Searcharxiv⌕ Search

arXiv subjects

Li Wang

Publications and source records attributed to Li Wang.

At least 55 records · Page 3Linked to original sources
arXiv2026

Homogenization of Time-Discrete Gradient Flows

In this work, we study the homogenization limit for Fokker--Planck equations on the flat torus with rapidly oscillating diffusion coefficients at the time-discrete level. We discretize the equation in time using a minimizing movement scheme and compare two choices of metric: a transport-type metric and a weighted L2 metric. We prove that this choice is crucial: the two schemes lead to different homogenization limits as the scaling parameter tends to zero, and consequently to different evolution equations as the time step is sent to zero. In particular, only the scheme based on the weighted L2 metric recovers the correct effective dynamics.

math.AP↗
arXiv2026

Isolated but Exposed: Persistence-Based Memory Extraction Attack on LLM Agents

LLM-based agents extend large language models with long-term memory (LTM) that persists privacy-sensitive user data across sessions. Production systems mitigate extraction risks through memory isolation, binding each user's LTM to a unique identifier. This defense has blocked known attacks on shared storage, fostering the assumption that isolated LTM is secure. We identify the tool interface as an overlooked attack surface. Agents routinely embed LTM-retrieved data in tool invocation parameters, enabling a malicious tool to exfiltrate private memory without violating user-level isolation. Naive adaptations of user-side extraction techniques fail because the adversarial command's semantics interfere with retrieval precision, and platform-imposed tool-call limits constrain the extraction budget per trigger. We present SPORE, the first extraction attack designed for this threat model. SPORE decouples the adversarial command from retrieval anchors by persisting the command in short-term memory and emitting semantically pure anchors in tool responses. The restored retrieval precision enables a geometric coverage optimization over the embedding space that systematically steers anchors toward unexplored memory regions. To sustain extraction beyond tool-call limits, SPORE persists reactivation payloads in memory that automatically resume the attack within and across sessions without additional user triggers. SPORE achieves an 80.0% record extraction rate with unlimited triggers and 47.0% with only 20 triggers. In multi-user deployments, attackers can link extracted records to user identities, enabling targeted surveillance. These results demonstrate that memory isolation alone is insufficient and call for reexamining tool-side trust boundaries in agent architectures.

cs.CR↗
arXiv2026

Geometry-Induced Vacuum Polarization and Mode Shifts in Maxwell-Klein-Gordon Theory

Geometric confinement is known to modify single-particle dynamics through effective potentials, yet its imprint on the interacting quantum vacuum remains largely unexplored. In this work, we investigate the Maxwell--Klein--Gordon system constrained to curved surfaces and demonstrate that the geometric potential $Σ_{\mathrm{geom}}(\mathbf{r})$ acts as a local renormalization environment. We show that extrinsic curvature modifies the scalar loop spectrum, entering the vacuum polarization as a position-dependent mass correction $M^2(\mathbf{r}) \to m^2 + Σ_{\mathrm{geom}}(\mathbf{r})$. This induces a finite, gauge-invariant ``geometry-induced running'' of the electromagnetic response. In the long-wavelength regime ($|{\bf Q}|R \ll 1$), we derive a closed-form expression for the relative frequency shift $Δω/ω$, governed by the overlap between the electric energy density and the geometric potential. Applying this formalism to Gaussian bumps, cylindrical shells, and tori, we identify distinct spectral signatures that distinguish these quantum loop corrections from classical geometric optics. Our results suggest that spatial curvature can serve as a tunable knob for ``vacuum engineering,'' offering measurable shifts in high-$Q$ cavities and plasmonic systems.

physics.optics↗
arXiv2026

PAGE-RAG: Evidence-Grounded Adaptive Graph Retrieval for Long-Document Question Answering

GraphRAG improves long-document question answering by introducing structured representations beyond conventional retrieval. However, automatically constructed graphs are inherently incomplete projections of source documents, and treating them as independent knowledge sources may lead to unreliable retrieval and generation. We propose PAGE-RAG, a projection-aware adaptive graph retrieval framework for reliable long-document question answering. PAGE-RAG views graph structures as semantic skeletons that organize and navigate document knowledge, rather than replacing the original knowledge source. Based on this perspective, PAGE-RAG introduces a task-adaptive retrieval routing strategy that dynamically selects appropriate retrieval behaviors according to query requirements. Furthermore, PAGE-RAG incorporates strict knowledge boundary control, ensuring that generated responses remain grounded within available evidence and abstaining from unsupported information beyond the accessible knowledge scope. Experiments demonstrate that PAGE-RAG achieves competitive answer quality while improving retrieval efficiency and knowledge reliability, highlighting the importance of projection-aware graph modeling, adaptive retrieval, and explicit knowledge boundary control for trustworthy GraphRAG systems. The source code is publicly available at https://github.com/CXY0112/PAGE-RAG.

cs.IR↗
arXiv2026

Non-invasive monitoring of process-relevant plasma parameters by Fiber PROES in capacitively coupled Ar/CF$_4$ plasmas

Monitoring changes of process-relevant plasma parameters, such as the electron density and ion flux to the wafer, is essential for the development and control of plasma processes. However, invasive plasma diagnostics, such as probe measurements, typically cannot be applied to commercial reactors. At the example of a low pressure capacitive radio frequency discharge operated in different mixtures of CF$_4$ and Ar, we demonstrate that changes of such plasma parameters can be monitored non-invasively by phase resolved optical emission spectroscopy via optical fibers (Fiber PROES), for which the ports are usually available at industrial plasma sources. In this way, the spatio-temporally resolved dynamics of energetic electrons are tracked by observing a selected emission line. By measuring the electron density and ion flux directly via probe and retarding field energy analyzer diagnostics as a function of driving voltage and pressure, changes of these plasma parameters, including hysteresis effects, are found to be correlated with transitions of the electron power absorption mode revealed by Fiber PROES. Such mode transitions cause the electron energy distribution function (EEDF) to change and, thus, affect such plasma parameters. Based on these findings, Fiber PROES can be used as a non-invasive diagnostic for the monitoring and knowledge-based development of plasma processes.

physics.plasm-ph↗
arXiv2026

Spatio-Temporal Scheduling Prediction Under Backhaul Delay for Resilient Coordinated Beamforming

Coordinated beamforming in distributed 5G networks relies on the timely exchange of inter-cell scheduling information, but backhaul latency makes this information stale. Even a single transmission time interval (TTI) of delay can reduce CBF-SLNR performance below the uncoordinated baseline, because the precoder suppresses interference toward users that are no longer active. Coordination on stale information is therefore worse than no coordination at all. To address this, we propose a two-stage predictive framework in which a Spectral Temporal Graph Neural Network (StemGNN) predicts future user equipment (UE) scheduling states from delayed historical observations, and the predictions replace stale inputs to the CBF-SLNR precoder. Evaluated on a three-cell massive MIMO downlink with 60 UEs and 64 antennas per base station under Quadriga Urban Micro (UMi) channels and a proportional fair scheduler, StemGNN achieves a mean scheduling prediction accuracy of 87.57%, outperforming LSTM, GRU, Simple RNN, and Markov chain baselines at all evaluated horizons, with gains of up to 7.71% over LSTM at longer horizons where inter-UE structural dependencies dominate over temporal autocorrelation. When integrated into coordinated beamforming, the predictions recover 57-73% of the sum rate loss caused by one TTI of backhaul delay, improving sum rate by 9.58-14.35% over the no-prediction baseline and recovering up to 83% of the Lag-1 fairness loss for cell-edge users, with fairness gains persisting at higher lag values where throughput gains diminish. These results show that treating backhaul latency as a spatio-temporal forecasting problem is an effective approach for robust inter-cell coordination in delay-constrained networks.

cs.NI↗
arXiv2026

Star Clusters in the Ultraviolet

Ultraviolet (UV) observations provide a powerful window into the hot and evolved stellar populations that shape the structure, evolution and integrated light of star clusters. Because UV wavelengths are highly sensitive to massive main-sequence stars, blue straggler stars (BSS), extreme-horizontal branch (HB) stars, post-HB stars, interacting binaries and compact remnants, they probe key evolutionary processes that are inaccessible at optical and infrared wavelengths. This review synthesises five decades of UV studies of star clusters across the Milky Way, the Magellanic Clouds (MCs) and nearby galaxies, drawing on results from early space missions, wide-field surveys, and high-resolution imaging. In Galactic open clusters, UV studies have revealed compact companions$-$including white dwarfs, hot subdwarfs and stripped stars$-$and have established the mass-transfer origins of BSS, Blue Lurkers and yellow stragglers. In globular clusters, UV imaging has identified multiple stellar populations through UV-sensitive molecular bands, probed helium enrichment and mapped HB morphologies. Wide-field UVIT surveys have extended HST studies with homogeneous catalogues of HB and post-HB stars across entire clusters. In the MCs, UV observations have transformed our understanding of multiple populations, rotation-driven extended main-sequence turn-offs and the UV-dim phenomenon, while spectroscopic surveys have constrained massive-star evolution, stellar winds and binarity at low metallicity. UV mapping of the Magellanic Bridge has revealed ongoing massive-star formation in low-density tidal environments. Beyond the Local Group, UV studies of extragalactic clusters constrain star-formation histories, stellar feedback and population synthesis across galactic environments. Collectively, UV observations now form a cornerstone of star cluster astrophysics and will continue to do so with upcoming missions.

astro-ph.GA↗
arXiv2026

SpatialFly: Implicit 3D Prior-Guided Visual Reparameterization for Continuous UAV Vision-and-Language Navigation

UAVs play an important role in applications such as autonomous exploration, disaster response, and infrastructure inspection. However, UAV VLN in complex 3D environments remains challenging. A key difficulty is the structural representation mismatch between 2D visual perception and the 3D trajectory decision space, which limits spatial reasoning. To this end, we propose SpatialFly, a geometry-guided spatial representation framework for UAV VLN. Operating on RGB observations without explicit 3D reconstruction, SpatialFly introduces a geometry-guided 2D adaptive representation mechanism. Specifically, the geometric prior injection module injects global structural cues into 2D semantic tokens to provide scene-level geometric guidance. The geometry-aware reparameterization module then uses geometry-conditioned cross-modal attention and gated residual fusion to adaptively reparameterize the visual tokens. Experimental results show that SpatialFly consistently outperforms state-of-the-art UAV VLN baselines across both seen and unseen environments, reducing NE by 4.03m and improving SR by 1.27% over the strongest baseline on the unseen Full split. Additional trajectory-level analysis shows that SpatialFly produces trajectories with better path alignment and smoother, more stable motion.

cs.CV↗
arXiv2026

DynFly: Dynamic-Aware Continuous Trajectory Generation for UAV Vision-Language Navigation in Urban Environments

Recent advances in multimodal large models have significantly improved UAV vision-language navigation (UAV-VLN) by enhancing high-level perception and reasoning. However, existing methods mainly focus on predicting discrete actions, local targets, or sparse waypoints, while the continuous transition from navigation intent to executable UAV motion remains weakly modeled. This motion-interface gap limits the continuity, stability, and executability of generated UAV trajectories. To address this gap, we propose DynFly, a dynamic-aware continuous trajectory generation framework that bridges high-level navigation reasoning and executable UAV motion. DynFly bridges high-level navigation intent and continuous UAV motion through a lightweight trajectory generation layer. Specifically, it represents expert trajectories in B-spline control-point space and employs a Spline-DiT generator to learn conditional trajectory generation via flow matching. Furthermore, we introduce UAV-oriented dynamic-aware supervision over position, finite-difference velocity, finite-difference acceleration, heading consistency, and local target alignment, enabling the generated trajectories to better satisfy UAV motion characteristics. And our trajectory generation framework can also be integrated with an existing UAV-VLN framework while preserving its original visual-language reasoning pipeline. Extensive experiments on the OpenUAV UAV-VLN benchmark show that DynFly improves both navigation performance and trajectory quality. On the Test Unseen Full split, DynFly improves the strongest baseline by 4.69 NDTW, 2.40 SDTW, 2.14 SR points and 4.87 OSR points, while reducing NE by 4.51 m.

cs.RO↗
arXiv2026

JD Oxygen AI Item Center (Oxygen AIIC) V1: An Industrial-Scale LLM/VLM-Centric Solution for Item Understanding, Management, and Applications

JD$.$com, one of the world's largest e-commerce platforms, serves over 700 million active users and millions of merchants, with a catalog of tens of billions of SKUs. At this scale, high-quality, structured item knowledge underpins a better consumer experience, lower management costs, and higher operational efficiency-yet producing and serving it poses three industrial-scale challenges: fast-emerging concepts, high-quality knowledge production for massive SKUs, and diverse downstream requirements. To address these challenges, we present the JD Oxygen AI Item Center (Oxygen AIIC), an industrial-scale platform built on LLMs/VLMs for item-knowledge production and service. Oxygen AIIC is built around four core pillars: (i) ontology engineering driven by efficient human-AI collaboration, which supports the dynamic evolution and agile expansion of an ontology with millions of entries; (ii) a "Semantic Search then Discrimination"(S2D) knowledge identification architecture that, combined with throughput improvement strategies, enables scalable, extensible, and high-throughput AI Item Library production for tens of billions of SKUs; (iii) self-evolving item-understanding LLMs/VLMs that improve in a stable and controllable manner, enabling knowledge production with 94.2% precision and 82.8% recall; and (iv) a unified item tunnel that serves as the data and service hub. Oxygen AIIC now covers tens of thousands of JD categories and processes hundreds of millions of item updates per day on Huawei Ascend NPUs. It has accumulated hundreds of billions of item-knowledge assets. Deployed across core business scenarios-including search, recommendation, operations, category planning-Oxygen AIIC has delivered measurable gains at scale. Search-traffic coverage reaches 80.4%, item-information quality issues drop by 37%, the automated fill rate of core attributes during item listing exceeds 80%.

cs.AI↗
arXiv2026

Limit behavior of linearly edge-reinforced random walks on the half-line

Motivated by the article [M. Takei, Electron. J. Probab. 26 (2021), article no. 104], we study the limit behavior of linearly edge-reinforced random walks on the half-line $\mathbb{Z}_+$ with reinforcement parameter $δ>0$, and each edge $\{x,x+1\}$ has the initial weight $x^α\ln^βx$ for $x > 1$ and $1$ for $x = 0, 1$. The aim of this paper is to study the almost sure limit behavior of the walk in the recurrent regime, and extend the results of Takei mentioned above.

math.PR↗
arXiv2026

HybridCodeAuthorship: A Benchmark Dataset for Line-Level Code Authorship Detection

Thanks to the rapid adoption of AI code assistants powered by large language models (LLMs), industry codebases are, increasingly, a hybrid of AI- and human-authored code. For risk management and productivity analysis purposes, it is crucial to enable fine-grained location detection of AI-generated code. To develop algorithms for this task, quality benchmarks are needed to assess performance. However, existing benchmarks tend to comprise academic, LeetCode-style problems and presume a code snippet is either completely human-authored or completely AI-authored, which is not reflective of the diverse intents and styles of industry codebases utilizing AI code assistants. To fill these gaps, we introduce HybridCodeAuthorship, a novel benchmark of Python code files with interleaved human- and AI-authored lines of code to simulate authentic utilization of AI code assistants. In this paper, we first present our dataset construction pipeline, which leverages CodeSearchNet, a massive collection of links to open sourced repositories on GitHub. We then benchmark the performance of two state-of-the-art AI-generated code detection algorithms at both the line- and chunk-level. Experimental results demonstrate that HybridCodeAuthorship is a challenging benchmark with a top-scoring algorithm, AIGCode Detector, obtaining a highest F1 score of 0.48 and 0.56 on chunk-level and line-level code detection tasks, respectively.

cs.SE↗
arXiv2026

ProjQ: Project-and-Quantize for Adapter-Aware LLM Compression

Post-Training Quantization (PTQ) and Low-Rank Adaptation (LoRA) constitute the standard pipeline for efficient Large Language Model (LLM) deployment. However, applying them sequentially poses a problem: PTQ often leaves behind random noise that is spread out (across the model's weights) in a way LoRA can't easily fix, meaning that LoRA ends up wasting its limited capacity trying to fix uncorrectable noise instead of improving task performance. In this paper, we propose \textbf{ProjQ}, a novel framework for constraining quantization noise to the low-rank manifold via orthogonal subspace projection. We derive an efficient alternating algorithm that shapes the quantization noise into a low-rank structure, effectively offloading dominant error components to the subsequent adapter while minimizing the residual error in the orthogonal "uncorrectable" subspace. Our theoretical analysis demonstrates that ProjQ preserves strictly greater model plasticity for downstream tasks compared to standard PTQ. Extensive experiments on LLaMA-2, Qwen2.5 and Qwen3 confirm that ProjQ consistently outperforms existing methods in both quantization error compensation and downstream task fine-tuning, achieving up to $2\times$ lower evaluation loss for compensation and matching the performance of standard 4-bit baselines on language modeling tasks with only 3 bits. The code is available on https://github.com/yy9301/ProjQ .

cs.LG↗
arXiv2026

ZipRL: Adaptive Multi-Turn Context Compression with Hindsight Response Replay

Adaptive context compression is vital for scaling Large Language Models (LLMs) to complex, multi-turn agent tasks. However, rule-based compression methods may discard task-critical nuances, while Reinforcement Learning (RL) approaches usually struggle to balance information retention and token efficiency under the sparse rewards inherent to long-horizon workflows. To bridge this gap, we propose ZipRL, a novel adaptive compression framework tailored for Reinforcement Learning from Verifiable Rewards (RLVR). ZipRL features a multi-granularity compression mechanism for active, non-uniform information reduction, coupled with Hindsight Response Replay (HRR), a technique designed to densify training signals during RLVR optimization. Theoretically, we prove ZipRL's superior task-relevant utility over uniform methods. Concretely, ZipRL utilizes coarse-to-fine prompts for macro-compression and incorporates HRR into GRPO via generalized advantage reshaping. Multiple models of varying versions and parameter scales validate the effectiveness of our approach. Benchmarks on five agent tasks show ZipRL outperforms state-of-the-art approaches by 27.9% and 34.7% across Qwen3-4B and Qwen3-8B models, while maintaining exceptional token efficiency and robustness under extreme 256-turn extrapolation stress tests.

cs.AI↗
arXiv2026

Learning to Adapt SFT Data for Better Reasoning Generalization

Large language models (LLMs) have achieved remarkable progress, with post-training playing a crucial role in enhancing their reasoning capabilities. Among post-training paradigms, supervised fine-tuning (SFT) is widely used: it leverages external data to provide dense supervision and enables efficient training. However, directly fine-tuning on expert data can hurt generalization when the data distribution is mismatched with the target model's own distribution. In this work, we propose Data Adaptation for Reasoning Tuning (DART), which formulates the use of a fixed, potentially distributionally misaligned SFT dataset as an optimization problem over demonstration transformations. DART trains a mapper model with reinforcement learning to convert original SFT data into model-adapted supervision that better matches the target model's distribution and learning preferences. The transformed data are then used for SFT, allowing the target model to better exploit external supervision. Experiments across multiple models and datasets show that DART improves generalization, achieves higher training efficiency than direct RL, and helps models surpass standard SFT. Our code is available at https://anonymous.4open.science/r/DART525E50D.

cs.CL↗
arXiv2026

Fill the GAP: A Granular Alignment Paradigm for Visual Reasoning in Multimodal Large Language Models

Visual latent reasoning lets a multimodal large language model (MLLM) create intermediate visual evidence as continuous tokens, avoiding external tools or image generators. However, existing methods usually follow an output-as-input latent paradigm and yield unstable gains. We identify evidence for a feature-space mismatch that can contribute to this instability: dominant visual-latent models build on pre-norm MLLMs and reuse decoder hidden states as predicted latent inputs, even though these states occupy a substantially different norm regime from the input embeddings the model was trained to consume (Xie et al., 2025; Li et al., 2026; Team et al., 2026). This mismatch can make direct latent feedback unreliable. Motivated by this diagnosis, we propose GAP, a Granular Alignment Paradigm for visual latent modeling. GAP aligns visual latent reasoning at three levels: feature-level alignment maps decoder outputs into input-compatible visual latents through a lightweight PCA-aligned latent head; context-level alignment grounds latent targets with inspectable auxiliary visual supervision; and capacity-guided alignment assigns latent supervision selectively to examples where the base MLLM struggles. On Qwen2.5-VL 7B, the resulting model achieves the best mean aggregate perception and reasoning performance among our supervised variants. Inference-time intervention probing further suggests that generated latents provide task-relevant visual signal beyond merely adding token slots.

cs.CV↗
arXiv2026

When Self-Belief Misleads: Active Label Acquisition for Reinforcement Learning with Verifiable Rewards

Large Language Models (LLMs) have achieved remarkable advancements in reasoning capabilities empowered by Reinforcement Learning with Verifiable Rewards (RLVR). Nonetheless, RLVR intrinsically relies on ground-truth labels for reward computation, the acquisition of which is often prohibitively expensive in real-world scenarios. While unsupervised RLVR paradigms attempt to circumvent this by training on pseudo-labels, they are notoriously susceptible to training collapse. Moreover, different samples often exhibit varying annotation values. In this paper, we propose Reinforcement Learning with Active Verifiable Rewards (RLAVR), which actively acquires ground-truth labels for a small set of selected samples and integrates them with pseudo-labels, thereby stabilizing training dynamics and improving performance under limited annotation budgets. To identify valuable samples, we propose the Corrective Advantage Gap (CAG) metric and analyze the sample-level supervision value. Building on this, we introduce Correction-Aware Reliability Estimation for RLAVR (CARE), which translates the oracle CAG criterion into a practical pre-query acquisition policy to substantially improve training stability. Extensive experiments across diverse domains, model families, and model scales demonstrate the effectiveness and generality of our approach. Our code is available at https://github.com/Lumina04/CARE.

cs.LG↗
arXiv2026

Single-Photon Fourier Transform

The extraction of information carried by light plays an increasingly important role in optical communication, imaging, and detection. However, the information can only be successfully extracted when the light pulse is comparably strong, leaving untouched scenarios where survived photons are extremely sparse. Here, we propose and experimentally demonstrate a single-photon Fourier transform scheme. By retrieving the implicit correlation shared in the sparse singlephoton stream globally, we are able to precisely classify each photon and synchronously extract multiple ultra-weak signals with high fidelity against extreme environments. Our experiment results give a full picture of the scheme in terms of multi-terminal expandability, wide frequency adaptability, 125 dB loss tolerance, and -10.4 dB signal-to-noise ratio robustness. Even when the pulse repetition frequencies of all terminals are the same, we can still recognize the free-running clock drift and separate different messages. Our work can be a general scheme to extend the capability boundary for all the extremely low-light-flux scenarios, and makes many challenging tasks possible, such as in-orbit optical communication network with complex topology, navigation in extremely lossy and noisy environments, and wide-range single-photon imaging with multi-source illumination.

physics.optics↗