Search arXiv⌕ Search

arXiv subjects

Yang Zhang

Publications and source records attributed to Yang Zhang.

At least 127 records · Page 7Linked to original sources
arXiv2026

Transition of Magnetic Reconnection Regimes in Partially Ionized Plasmas

Magnetic reconnection in partially ionized plasmas plays a crucial role in a wide range of solar, astrophysical, and laboratory environments. While reconnection in such plasmas is commonly characterized by the ion-neutral coupling strength and the ionization fraction $χ=n_{i}/(n_{i}+n_{n})$, most previous studies have focused primarily on the former. A systematic exploration of the ionization fraction, particularly in combination with ion-neutral coupling, is still lacking. This study presents the first systematic scan of the two-dimensional parameter space defined by ion-neutral collisionality and ionization fraction, enabling investigation of the transition from strongly coupled reconnection to faster, decoupled reconnection. To achieve this, we employ a new three-fluid, five-moment numerical model that treats electrons, ions, and neutrals as separate species on an equal footing. We find that in the strongly coupled regime, the reconnection rate is consistent with a $χ^{1/4}$ scaling. As collisionality decreases, the system transitions to a fast, ionization-independent regime. On the other hand, in the weakly coupled and fast-reconnection regimes, the current sheet approaches an ion-inertial-scale thickness rather than the expanded hybrid scale $d_{i}χ^{-1/2}$ predicted by fully coupled analytic fluid theories. The identified critical thickness and the resulting onset of fast reconnection agree reasonably well with recent fully kinetic simulations and laboratory experiments. In addition, we show that, over a wide range of coupling strengths, the ion outflow velocities remain Alfvénic, scaling with the appropriate ion or hybrid Alfvén speed, while the hybrid outflow velocity scales as $χ^{1/2}$ when normalized by ion Alfvén speed.

physics.plasm-ph↗
arXiv2026

Metis: Memory Foundation Model

Recent advances in AI agents have increasingly internalized native capabilities into their underlying foundation models, giving rise to multimodal foundation models and large reasoning models. However, agent memory is still primarily implemented through external modules, leaving the native memory capability largely unexplored. In this paper, we take a first step toward this direction by introducing memory foundation models, which empower foundation models with native memory capabilities. We formalize native memory from two perspectives: a persistent and dynamically evolving memory state within the backbone, and native memory procedures that autonomously store and utilize information through model computation. We show that native memory offers advantages in architecture, end-to-end optimization, and efficiency. Based on this formulation, we propose Metis, the first prototype of memory foundation models. Metis introduces a new architecture that equips a foundation model with a native memory state, allowing historical information to be compressed into the model and accessed through memory attention. We construct large-scale memory-specific training data and introduce multiple optimization objectives to acquire these native memory procedures through mid-training. The online memory maintenance of Metis is gradient-free, and the memory update requires only a forward pass. At inference time, all learned model weights remain frozen, while the native memory states are autonomously transformed through standard forward computation. Through extensive experiments, we show that Metis exhibits native memory capabilities and further provide a detailed analysis of its strengths, limitations, and behaviors. To facilitate future research on memory foundation models, we release our project and model checkpoints.

cs.CL↗
arXiv2026

Observation of quantum nonclassicality without freedom of choice in a minimal causal network

Quantum causal networks enable tests of nonclassicality beyond Bell nonlocality while relaxing some physically unwarranted assumptions. By relaxing the freedom-of-choice and spacelike-separation assumptions, the unrelated-confounders causal networks provide a simple and robust route to certify quantum nonclassicality. Here, we implement the minimal three-node unrelated-confounders network in an optical experiment using two independent polarization-entangled photon sources and an intervention at the central node, experimentally achieved with a high-fidelity entangling measurement. We employ a causal data-fusion protocol that combines observational and interventional data to significantly improve the protocol's noise tolerance, and observe a violation of the corresponding hybrid causal inequality by more than three standard deviations. Our results provide deeper insights into quantum nonlocality in networks and highlight the UC network as a compact, experimentally accessible platform for device-independent quantum protocols that do not require actively chosen measurement settings.

quant-ph↗
arXiv2026

Every Wrong Answer Counts: Option-Level Psychometrics for LLM Multiple-Choice Benchmarks

Most multiple-choice question (MCQ) benchmarks evaluate Large Language Models (LLMs) only by whether they select the correct answers. This binary scoring treats all incorrect responses alike, even though an LLM's preferences among incorrect options may contain systematic and useful information about its behavior and ability. We introduce the LLM Nominal Response Model (LLM-NRM), an option-aware psychometric framework that models the full distribution over answer choices to jointly estimate LLM ability and option-level item characteristics, while separating model-specific response calibration sharpness, positional preference, and difficulty-dependent fallback behavior. Across 189 LLMs and 31,554 items from 14 benchmarks, LLM-NRM predicts held-out LLM-item interactions more accurately than binary Item Response models and conventional nominal-response baselines, and its ability estimates achieve the strongest Spearman correlation of 0.920 with the external human-preference Arena.ai Elo leaderboard. Distractor identity contributes +101% additional Fisher Information per item beyond correctness, and incorrect responses alone recover full-information ability estimates with Spearman 0.943. The learned item parameters also enable efficient benchmarking, where 41 selected items preserve the full-bank ranking with Kendall's correlation 0.85, corresponding to a 770 times reduction. In conclusion, we show that incorrect answers carry distinct and useful measurement information rather than representing equivalent mistakes.

cs.CL↗
arXiv2026

Question Begets Question: Self-Evolving Curriculum for Reinforcement Fine-Tuning on Competition Mathematics

Teaching a language model a skill it has not mastered is obstructed by three recurring difficulties: training data is scarce, ground-truth reasoning traces are usually unavailable, and models often exhibit an apparent ceiling beyond which additional data yields no further improvement. We study these difficulties in a controlled setting, fine-tuning Qwen2.5-Math-7B on competition mathematics (AIME), a task on which it initially solves only 5.6\% of problems (pass@1). To address data scarcity, we introduce Question-begets-Question (QbQ), a scalable procedure in which a teacher transforms existing problems into diverse variants that probe the same underlying skills; to model the absence of oracle reasoning, we train exclusively via reinforcement learning on problem statements and final answers, never on teacher reasoning traces. Static training on such data, however, plateaus well short of the task: real-plus-synthetic augmentation and non-curriculum QbQ generated synthetic data training cap pass@1 at 12.5\% and 14.5\% respectively, despite large increases in data. Our central finding is that this ceiling is not intrinsic to the model. We propose a self-evolving curriculum that, each round, evaluates the current checkpoint, seeds QbQ from the problems it can mostly get right, and trains on the resulting variants; under an identical data budget, this breaks the ceiling and lifts pass@1 to 16.5\% with no sign of saturation after 20 rounds. Counterintuitively, we find that models improve when trained on variants of problems they can mostly get right, and that models trained this way go on to solve harder problems never seen during training.

cs.LG↗
arXiv2026

Seeing the Unseen: Towards Training-Free Inspection for Wind Turbine Blades Using Knowledge-Augmented Vision Language Models

Wind turbine blades operate in harsh environments, making timely damage detection essential for preventing failures and optimizing maintenance. Drone-based inspection and deep learning are promising, but typically depend on large, labeled datasets, which limit their ability to detect rare or evolving damage types. To address this, we propose a zero-shot-oriented inspection framework that integrates Retrieval-Augmented Generation (RAG) with Vision-Language Models (VLM). A multimodal knowledge base is constructed, comprising technical documentation, representative reference images, and domain-specific guidelines. A hybrid text-image retriever with keyword-aware reranking assembles the most relevant context to condition the VLM at inference, injecting domain knowledge without task-specific training. We evaluate the framework on 30 labeled blade images covering diverse damage categories. Although the dataset is small due to the difficulty of acquiring verified blade imagery, it covers multiple representative defect types. On this test set, the RAG-grounded VLM correctly classified all samples, whereas the same VLM without retrieval performed worse in both accuracy and precision. We further compare against open-vocabulary baselines and incorporate uncertainty Clopper-Pearson confidence intervals to account for the small-sample setting. Ablation studies indicate that the key advantage of the framework lies in explainability and generalizability: retrieved references ground the reasoning process and enable the detection of previously unseen defects by leveraging domain knowledge rather than relying solely on visual cues. This research contributes a data-efficient solution for industrial inspection that reduces dependence on extensive labeled datasets.

cs.CV↗
arXiv2026

Intertwined magnetoresistance and Hall multifunctionality in a non-coplanar magnetic Weyl semimetal DyB4

Anomalous magneto-transport responses provide complementary probes of orbital motion, momentum-space topology, and real-space spin chirality, yet their integration into a single material remains rare because their underlying requirements often compete. A promising materials-design strategy is to realize a magnetic Weyl semimetal that combines linearly dispersive high-mobility bands with tunable non-coplanar magnetism while limiting spin-dependent scattering. Here we identify DyB4, a frustrated rare-earth tetraboride, as a magnetic Weyl semimetal candidate that embodies this strategy and hosts intertwined magnetoresistance and Hall multifunctionality. Neutron diffraction reveals a sequence of field-tunable magnetic states, including non-coplanar spin configurations and PT-symmetry-broken phases. First-principles calculations identify steep linear dispersions and field-induced Weyl points near the Fermi level. Magneto-transport measurements establish a rare fourfold combination of extremely large magnetoresistance, chiral-anomaly-like negative magnetoresistance, large anomalous Hall conductivity arising from cooperative intrinsic Berry curvature and skew scattering, and scalar-spin-chirality-driven topological Hall responses. This multifunctionality arises from the distinct yet weakly coupled roles of itinerant carriers and localized 4f moments, which enable high-mobility transport, field-induced Weyl topology, and non-coplanar magnetism. DyB4 therefore provides a 4f-electron platform for correlating orbital transport, momentum-space Berry curvature, and real-space spin chirality, suggesting a route toward multifunctional magnetic topological materials.

cond-mat.mtrl-sci↗
arXiv2026

Learning from the Future: Privileged Self-Distillation for Sequential Recommendation

Sequential recommenders are commonly trained with one-hot next-item labels under a causal (prefix-only) objective aligned with inference. While deployment-compatible, this supervision offers little insight into relative preferences among non-target items. Yet logged interaction sequences contain an additional supervisory source: interactions following the target often reveal how user intent evolves, making the target easier to interpret. We treat these future interactions as training-only privileged information, available during learning but not at inference. This raises a natural question: can future interactions provide richer supervision while keeping training aligned with inference-time prediction? We propose Privileged Self-Distillation (PSD), a framework that separates learning-time information from inference-time information. PSD applies two attention masks to the same backbone: a future-aware view yields a privileged teacher distribution conditioned on past and future interactions, while a prefix-only view yields the student distribution used for deployment. Distilling the privileged distribution converts future interactions into training-only supervision rather than inference-time inputs. Since both views share a backbone, the teacher's advantage is purely informational, not architectural, removing the need for a separately pretrained teacher and letting its supervision adapt as the student evolves. PSD further uses an advantage-reachability gate to focus distillation on teacher signals likely supported by the observed prefix, along with a momentum-averaged teacher for stable targets. The framework is optimized end-to-end in a single stage, leaving the deployed model and inference cost unchanged. Experiments across public benchmarks and diverse backbones show consistent improvements.

cs.IR↗
arXiv2026

MPP-GNN: Subject-Adaptive Community Detection for fMRI-Based Alzheimer's Disease Classification

Functional magnetic resonance imaging (fMRI) is a widely used technique for studying the brain. Recent methods that utilize graph neural networks (GNNs) for analysis of brain functional connectivity have shown great potential for the classification of brain disorders, such as Alzheimer's disease (AD). However, these methods often assume a preset number of functional modules across all subjects, which overlooks inter-subject variability. In addition, the discovered modules are rarely used to directly guide the learned connectivity patterns. Here, to address these issues, we propose a Meta Probabilistic Pooling GNN (MPP-GNN). We frame the model's task as a coupled, bilevel optimization that performs adaptive graph partitioning hierarchically to discover subject-specific modules and then uses the discovered brain modules as an explicit prior to guide edge refinement and representation learning. We validate MPP-GNN on two public datasets for AD classification, achieving the highest AUC in comparison to established baselines for both datasets. Furthermore, our analysis demonstrates that MPP-GNN shows significant alignment with the canonical functional-network organization defined by the Yeo brain atlas and reveals a network-level dedifferentiation pattern for AD.

cs.LG↗
arXiv2026

Reason4Rec: Deliberative User Preference Alignment of Large Language Models for Recommendation

Aligning Large Language Models (LLMs) with recommendation tasks represents an emerging paradigm in recommendation domain, exhibiting promising performance overall. However, these aligned recommendation LLMs often struggle with complex scenarios due to limitations in the current alignment task formulation, which optimizes LLMs to directly generate user feedback without deliberation. To develop more reliable recommendation LLMs, we introduce a new \textit{Deliberative Recommendation} task, which enforces explicit reasoning about user preferences as an additional alignment objective. To address this task, we propose a \textit{Reasoning-powered Recommender} framework designed to enhance reasoning capabilities by leveraging verbalized user feedback in a step-wise manner. Specifically, this framework employs collaborative step-wise experts alongside specifically crafted expert-wise training strategies. Extensive experiments conducted on three real-world datasets demonstrate the rationality of the deliberative task formulation and the effectiveness of the proposed framework in improving both prediction accuracy and reasoning quality. Our implementation is publicly available on GitHub: https://github.com/Peter-Fy/Reason4Rec.

cs.IR↗
arXiv2026

P3: Probabilistic Policy Propagation for Stable VAE-Based Robot Learning

Variational Autoencoders are widely used to encode high-dimensional and noisy observations in robotics. However, their stochastic latent creates a mismatch with Proximal Policy Optimization (PPO): an effective policy marginalizes over the latent distribution, whereas former implementations estimate its probability ratio and KL divergence using only one latent sample. We identify a fundamental but overlooked theoretical cause: naive single-sample approximations in stochastic latent space induce significant variance and bias in the surrogate loss. To address this, we introduce P^3 (Probabilistic Policy Propagation), a distribution-aware optimization framework for VAE-based policies. $P^3$ couples moment-based probabilistic method for stable and efficient learning with sampling-based calibration for robust policy behavior under latent uncertainty. In our experiments, P^3 boosts data efficiency from 64.6% to >96%, reduces convergence steps by >20%. Furthermore, P^3 is evaluated on challenging humanoid parkour tasks and shows an effective foundation for VAE-based PPO. Code is available at https://github.com/ylyem9x/P3_Open.

cs.RO↗
arXiv2026

Do Pathology Vision-Language Models Truly See Pathology?

Pathology vision-language models (VLMs) have recently progressed rapidly and are commonly evaluated by answer accuracy on pathology VQA benchmarks. However, we dig into current evaluations and identify three overlooked issues: 1) Visual evidence is not always necessary. For instance, Gemini-3-Pro achieves 53.5% average accuracy across 5 VQA benchmarks without any visual input. 2) Domain training can improve accuracy without proportional gains in visual binding. Compared with Qwen2.5-VL-7B, Patho-R1-7B exhibits a 5.8-point lower multimodal gain and a 3.7-point lower attention IoU. 3) Entity-level attention is diffuse and weakly query-specific. On PathVG, attention maps remain highly correlated across different entity queries. These issues can lead to substantial misjudgments of pathology VLMs' actual multimodal capabilities. To this end, we present PathBind, a benchmark comprising 2,600 samples: PathBind-VQA with 1,500 questions across six dimensions, PathBind-PTA with 600 questions from a private pathology teaching atlas, and PathBind-Grounding with 500 expert-curated region-level samples. Each component undergoes task-specific automated filtering and expert review to reduce textual shortcuts and improve entity-region correspondence. We evaluate 18 representative VLMs on VQA samples of PathBind and five existing pathology VQA benchmarks, and further evaluate 10 VLMs on PathBind-Grounding and PathVG. Results show that current pathology VLMs still exhibit a substantial gap between answer-side performance and visual-semantic binding.

cs.CV↗
arXiv2026

PrefReward: Learning User Preference Matrix for Personalized Text Generation

Large Language Models (LLMs) have demonstrated remarkable ability in generating personalized content by leveraging user histories and contextual cues. However, most existing personalization approaches rely on implicit representations within model parameters, making it difficult to interpret user-specific preferences or effectively handle long-context dependencies. To address these challenges, we propose PrefReward, a novel preference-aware generative framework that explicitly models user styles through a structured preference matrix and integrates it into the decoding process as a reward signal. PrefReward consists of two stages: (1) extracting a user-specific preference matrix that summarizes individual stylistic tendencies, and (2) using the matrix to guide generation via a KL-divergence-based reward function. Experiments on the LongLaMP dataset show that PrefReward outperforms non-personalized and retrieval-based baselines in both generation quality and personalization interpretability.

cs.CL↗
arXiv2026

KineBench: Benchmarking Embodied World Models via IDM-Free Kinematic Grounding

Evaluating the physical consistency of embodied world models(EWMs) is a critical open challenge. While closed-loop evaluation via simulator rollouts offers a more faithful assessment of physical plausibility than open-loop alternatives, existing frameworks almost exclusively rely on Inverse Dynamics Models(IDMs) for action extraction. Due to the intricate mapping from 2D pixel space to 3D kinematic space, the learned IDMs can be brittle to data outside their training distribution, resulting in unreliable action extraction from the generated videos with novel objects and scenarios. This creates an unavoidable attribution ambiguity between world model inaccuracies and extractor errors. To reduce this ambiguity, we present KineBench, an IDM-free closed-loop benchmark for EWMs, built upon an explicit kinematic grounding pipeline. Given a generated video, KineBench employs cascaded visual foundation models to directly extract 6D end-effector poses from individual frames, which are then executed in a physics simulator for closed-loop validation. Beyond execution-based task success, KineBench incorporates two classical 3D kinematic metrics--Spectral Arc Length (SPARC) and the Maruyama Manipulability Index--to characterize trajectory smoothness and kinematic feasibility from a robot-centric perspective. Built on 20 diverse manipulation tasks in ManiSkill3, KineBench evaluates EWMs across four progressive suites: basic execution, task transfer, visual out-of-distribution generalization, and complexity-conditioned scaling. Evaluation across frontier models reveals task-complexity-bounded nonlinear scaling in embodied video generation, providing empirical guidance for future data-scaling strategies.

cs.RO↗
arXiv2026

Conjugate Gradient Unrolled Network with PSF Conditioning for Non-Diagonal Data Fidelity in CASSI Reconstruction

Deep unfolding methods achieve state-of-the-art performance in coded aperture snapshot spectral imaging (CASSI) reconstruction but typically rely on closed-form data-fidelity updates that assume an ideal optical system. In practical CASSI systems, the field-dependent and wavelength-dependent point spread function (PSF) introduces convolutional coupling that breaks the diagonal structure of the normal equation, rendering closed-form updates inapplicable. We propose a deep unfolding framework that addresses this fundamental algorithmic challenge through three contributions: (1)~a $K$-step conjugate gradient (CG) unrolling that explicitly solves the non-diagonal normal equation under PSF-inclusive forward operators; (2)~a learned gradient refinement module with wavelength-adaptive step sizes generated from a per-wavelength PSF embedding; and (3)~a PSF-conditioned penalty estimator that adapts the ADMM regularization strength to the optical degradation severity. A Monte Carlo PSF training strategy further improves robustness to manufacturing-induced PSF variations. Our method achieves 30.53~dB on the KAIST dataset, +2.73~dB over the DPU baseline (1.27M) using a comparable number of parameters (1.42M), and +1.70~dB over a larger baseline (DPU-B+, 2.12M) using 33\% fewer parameters.

physics.optics↗
arXiv2026

Exact Nagaoka-to-spiral transition in the doped infinite-U triangular lattice

We study the single-hole infinite-$U$ Hubbard model on the triangular lattice with nearest- and next-nearest-neighbor hoppings $t_1$ and $t_2$. A frustrating $t_2$ destabilizes Nagaoka ferromagnetism through a Lifshitz transition to a long-wavelength coplanar spiral at $t_{2,c}/t_1=-0.182$. We determine the critical point analytically by reducing the many-body problem to an effective two-body hole--magnon scattering problem in which the hard-core hole--magnon constraint is treated exactly. Numerical calculations confirm the spiral ground state and reveal a coherent spin polaron with quasiparticle weight $Z\simeq0.92$ at the Lifshitz-shifted momentum $\mathbf{K}=-\mathbf{Q}^*$. These results establish the first quantum instability of the triangular-lattice Nagaoka ferromagnet and provide the magnetic baseline for magnon-mediated pairing at finite doping.

cond-mat.str-el↗
arXiv2026

Measuring gravitational wave spectrum from electroweak phase transition and Higgs self-couplings

In this work, we demonstrate the complete process of using space-based gravitational wave detectors to measure properties of the stochastic gravitational wave background arising from a first-order electroweak phase transition. Based on frequency-domain simulations of the Taiji mission, including instrumental noise and astrophysical foregrounds, we perform parameter inference using both the Fisher information matrix and Bayesian Markov Chain Monte Carlo sampling. We show how the reconstructed spectrum constrains the macroscopic parameters of the phase transition, and further how these constraints map onto the underlying particle-physics parameters in a singlet-extended Standard Model. Our results demonstrate that the Higgs cubic and quartic self-couplings can be significantly constrained using gravitational wave observations, despite limitations arising from parameter degeneracy.

hep-ph↗
arXiv2026

BADTV: Unveiling Backdoor Threats in Third-Party Task Vectors

Task arithmetic in large-scale pre-trained models enables agile adaptation to diverse downstream tasks without extensive retraining. By leveraging task vectors (TVs), users can perform modular updates through simple arithmetic operations like addition and subtraction. Yet, this flexibility presents new security challenges. In this paper, we investigate how TVs are vulnerable to backdoor attacks, revealing how malicious actors can exploit them to compromise model integrity. By creating composite backdoors that are designed asymmetrically, we introduce BadTV, a backdoor attack specifically crafted to remain effective simultaneously under task learning, forgetting, and analogy operations. Extensive experiments show that BadTV achieves near-perfect attack success rates across diverse scenarios, posing a serious threat to models relying on task arithmetic. We also evaluate current defenses, finding they fail to detect or mitigate BadTV. Our results highlight the urgent need for robust countermeasures to secure TVs in real-world deployments.

cs.LG↗