arXiv subjects

Xinyang Zhang

Publications and source records attributed to Xinyang Zhang.

At least 19 recordsLinked to original sources

arXiv2026

PIML-OFEM: A New Large-Scale Structural Analysis Method Based on Problem-Independent Machine Learning and Overlapping Finite Element Technique

High-resolution analysis and design of large-scale heterogeneous structures require accurate reduced-order models and efficient online computation. Existing multiscale methods must repeatedly construct local basis functions for different material distributions, whereas substructure-based problem-independent machine learning (PIML) methods can be limited by prescribed boundary displacement interpolation. We propose PIML-OFEM, an overlapping finite element method accelerated by problem-independent machine learning. Each substructure retains only its corner-node degrees of freedom. Oversampled numerical basis functions are constructed by solving local elasticity problems on extended domains and restricting the solutions to the target substructure, eliminating prescribed displacement interpolation on its boundary. Independently constructed local bases are blended through a partition-of-unity overlapping formulation to obtain a globally continuous displacement field. A U-Net learns the mapping from local Young's modulus distributions to numerical basis functions, replacing repeated online local solves and allowing the model to be reused across load cases and global boundary conditions. Numerical examples show close agreement with fine-scale finite element results in displacement and elemental strain energy. PIML-OFEM reduces online computational cost relative to direct finite element analysis and improves accuracy over PIML substructure models based on linear boundary interpolation. In topology optimization, the method supports stable high-resolution iterations with small filter radii and preserves fine-scale features, including local patterns resembling rank-2 microstructures. The framework provides an efficient physics-data approach for large-scale heterogeneous structural analysis and high-resolution topology optimization.

math.NA↗

arXiv2026

Thermal Hall Resistivity as a Unifying Description of Phonon Thermal Hall Effect in Various Insulators

A considerable phonon thermal Hall effect was recently discovered across a diverse collection of materials. To clarify this enigmatic thermal Hall response in various insulators, we investigate the doped Mott insulator La$_{5/3}$Sr$_{1/3}$NiO$_4$ as an example material system and reveal a characteristic phonon-dominated thermal Hall effect. As a sensitive probe of the transverse thermal response, the thermal Hall resistivity $w_{xy}$ exhibits an insulating-like temperature dependence $w_{xy}(T)$, a linear magnetic-field dependence $w_{xy}(H)$ near $H=0$, and a $T$-linear thermal Hall angle at low temperatures. The presence of similar phenomena across a series of insulators suggests that $w_{xy}$ serves as a unifying description of phonon thermal Hall effect, corroborated by an apparent correlation between the insulating-like $w_{xy}(T)$ and material's localized electronic state.

cond-mat.str-el↗

arXiv2026

CoMem: Context Management with A Decoupled Long-Context Model

Context management enables agentic models to solve long-horizon tasks through iterative summarization of previous interaction histories. However, this process typically incurs substantial decoding overhead for the extra summarization tokens, which significantly affect the end-to-end response latency at deployment. In this paper, we introduce CoMem, a novel framework that decouples memory management from the primary agent workflow, enabling these processes to execute in parallel. We propose a $k$-step-off asynchronous pipeline that overlaps the memory model's summarization with the agent's inference, effectively masking the latency of context processing. To ensure robustness under this asynchronous setting, we introduce a reward-driven training strategy that aligns the memory model to capture sufficient statistics for the agent's decision-making. Theoretical analysis confirms that CoMem offers a superior efficiency-effectiveness trade-off compared to coupled architectures. Our extensive experimental results on SWE-Bench-Verified show that CoMem provides 1.4x latency improvements upon vanilla long-context solutions while preserving most of the performance. Furthermore, we demonstrate that these latency gains scale favorably with increased system throughput, offering a modular path forward for the independent optimization of agent reasoning and memory compression.

cs.LG↗

arXiv2026

END: Early Noise Dropping for Efficient and Effective Context Denoising

Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing tasks. However, they are often distracted by irrelevant or noisy context in input sequences that degrades output quality. This problem affects both long- and short-context scenarios, such as retrieval-augmented generation, table question-answering, and in-context learning. We reveal that LLMs can implicitly identify whether input sequences contain useful information at early layers, prior to token generation. Leveraging this insight, we introduce Early Noise Dropping (\textsc{END}), a novel approach to mitigate this issue without requiring fine-tuning the LLMs. \textsc{END} segments input sequences into chunks and employs a linear prober on the early layers of LLMs to differentiate between informative and noisy chunks. By discarding noisy chunks early in the process, \textsc{END} preserves critical information, reduces distraction, and lowers computational overhead. Extensive experiments demonstrate that \textsc{END} significantly improves both performance and efficiency across different LLMs on multiple evaluation datasets. Furthermore, by investigating LLMs' implicit understanding to the input with the prober, this work also deepens understanding of how LLMs do reasoning with contexts internally.

cs.CL↗

arXiv2026

T$^2$PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning

Recent progress in multi-turn reinforcement learning (RL) has significantly improved reasoning LLMs' performances on complex interactive tasks. Despite advances in stabilization techniques such as fine-grained credit assignment and trajectory filtering, instability remains pervasive and often leads to training collapse. We argue that this instability stems from inefficient exploration in multi-turn settings, where policies continue to generate low-information actions that neither reduce uncertainty nor advance task progress. To address this issue, we propose Token- and Turn-level Policy Optimization (T$^2$PO), an uncertainty-aware framework that explicitly controls exploration at fine-grained levels. At the token level, T$^2$PO monitors uncertainty dynamics and triggers a thinking intervention once the marginal uncertainty change falls below a threshold. At the turn level, T$^2$PO identifies interactions with negligible exploration progress and dynamically resamples such turns to avoid wasted rollouts. We evaluate T$^2$PO in diverse environments, including WebShop, ALFWorld, and Search QA, demonstrating substantial gains in training stability and performance improvements with better exploration efficiency. Code is available at: https://github.com/WillDreamer/T2PO.

cs.AI↗

arXiv2026

Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards

Large Language Models demonstrate strong capabilities in single-turn instruction following but suffer from Lost-in-Conversation (LiC), a degradation in performance as information is revealed progressively in multi-turn settings. Motivated by the current progress on Reinforcement Learning with Verifiable Rewards (RLVR), we propose Curriculum Reinforcement Learning with Verifiable Accuracy and Abstention Rewards (RLAAR), a framework that encourages models not only to generate correct answers, but also to judge the solvability of questions in the multi-turn conversation setting. Our approach employs a competence-gated curriculum that incrementally increases dialogue difficulty (in terms of instruction shards), stabilizing training while promoting reliability. Using multi-turn, on-policy rollouts and a mixed-reward system, RLAAR teaches models to balance problem-solving with informed abstention, reducing premature answering behaviors that cause LiC. Evaluated on LiC benchmarks, RLAAR significantly mitigates LiC performance decay (62.6% to 75.1%) and improves calibrated abstention rates (33.5% to 73.4%). Together, these results provide a practical recipe for building multi-turn reliable and trustworthy LLMs.

cs.CL↗

arXiv2026

Autoregressive Image Generation with Masked Bit Modeling

This paper challenges the dominance of continuous pipelines in visual generation. We systematically investigate the performance gap between discrete and continuous methods. Contrary to the belief that discrete tokenizers are intrinsically inferior, we demonstrate that the disparity arises primarily from the total number of bits allocated in the latent space (i.e., the compression ratio). We show that scaling up the codebook size effectively bridges this gap, allowing discrete tokenizers to match or surpass their continuous counterparts. However, existing discrete generation methods struggle to capitalize on this insight, suffering from performance degradation or prohibitive training costs with scaled codebook. To address this, we propose masked Bit AutoRegressive modeling (BAR), a scalable framework that supports arbitrary codebook sizes. By equipping an autoregressive transformer with a masked bit modeling head, BAR predicts discrete tokens through progressively generating their constituent bits. BAR achieves a new state-of-the-art gFID of 0.99 on ImageNet-256, outperforming leading methods across both continuous and discrete paradigms, while significantly reducing sampling costs and converging faster than prior continuous approaches. Project page is available at https://bar-gen.github.io/

cs.CV↗

arXiv2025

Heteroscedasticity of Denoising Score Matching with Generalised Smooth Noise

Score Matching (SM) is a powerful framework for estimating the log-density derivatives of a distribution without calculating its normalizing constants. This capability has made it a cornerstone across multiple domains, from classical sta- tistical estimation and energy-based models to modern diffusion-based generative models. In practice, these models rely almost exclusively on Denoising Score Matching (DSM) as a tractable proxy for score matching. This ubiquity naturally raises a fundamental question: Is DSM truly "score matching for free"? In this work, we demonstrate that DSM is not a perfect substitute. We prove that the denoising objective is inherently heteroscedastic, the variance of model parame- ters fluctuates unpredictably based on both noise levels and the underlying data geometry. This instability is baked into the mathematical structure of the DSM. To address this, we derive an ideal weighting function that equalizes this variance, yielding a homoscedastic generalization of DSM. Since the ideal weights are of- ten empirically inaccessible, we show that a practical approximation weighting function via Taylor expansion reduces gradient variance during training, at the cost of statistical optimality. Notably, this provides a theoretical justification for an existing heuristic weight used in Isotropic Gaussian Diffusion. We validate our theory across different perturbed distributions and for higher-order scores.

cs.LG↗

arXiv2025

From Web Search towards Agentic Deep Research: Incentivizing Search with Reasoning Agents

Information retrieval is a cornerstone of modern knowledge acquisition, enabling billions of queries each day across diverse domains. However, traditional keyword-based search engines are increasingly inadequate for handling complex, multi-step information needs. Our position is that Large Language Models (LLMs), endowed with reasoning and agentic capabilities, are ushering in a new paradigm termed Agentic Deep Research. These systems transcend conventional information search techniques by tightly integrating autonomous reasoning, iterative retrieval, and information synthesis into a dynamic feedback loop. We trace the evolution from static web search to interactive, agent-based systems that plan, explore, and learn. We also introduce a test-time scaling law to formalize the impact of computational depth on reasoning and search. Supported by benchmark results and the rise of open-source implementations, we demonstrate that Agentic Deep Research not only significantly outperforms existing approaches, but is also poised to become the dominant paradigm for future information seeking. All the related resources, including industry products, research papers, benchmark datasets, and open-source implementations, are collected for the community in https://github.com/DavidZWZ/Awesome-Deep-Research.

cs.IR↗

arXiv2025

PersonaAgent: Bridging Memory and Action for Personalized LLM Agents

Large Language Model (LLM) empowered agents have recently emerged as advanced paradigms that exhibit impressive capabilities in a wide range of domains and tasks. Despite their potential, current LLM agents often adopt a one-size-fits-all approach, lacking the flexibility to respond to users' varying needs and preferences. This limitation motivates us to develop PersonaAgent, the first personalized LLM agent framework designed to address versatile personalization tasks. Specifically, PersonaAgent integrates two complementary components - a personalized memory module that includes episodic and semantic memory mechanisms; a personalized action module that enables the agent to perform tool actions tailored to the user. At the core, the persona (defined as unique system prompt for each user) functions as an intermediary: it leverages insights from personalized memory to control agent actions, while the outcomes of these actions in turn refine the memory. Based on the framework, we propose a test-time user-preference alignment strategy that simulate the latest n interactions to optimize the persona prompt, ensuring real-time user preference alignment through textual loss feedback between simulated and ground-truth responses. Experimental evaluations demonstrate that PersonaAgent significantly outperforms other baseline methods by not only personalizing the action space effectively but also scaling during test-time real-world applications. These results underscore the feasibility and potential of our approach in delivering tailored, dynamic user experiences.

cs.AI↗

arXiv2025

Possible Unconventional Surface Superconductivity in the Half-Heusler YPtBi

We report an extensive study of the noncentrosymmetric half-Heusler superconductor YPtBi, revealing an unusual relation between bulk superconductivity and the possible appearance of surface superconductivity on the (111) oriented surface, at temperatures up to 3 times the bulk transition temperature. Transport measurements confirmed the low carrier density of the material and its bulk superconducting transition, which was also observed in ac susceptibility through mutual inductance (MI) measurements. However, a weak signature of superconductivity in the MI measurements appeared much above the bulk transition temperature, which was further observed in scanning tunneling spectroscopy, pointing to a possible surface superconducting state. Polar Kerr effect measurements suggest that while the bulk superconductor may exhibit an unusual nodal superconducting state, only the surface state breaks time reversal symmetry. Complementary tunneling measurements on LuPtBi are used to establish the observations on YPtBi, while density-functional theory calculations may shed light on the origin of this unusual surface state.

cond-mat.supr-con↗

arXiv2025

A counterexample of the Fredholm of Toeplitz operator

In this paper we study the essential spectra of the Toeplitz operator on the Hardy space $H^1$. We give a counterexample to show that the Toeplitz operator with symbol is not Fredholm, which gives a counterexample to the conjecture by J.A. Virtanen J A in 2006.

math.FA↗

arXiv2025

The Spectrum of C/2023 A3 Indicates A Depleted Composition

We report a spectroscopic observation of comet C/2023 A3 using an 80 mm apochromatic (apo) refractor equipped with a custom-built spectrometer with a resolution of R~2,500 on the night of 4 October 2024. Sodium D lines were detected prominently, while no other emission lines, particularly carbon-bearing species, were observed, which suggests that comet C/2023 A3 may be carbon-depleted. The mobility and flexibility of our observational setup highlight the value of amateur telescopes in observing low-altitude targets like C/2023 A3 as a complement to professional facilities.

astro-ph.EP↗

arXiv2024

Prompt-Guided Mask Proposal for Two-Stage Open-Vocabulary Segmentation

We tackle the challenge of open-vocabulary segmentation, where we need to identify objects from a wide range of categories in different environments, using text prompts as our input. To overcome this challenge, existing methods often use multi-modal models like CLIP, which combine image and text features in a shared embedding space to bridge the gap between limited and extensive vocabulary recognition, resulting in a two-stage approach: In the first stage, a mask generator takes an input image to generate mask proposals, and the in the second stage the target mask is picked based on the query. However, the expected target mask may not exist in the generated mask proposals, which leads to an unexpected output mask. In our work, we propose a novel approach named Prompt-guided Mask Proposal (PMP) where the mask generator takes the input text prompts and generates masks guided by these prompts. Compared with mask proposals generated without input prompts, masks generated by PMP are better aligned with the input prompts. To realize PMP, we designed a cross-attention mechanism between text tokens and query tokens which is capable of generating prompt-guided mask proposals after each decoding. We combined our PMP with several existing works employing a query-based segmentation backbone and the experiments on five benchmark datasets demonstrate the effectiveness of this approach, showcasing significant improvements over the current two-stage models (1% ~ 3% absolute performance gain in terms of mIOU). The steady improvement in performance across these benchmarks indicates the effective generalization of our proposed lightweight prompt-aware method.

cs.CV↗

arXiv2024

Gapless superconductivity in the low-frequency electrodynamic response of two-dimensional granular In/InO$_x$ composites

We measured the full complex ac conductance of two-dimensional granular In/InO$_x$ composites using the mutual inductance technique to explore the transition from a "failed-superconductor-turned anomalous metal" to a robust superconductor. In this system, room-temperature annealing was adopted to tune the InO$_x$-mediated coupling between In grains, allowing for the observation of both a "true" superconductor-to-insulator transition and the emergence of an intervening anomalous metallic state. In this paper, we show that further annealing increases the inter-grain coupling, which eliminates the anomalous metallic phase, but at the same time prevent the emergence of strong Bose-dominated insulating phase. The complex ac conductance revealed a $T\to0$ saturating dissipative response in a finite magnetic field, coexisting with a robust superfluid density. The anomalous power-law spectra for the dissipative response appear to indicate quantum critical behavior proximate to a quantum superconductor to anomalous-metal transition as probed in the kilo-Hertz range, and point to signatures of gapless superconductivity in our granular superconducting system.

cond-mat.supr-con↗

arXiv2024

SoK: On the Semantic AI Security in Autonomous Driving

Autonomous Driving (AD) systems rely on AI components to make safety and correct driving decisions. Unfortunately, today's AI algorithms are known to be generally vulnerable to adversarial attacks. However, for such AI component-level vulnerabilities to be semantically impactful at the system level, it needs to address non-trivial semantic gaps both (1) from the system-level attack input spaces to those at AI component level, and (2) from AI component-level attack impacts to those at the system level. In this paper, we define such research space as semantic AI security as opposed to generic AI security. Over the past 5 years, increasingly more research works are performed to tackle such semantic AI security challenges in AD context, which has started to show an exponential growth trend. In this paper, we perform the first systematization of knowledge of such growing semantic AD AI security research space. In total, we collect and analyze 53 such papers, and systematically taxonomize them based on research aspects critical for the security field. We summarize 6 most substantial scientific gaps observed based on quantitative comparisons both vertically among existing AD AI security works and horizontally with security works from closely-related domains. With these, we are able to provide insights and potential future directions not only at the design level, but also at the research goal, methodology, and community levels. To address the most critical scientific methodology-level gap, we take the initiative to develop an open-source, uniform, and extensible system-driven evaluation platform, named PASS, for the semantic AD AI security research community. We also use our implemented platform prototype to showcase the capabilities and benefits of such a platform using representative semantic AD AI attacks.

cs.CR↗

arXiv2023

Rotating dark matter admixed neutron stars

We study rotating compact stars that are mixtures of the ordinary nuclear matter in a neutron star and fermionic dark matter. After deriving equations describing a slowly rotating system made up of an arbitrary number of perfect fluids, we specialize to the two-fluid case, where the first fluid describes ordinary matter and the second fluid describes dark matter. Electromagnetic observations of the moment of inertia and angular momentum directly probe ordinary matter and not dark matter. Upon taking this into account, we show that the I-Love-Q relations for dark matter admixed neutrons stars can deviate significantly from the standard single-fluid relationships.

gr-qc↗

arXiv2023

TwHIN-BERT: A Socially-Enriched Pre-trained Language Model for Multilingual Tweet Representations at Twitter

Pre-trained language models (PLMs) are fundamental for natural language processing applications. Most existing PLMs are not tailored to the noisy user-generated text on social media, and the pre-training does not factor in the valuable social engagement logs available in a social network. We present TwHIN-BERT, a multilingual language model productionized at Twitter, trained on in-domain data from the popular social network. TwHIN-BERT differs from prior pre-trained language models as it is trained with not only text-based self-supervision, but also with a social objective based on the rich social engagements within a Twitter heterogeneous information network (TwHIN). Our model is trained on 7 billion tweets covering over 100 distinct languages, providing a valuable representation to model short, noisy, user-generated text. We evaluate our model on various multilingual social recommendation and semantic understanding tasks and demonstrate significant metric improvement over established pre-trained language models. We open-source TwHIN-BERT and our curated hashtag prediction and social engagement benchmark datasets to the research community.

cs.CL↗