the-art semantic anomaly detection methods in terms of both generalization ability and learning of given processes. Users can directly apply DABL to detect semantic anomalies in their own datasets without the need for additional training. Furthermore, DABL offers the capability to interpret the causes of anomalies in natural language, providing valuable insights into the detected anomalies.

cs.CL↗

arXiv2025

LLM-SrcLog: Towards Proactive and Unified Log Template Extraction via Large Language Models

Log parsing transforms raw logs into structured templates containing constants and variables. It underpins anomaly detection, failure diagnosis, and other AIOps tasks. Current parsers are mostly reactive and log-centric. They only infer templates from logs, mostly overlooking the source code. This restricts their capacity to grasp dynamic log structures or adjust to evolving systems. Moreover, per-log LLM inference is too costly for practical deployment. In this paper, we propose LLM-SrcLog, a proactive and unified framework for log template parsing. It extracts templates directly from source code prior to deployment and supplements them with data-driven parsing for logs without available code. LLM-SrcLog integrates a cross-function static code analyzer to reconstruct meaningful logging contexts, an LLM-based white-box template extractor with post-processing to distinguish constants from variables, and a black-box template extractor that incorporates data-driven clustering for remaining unmatched logs. Experiments on two public benchmarks (Hadoop and Zookeeper) and a large-scale industrial system (Sunfire-Compute) show that, compared to two LLM-based baselines, LLM-SrcLog improves average F1-score by 2-17% and 8-35%. Meanwhile, its online parsing latency is comparable to data-driven methods and about 1,000 times faster than per-log LLM parsing. LLM-SrcLog achieves a near-ideal balance between speed and accuracy. Finally, we further validate the effectiveness of LLM-SrcLog through practical case studies in a real-world production environment.

cs.SE↗

arXiv2025

Adaptive Digital Twin of Sheet Metal Forming via Proper Orthogonal Decomposition-Based Koopman Operator with Model Predictive Control

Digital Twin (DT) technologies are transforming manufacturing by enabling real-time prediction, monitoring, and control of complex processes. Yet, applying DT to deformation-based metal forming remains challenging because of the strongly coupled spatial-temporal behavior and the nonlinear relationship between toolpath and material response. For instance, sheet-metal forming by the English wheel, a highly flexible but artisan-dependent process, still lacks digital counterparts that can autonomously plan and adapt forming strategies. This study presents an adaptive DT framework that integrates Proper Orthogonal Decomposition (POD) for physics-aware dimensionality reduction with a Koopman operator for representing nonlinear system in a linear lifted space for the real-time decision-making via model predictive control (MPC). To accommodate evolving process conditions or material states, an online Recursive Least Squares (RLS) algorithm is introduced to update the operator coefficients in real time, enabling continuous adaptation of the DT model as new deformation data become available. The proposed framework is experimentally demonstrated on a robotic English Wheel sheet metal forming system, where deformation fields are measured and modeled under varying toolpaths. Results show that the adaptive DT is capable of controlling the forming process to achieve the given target shape by effectively capturing non-stationary process behaviors. Beyond this case study, the proposed framework establishes a generalizable approach for interpretable, adaptive, and computationally-efficient DT of nonlinear manufacturing systems, bridging reduced-order physics representations with data-driven adaptability to support autonomous process control and optimization.

eess.SY↗

arXiv2025

First GNSS-deployed optical clock for local time scale upgrade

Precise time scale is the universal base for all measurements. Here we report the deployment of a compact and transportable optical clock to a timekeeping institution and steering an active hydrogen maser to generate an optical time scale, realizing the upgrade of the local time scale in the Global Navigation Satellite System. The optical clock was transported over 1200 km by express delivery and resume work as normal promptly, and its extremely high uptime of 93.6% in the half-year enabled us to precisely correct the frequency drift of hydrogen maser, ultimately achieving an unprecedented monthly instability of 4E-17. This steering experiment with a deployable optical clock marks a significant advancement, demonstrating that a timing accuracy below 100 ps per month can be achieved feasibly in various timekeeping institutions where hydrogen masers are typically employed as the primary contributor to timekeeping. In the future, mobile optical time scale based on such transportable optical clock can be deployed flexibly and rapidly, which is particularly important in scenarios lacking International Atomic Time reference.

physics.atom-ph↗

arXiv2025

Towards a compact transportable optical clock based on the octupole transition in 171Yb+

Optical clocks have extremely attractive applications in many fields, including time-frequency metrology, validation of fundamental physical principles, and relativistic geodesy. The 467 nm octupole transition in 171Yb+ ion exhibits intrinsic insensitivity to magnetic field and an ultra-long clock state lifetime of 1.6 years. In addition, the entire laser system can be realized by semiconductor technologies, rendering this platform uniquely advantageous for developing high-precision, compact and transportable optical clocks. Here, we report the development of a compact optical clock based on the 467 nm transition of a single 171Yb+ ion. Using a narrow linewidth 467 nm laser to interrogate the clock transition, we obtain a near-Fourier-limited linewidth of 2.3 Hz in an integrated ion trapping system. Self-comparison demonstrated a frequency instability of 2.2E-15/sqrt(tau/s) with an interrogation time of 180 ms, which reaches the high parts in E-18 level with an averaging time of only one day. These work laid the technical foundation for the subsequent clock systematic evaluation and the packaging of each subsystem into an engineering prototype with high-precision at the level of E-18.

physics.atom-ph↗

arXiv2025

QCD phase transition at finite temperature and chemical potential with the non-extensive statistics

The intrinsic fluctuations, memory effects and long-range color interactions in high energy nuclear collisions imply the presence of non-Markovian processes in the fireball evolution, which affects the thermalization process towards equilibrium and produces a non-extensive behavior. In order to investigate the non-equilibrium effect on the quantum chromodynamics (QCD) phase transition at finite temperature ($T$) and chemical potential ($μ$), we apply a non-extensive correction to the equation of state in the parton (hadron resonance) gas at high (low) temperature and interpolate these two equation of states with a smooth crossover. The non-extensive statistics is characterized by a non-extensivity parameter $q$, which measures the degrees of deviation from the thermal equilibrium. It is found that the dimensionless thermodynamic quantities such as the entropy density, the pressure, the energy density, the specific heat at constant volume and the trace anomaly are sensitive to the deviation of $q$ from unity and they become large both in the hadronic and quark-gluon plasma phases with the increase of $q$. Moreover, this deviation leads to nontrivial corrections of the squared speed of sound ($(c_s^2)_q$) in the vicinity of the critical point ($T_c$) and at lower temperatures. Additionally, these thermodynamic quantities are sensitive to the deviation of $μ$ from zero. With increasing $μ$, they become enhanced in both phases. Specifically, for $(c_s^2)_q$, the value increases near $T_c$ but decreases at lower temperatures. Finally, we observe that our results with $q=1$ agree well with those from the Lattice QCD, the hadron resonance gas model, and the Thermal-Fist fit to the hadron yields in high energy nuclear collisions in the low temperature region up to $T\sim 150$ MeV.

hep-ph↗

arXiv2025

GFS: A Preemption-aware Scheduling Framework for GPU Clusters with Predictive Spot Instance Management

The surge in large language models (LLMs) has fundamentally reshaped the landscape of GPU usage patterns, creating an urgent need for more efficient management strategies. While cloud providers employ spot instances to reduce costs for low-priority (LP) tasks, existing schedulers still grapple with high eviction rates and lengthy queuing times. To address these limitations, we present GFS, a novel preemptive scheduling framework that enhances service-level objective (SLO) compliance for high-priority (HP) tasks while minimizing preemptions to LP tasks. Firstly, GFS utilizes a lightweight forecasting model that predicts GPU demand among different tenants, enabling proactive resource management. Secondly, GFS employs a dynamic allocation mechanism to adjust the spot quota for LP tasks with guaranteed durations. Lastly, GFS incorporates a preemptive scheduling policy that prioritizes HP tasks while minimizing the impact on LP tasks. We demonstrate the effectiveness of GFS through both real-world implementation and simulations. The results show that GFS reduces eviction rates by 33.0\%, and cuts queuing delays by 44.1\% for LP tasks. Furthermore, GFS enhances the GPU allocation rate by up to 22.8\% in real production clusters. In a production cluster of more than 10,000 GPUs, GFS yields roughly \$459,715 in monthly benefits.

cs.DC↗

arXiv2025

OmniTry: Virtual Try-On Anything without Masks

Virtual Try-ON (VTON) is a practical and widely-applied task, for which most of existing works focus on clothes. This paper presents OmniTry, a unified framework that extends VTON beyond garment to encompass any wearable objects, e.g., jewelries and accessories, with mask-free setting for more practical application. When extending to various types of objects, data curation is challenging for obtaining paired images, i.e., the object image and the corresponding try-on result. To tackle this problem, we propose a two-staged pipeline: For the first stage, we leverage large-scale unpaired images, i.e., portraits with any wearable items, to train the model for mask-free localization. Specifically, we repurpose the inpainting model to automatically draw objects in suitable positions given an empty mask. For the second stage, the model is further fine-tuned with paired images to transfer the consistency of object appearance. We observed that the model after the first stage shows quick convergence even with few paired samples. OmniTry is evaluated on a comprehensive benchmark consisting of 12 common classes of wearable objects, with both in-shop and in-the-wild images. Experimental results suggest that OmniTry shows better performance on both object localization and ID-preservation compared with existing methods. The code, model weights, and evaluation benchmark of OmniTry will be made publicly available at https://omnitry.github.io/.

cs.CV↗

arXiv2025

TOBACO: Topology Optimization via Band-limited Coordinate Networks for Compositionally Graded Alloys

Compositionally Graded Alloys (CGAs) offer unprecedented design flexibility by enabling spatial variations in composition; tailoring material properties to local loading conditions. This flexibility leads to components that are stronger, lighter, and more cost-effective than traditional monolithic counterparts. The fabrication of CGAs have become increasingly feasible owing to recent advancements in additive manufacturing (AM), particularly in multi-material printing and improved precision in material deposition. However, AM of CGAs requires imposition of manufacturing constraints; in particular limits on the maximum spatial gradation of composition. This paper introduces a topology optimization (TO) based framework for designing optimized CGA components with controlled compositional gradation. In particular, we represent the constrained composition distribution using a band-limited coordinate neural network. By regulating the network's bandwidth, we ensure implicit compliance with gradation limits, eliminating the need for explicit constraints. The proposed approach also benefits from the inherent advantages of TO using coordinate networks, including mesh independence, high-resolution design extraction, and end-to-end differentiability. The effectiveness of our framework is demonstrated through various elastic and thermo-elastic TO examples.

cs.CE↗