arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 19 recordsLinked to original sources

arXiv2026

Prospects for exploring quantum-correlated neutral $D\overline{D}$ decays at STCF

We investigate the prospects for determining charm mixing, indirect $C\!P$ violation and hadronic parameters in charm decays with quantum-correlated neutral charm mesons at the proposed Super Tau-Charm Facility. The study is based on time-integrated double tag rates of coherent $D\overline{D}$ pairs produced at $\sqrt{s}=3.773$ and $4.030$ GeV, from $e^{+}e^{-}\to D^{(*)}\overline{D}^{(*)}$ processes. Here, one $D$ is tagged with the quasi flavour-specific decays $D\to K^-π^+π^+π^-$, $K^-π^+π^0$, and $K^-π^+$, while the other $\overline{D}$ is detected with the final states in flavour-specific mode, $C\!P$ eigenstate mode, or self-conjugate mode. The expected sensitivities are evaluated assuming an integrated luminosity of $1~{\rm ab}^{-1}$ at each energy point, demonstrating the capability on precise determination of the charm mixing parameters, the indirect $C\!P$ violation parameters, the strong-phase differences, and the coherence factors. The constraints of the hadronic-parameters provide unique inputs to the CKM angle $γ$ fit, while the measurements of charm mixing and indirect $C\!P$ violation parameters are complementary to the time-dependent measurements in the Belle II and LHCb experiments.

hep-ex↗

arXiv2026

Thermodynamics and phase structure of rotating Hayward-AdS black holes

We use the modified Newman--Janis algorithm to obtain the rotating ``mother'' black hole solution of the rotating Hayward-AdS black hole. We further derive the fundamental thermodynamic quantities of the ``mother'' black hole in the full unconstrained phase space $(S,J,q,α,P)$, and impose the regularity condition $M_1=q^3/α$ to obtain the thermodynamic quantities of the rotating Hayward-AdS regular black hole. This order of operations gives a Hawking temperature identical to the surface gravity temperature and produces the correct response functions of the regular black hole. For $P<P_c$, the Hawking temperature $T_2$ possesses a local maximum and a local minimum, which divides the black hole into small, intermediate, and large black hole branches, and the $P$--$V$ isotherms display oscillatory behavior characteristic of a liquid/gas phase transition. The small and large black hole branches are locally stable, whereas the intermediate black hole branch has negative heat capacity and is locally unstable. Increasing the rotation parameter $a$, magnetic charge $q$, or pressure $P$ reduces the region of the intermediate branch. The $G_2$--$T_2$ diagram shows that the phase structure, in contrast to the standard swallowtail of the Reissner--Nordström-AdS and Kerr--Newman-AdS black holes, is 8-shaped or C-shaped. These two shapes are divided by the characteristic pressure $P_z$. For $P<P_z$, the curve is 8-shaped, corresponding to a first-order small/large black hole transition. For $P_z<P<P_c$, the curve is C-shaped, corresponding to a zeroth-order transition. These results show that the 8-shaped/C-shaped phase structure characteristic of regular AdS black holes persists under rotation, indicating that it stems from the restructuring of the thermodynamic phase space by the regularity condition rather than from spherical symmetry.

gr-qc↗

arXiv2026

Zhengrong-Spatial-Aware 1.0: Advancing 3D Multimodal Intelligence

This report presents the implementation and validation of two end-to-end multimodal adaptation pipelines for 3D-aware visual question answering on a custom spatial dataset. The first pipeline fine-tunes HiSpatial-3B, which augments RGB observations with metric-scale 3D point maps; the second adapts Qwen3-VL through memory-efficient QLoRA-based supervised fine-tuning. The completed engineering stack covers dataset conversion, image-annotation alignment, monocular geometry preprocessing, instruction construction, mixed-domain training, checkpoint management, batch inference, and automated evaluation. On the target spatial benchmark, HiSpatial adaptation increases overall accuracy from 1.73% before fine-tuning to 79.89% at 3,600 steps. On a separately collected styled split, accuracy improves from 14.9% to 55.1%, indicating transfer beyond the original training set. The Qwen3-VL pipeline is likewise implemented end-to-end for both 8B- and 32B-scale models using mixed-domain QLoRA adaptation. The resulting accuracy improves from 20.74% to 45.27% for Qwen3-VL-8B and from 30.79% to 47.60% for Qwen3-VL-32B. Together, these results establish two complementary and reproducible adaptation pipelines for geometry-aware and general-purpose multimodal spatial reasoning. These results establish a reproducible engineering foundation for subsequent work on metric 3D reasoning, spatial grounding, and domain-adapted multimodal systems, while also identifying training-data composition as a central factor in preserving general multimodal capability during specialization.

cs.CV↗

arXiv2026

The QCD phase diagram for three-flavor Möbius domain-wall fermions

We investigate the phase transition of Quantum Chromodynamics (QCD) with three degenerate quark flavors at zero baryon chemical potential. Using Möbius domain-wall fermions as the lattice fermion formulation, we ensure excellent chiral symmetry preservation. Our simulations are performed at three different temporal lattice extents, $N_{t}=6, 8, 12$, with a fixed lattice spacing $a=0.1361(20)$ fm, corresponding to temperatures of 242(4), 181(3), and 121(2) MeV, respectively. We explore a range of quark masses and spatial volumes with aspect ratios $N_{s}/N_{t}$ spanning from 2 to 4. By analyzing the mass and volume dependencies of the plaquette, plaquette susceptibility, chiral condensate, chiral susceptibilities, and Binder cumulant, we identify the pseudocritical transition quark masses from our largest lattice volumes. For $N_t=6$, this is 184(10) MeV (determined from the plaquette susceptibility). For $N_t=8$ and 12, the transition points vary slightly depending on whether the total or disconnected chiral susceptibility is used, yielding ranges of 36(1)-39.1(9) MeV and 3.5(3)-3.7(2) MeV, respectively, in the $\overline{\text{MS}}$ scheme at a scale of $μ=2$ GeV. The negligible volume dependence at $N_t=6$ and 8, combined with finite-size scaling analysis at $N_t=12$ revealing volume growth significantly weaker than expected for a first- or second-order phase transition, points to a continuous crossover at these specific quark mass points. Additionally, we study the effects of residual chiral symmetry breaking on the chiral condensate and chiral susceptibilities using two different values of $L_s$.

hep-lat↗

arXiv2026

CRIM: Composable and Reusable Item Memory for Efficient Personalized Retrieval

Large language models incur substantial training and inference costs in personalized search and recommendation, as growing interaction histories lengthen request inputs and repeatedly expose the same item text to the model. We propose \textbf{C}omposable and \textbf{R}eusable \textbf{I}tem \textbf{M}emory (CRIM), a framework that separates item-level semantic encoding from request-level history modeling. A small set of learnable summary tokens compresses each item into compact continuous memories, while segment-level readout tokens (REC tokens) introduce predictive supervision within the interaction history. These memories are independent of users and queries and can therefore be reused across requests. At inference time, CRIM composes cached memories in interaction order and incorporates an optional query to produce a personalized retrieval representation, reducing repeated item-text encoding while retaining request-dependent contextual modeling. We further introduce a joint training mechanism that combines cached memories with sampled live re-encoding. Re-encoding only a few history items per request allows the retrieval objective to jointly optimize the item and request encoders at limited computational cost, while periodic cache refreshes maintain consistency. Extensive evaluation and cost analysis on MovieLens-1M, Qilin, and our new LazSeq dataset show that CRIM achieves state-of-the-art retrieval performance with substantially lower inference costs at the same backbone scale, while limiting item re-encoding during joint training. We will publicly release our code and dataset.

cs.IR↗

arXiv2026

GraphDancer: Training LLMs to Explore and Reason over Graphs via Curriculum Reinforcement Learning

Large language models (LLMs) increasingly rely on external knowledge to improve factuality, yet many knowledge sources on the Web, such as scholarly indexes and product catalogs, are organized as heterogeneous graphs. Reasoning over such graph-structured knowledge poses two key challenges: (1) navigating structured, schema-defined relations requires precise function calls beyond similarity-based retrieval, and (2) answering complex questions often demands multi-hop evidence aggregation through iterative information seeking. We study how LLMs can learn these interaction capabilities and transfer them to unseen domains. We propose GraphDancer, a reinforcement learning (RL) framework that teaches LLMs to navigate graphs by interleaving reasoning and function execution. To improve RL for moderate-sized LLMs, we introduce a graph-aware curriculum that schedules training by the structural difficulty of questions using an easy-to-hard biased sampler. We evaluate GraphDancer on the multi-domain GRBench benchmark, training only on Academic and Biomedical graphs. Testing covers held-out questions from these domains as well as unseen E-commerce, Literature, and Legal graphs. With either Qwen2.5-3B-Instruct or Qwen3-14B as the backbone, GraphDancer outperforms graph-agent baselines using the same backbone across all evaluation domains. These results indicate that the learned graph exploration and reasoning capabilities transfer across domains. On average across unseen domains, GraphDancer with a 14B backbone also outperforms baselines powered by the closed-source models GPT-4.1-mini and GPT-5-nano. Trajectory analyses further show that GraphDancer retrieves answer-bearing evidence more often and achieves higher accuracy when such evidence is found, while substantially reducing invalid-protocol outcomes.

cs.LG↗

arXiv2026

Spatial Audio Generation and Editing with Executable Plans and Self-Distilling Transfusion

Spatial audio authoring coordinates source decisions with soundfield synthesis. For editing, plans specify what to change and retain, while reference audio supplies further acoustic detail. We present AMBIT, a unified framework for native First-Order Ambisonics (FOA) generation and editing with three connected contributions. First, a shared Transfusion backbone couples autoregressive planning with rectified-flow rendering through executable ScenePlans that specify source content, activity, and 3D trajectories. For editing, the plan describes the full target scene, including retained sources; contrastive features ground the requested change, while full reference latents supply residual acoustic detail for preservation. Second, an FOA-aware continuous VAE combines shared-decoder omnidirectional recovery, asymmetric grouped KL, and spatial covariance matching to preserve content and inter-channel structure. Third, joint Transfusion on-policy self-distillation (OPSD) refines planning on student-generated prefixes and rendering at sampling states reached under student plans, with target-conditioned teachers supervising both modes of the shared student. Evaluations across speech, music, and sound show improved generation content and spatial quality, acoustic fidelity during instruction-guided editing, and unedited-source preservation over the compared baselines. Project page: https://wjm-george.github.io/AMBIT/

cs.SD↗

arXiv2026

MLIPDiff: Guiding Denoising Diffusion Models for Structure-Based Drug Design via Machine-Learned Interatomic Potentials

Diffusion models have become a dominant approach to structure-based drug design. However, because they are trained primarily to fit reference structures, they often place ligands in physically implausible poses that clash with the protein. We introduce MLIPDiff, an inference-time guidance method that can be applied to pretrained diffusion models for ligand generation. The method uses physics-informed guidance from UMA (Universal Models for Atoms), a machine-learned interatomic potential (MLIP). The guidance combines four weighted energy terms (protein-ligand interaction, ligand strain, steric clash, and pocket anchoring) and uses the original model checkpoint without retraining. On CrossDocked2020, MLIPDiff completely eliminates protein-ligand collisions while preserving the ligand's local geometry, improving bond-length distributions and physical plausibility relative to the base model. Compared to models such as NucleusDiff that are specifically designed to reduce atomic clashes, MLIPDiff achieves clash-free structures without compromising local geometry. It maintains the base model's PoseBusters pass rate and attains the best median Vina Score, Vina Min, and Vina Dock, surpassing both the base model and NucleusDiff, with a median of 59.6\% of ligands per pocket binding more strongly than the reference. MLIPDiff thus addresses a limitation of prior methods by generating physically plausible structures with favorable predicted binding affinity using a single frozen model and no additional training.

cs.LG↗

arXiv2026

PACE: Plan, Attend, Check, and Emit for Training-Free Response Timing in Streaming Video Understanding

Streaming video understanding requires a model to decide not only what to answer but also when to answer. Evidence arrives progressively, and recent benchmarks identify premature responses, issued before the queried event has concluded, as a recurring failure mode. Existing approaches often learn response timing through additional training, while training-free alternatives typically rely on scalar gating signals that do not explicitly represent when an answer becomes final. We organize chunk-wise streaming errors into three patterns: sampling variation can make an answer unstable, a stable answer may lack support in the observed video, or an answer may be supported by the current evidence while the event remains unfinished. We propose PACE (Plan, Attend, Check, and Emit), a training-free framework designed around these three failure modes. Given a question, PACE drafts a plan specifying what to observe and what will signal completion. As the stream unfolds, it invokes an off-the-shelf video LLM at informative moments, evaluates each draft for stability, grounding in its cited frames, and completion, and records verified observations in an event ledger. Experiments on ProactiveVideoQA and OVO-Bench Forward show that PACE improves over the official protocols by up to 5.9 average PAUC points and 6.2 average accuracy points across proprietary and open-source backbones, yielding state-of-the-art performance.

cs.CV↗

arXiv2026

Auto-Director: Leveraging Multi-Agent Collaboration for Autonomous Video Editing

Autonomous video editing from open-world resources remains a challenging problem in multimedia content creation. Retrieval-based methods are constrained by fixed source material and lack creative flexibility, while generative models struggle with long-horizon coherence and often diverge from user intent. Both paradigms fail to replicate the structured, multi-stage reasoning that underlies professional post-production. To address these limitations, we present Auto-Director, a tool-augmented multi-agent framework that mirrors professional editorial workflows: a Planning Agent decomposes abstract topics into fact-grounded scripts via retrieval-augmented planning; a Material Agent indexes open-world footage through shot segmentation and visual tagging; an Editor Agent assembles semantically aligned clips with MLLM-driven spatial adaptation; and a Critic Agent governs an iterative Reflect-and-Revise loop that checks scene-level alignment and pacing. To rigorously evaluate this new paradigm, we introduce VEdit-Bench, a benchmark of 100 expert-curated topics with human-annotated Golden Scripts and a five-dimensional metric suite covering Narrative Logic, Factual Accuracy, Semantic Alignment, Compositional Quality, and Stylistic Cohesion. Extensive experiments demonstrate that Auto-Director significantly outperforms baseline methods in both automated metrics and human evaluation, effectively bridging the gap between amateurs and professional-grade video production.

cs.CV↗

arXiv2026

Bezier-Guided Adaptive Flow for Cross-domain Sequential Recommendation

Cross-domain sequential recommendation aims to alleviate data sparsity and cold-start problems by transferring useful preference knowledge across domains. However, users exhibit varying degrees of preference consistency across domains, while existing methods often rely on global alignment objectives or coarse-grained transfer strategies. Such approaches may introduce irrelevant information and provide limited flexibility in adapting semantic representations to individual user preferences. To address these challenges, we propose BGAFlow, a personalized framework that combines state-adaptive semantic evolution with user-specific transfer control. BGAFlow first establishes a shared semantic space through unified dual-domain pretraining with user profiles inferred by a large language model (LLM). Based on latent interest overlap, it then estimates a personalized horizon to determine the appropriate transfer extent for each user. To enable semantic adaptation, BGAFlow constructs a conditional Bezier curve as a geometric reference and learns a residual flow that progressively refines the trajectory based on intermediate preference states. The learned flow incorporates dynamic preference feedback to adjust transfer directions, while recommendation supervision and trajectory regularization encourage effective and smooth adaptation. Extensive experiments on real-world datasets demonstrate that BGAFlow consistently outperforms state-of-the-art baselines, validating the benefits of personalized transfer control and state-adaptive semantic evolution.

cs.IR↗

arXiv2026

Differentiated Determinant Graphs Enable Large-Scale Multideterminant Auxiliary-Field Quantum Monte Carlo

Multideterminant trial wave functions can reduce phaseless bias in auxiliary-field quantum Monte Carlo (AFQMC), but the repeated evaluation of their overlap, force bias, and local energy restricts systematic trial improvement. In this work, we organize particle--hole excitation determinants as a shared arithmetic directed acyclic graph of canonical minors. A coefficient-weighted reverse pass and batched directional passes evaluate the complete estimator without determinant-resolved cofactor tensors while preserving the sparse $α$--$β$ coefficient coupling exactly. Controlled complex-valued tests verify the complete kernel to machine precision and show orders-of-magnitude acceleration over direct determinant summation. In molecular calculations, the observed cost remains sublinear through million-configuration selected-CI trials and the same estimator remains usable at one hundred million configurations, where the eventual exact coefficient-contraction boundary becomes visible. For Cr$_2$, an orbital-optimized CAS(12e,42o) trial in a double-$ζ$ basis yields a potential-energy curve (PEC) and spectroscopic constants in close agreement with experiment. Differentiated determinant graphs therefore convert very large correlated trials from a formal accuracy hierarchy into a practical route for systematic AFQMC studies of strongly correlated molecules.

physics.chem-ph↗

arXiv2026

CiteAnchor: Source-Aware Grounding for Hallucinated Citation Detection and Resolution

Hallucinated citations make it harder to trace sources and verify the evidence behind scientific claims. Detecting them is difficult because bibliographic records can be incomplete or differ across source types and publication versions. A plausible match alone neither validates the citation nor establishes its underlying source. We introduce CiteResolve, a benchmark of 423 real-world hallucinated citations annotated for source recoverability. We further propose CiteAnchor, a source-aware grounding framework that adapts retrieval and field-level verification to each source type and separately determines citation validity and whether a unique source can be recovered. On an existing detection benchmark, CiteAnchor achieves 0.841 F1 with GPT-5.6 Luna and 0.784 with an open-weight backbone, outperforming the baseline with the same backbones by 7.9 and 36.8 points. On CiteResolve, it achieves 74.7% and 70.7% end-to-end task success, outperforming the corresponding web search baselines by 16.3 and 65.3 percentage points. Our code and data are available at https://github.com/oldwizard7/CiteANCHOR_code.

cs.CL↗

arXiv2026

Chiral Cavity Vacua as Local Generators and Probes of Non-Abelian Kitaev Topology

We show how a gyrotropic cavity vacuum generates, probes, and spatially programs non-Abelian topology in the Kitaev honeycomb model. Gauge-consistent virtual-photon elimination produces a universal helicity-odd interaction $\mathrm{i}[P_x,P_y]$ alongside same-order even dressing. For the leading bond polarization, this commutator is exactly the Kitaev three-spin mass. Mode normalization imposes a thermodynamic constraint: one extended normalized mode gives a local mass that vanishes with sample area, whereas a localized mode writes a finite topological island and a finite mode density enables a bulk mass. The exact Majorana solution yields helicity-reversed Chern gaps and edge propagation, while integrated circular dichroism exactly measures the mass generator, $\langle\mathrm{i}[P_x,P_y]\rangle=-\sqrt3p^2\langle O_3\rangle$. Real-space calculations reveal a non-Abelian island and a helicity wall with two co-propagating Majorana modes. Uniform helicity reversal is a Dirac-mass quench with $n_{\rm ex}\proptoτ_Q^{-1}$, whereas moving a wall localizes the gap closing. The equilibrium vacuum kernel contains sum-frequency rather than resonant denominators. These results establish distinct criteria for local and bulk cavity-induced topology.

cond-mat.mes-hall↗

arXiv2026

RealMe: Immersive Volumetric Video from Minimal Captures

This paper addresses the problem of creating immersive volumetric videos from sparse and low-resolution capture resources. Conventional dense multi-view reconstruction methods can produce high-quality volumetric videos, but their reliance on costly capture setups makes them impractical in real-world scenarios. Recent approaches use generative models to relax capture requirements, but still suffer from limited resolution and 4D inconsistency. To overcome these constraints, we propose RealMe, a unified framework built around a high-resolution video diffusion model that enhances limited captures into high-resolution (2K), 4D-consistent videos for volumetric video reconstruction. Specifically, RealMe constructs auxiliary data from capture videos and encodes them into latents along with layout-aware positional embeddings. The video diffusion model adapts factorized spatio-temporal attention to the multi-view capture layout, denoises these latents across viewpoints and time, and decodes them into enhanced videos to refine the 4D representation. This adaptation helps the model capture temporal dynamics and multi-view relationships, leading to more 4D-consistent generation. To reduce the data gap between generated videos and real captures, we apply a latent frequency loss that provides frequency-domain supervision in latent space. Extensive experiments on the public DNA-Rendering benchmark and our self-captured RealVV dataset demonstrate that RealMe is able to create immersive volumetric videos and outperforms state-of-the-art methods in visual fidelity and 4D consistency.

cs.CV↗

arXiv2026

KPHarness: Co-Evolving Knowledge and Probing for Long-Horizon Gaming Harnesses

A failed gaming trajectory reveals that an agent needs intervention, but not necessarily how its execution harness should change. Effective harness adaptation therefore requires gathering diagnostic evidence, rather than merely proposing alternative recovery rules. We present KPHarness, a framework that adapts the external execution harness of a frozen vision-language-action model through knowledge-probing co-evolution. Recovery knowledge supplies context-specific intervention hypotheses that guide active probing; probe outcomes reveal when these hypotheses apply and provide evidence for refining both recovery knowledge and later probe selection. This reciprocal process allows the harness to improve not only which interventions it applies, but also how it investigates failures. KPHarness separates fast policy execution from a slow research loop that translates accumulated knowledge and probing evidence into bounded harness revisions. Each revision is compared with the incumbent under matched sandbox conditions and accepted only when the paired evaluation shows improvement. Recoveries supported by further validation are retained for cross-task reuse, while immediate interaction feedback and delayed episode-success credit update action- and knowledge-probing policies during adaptation. Experiments on Minecraft tasks spanning mining, combat, and crafting show improved task coverage over bare policy execution, a manually specified fixed harness, and knowledge-only harness evolution with non-adaptive probing. Evaluation on previously unseen tasks further shows improved task completion with the evolved harness, knowledge bank, and probing policies held fixed,

cs.AI↗

arXiv2026

AGORA: An Archive-Grounded Benchmark for Agentic Workplace Document Reasoning

Large language models are increasingly deployed as agents that reason over documents rather than answer from parametric knowledge. We study archive-grounded reasoning: locating sparse evidence across a large, messy collection of workplace files, reconciling inconsistent terminology, units, and time conventions, and computing an answer. Existing benchmarks address only parts of this setting and none jointly stresses archive-groundedness, agentic exploration, and cross-domain coverage. We introduce Agora, a benchmark pairing 362 questions with eight domain collections of 9,664 authentic documents and 372M tokens, far exceeding any model's context window, so agents must explore deliberately rather than scan exhaustively. Agora is built by an agentic pipeline combining cross-document task synthesis, leakage-preventing obfuscation, and difficulty filtering. Evaluating eight models, we find the task far from solved: even the strongest reaches only 59.4% accuracy, with notable variation across domains.

cs.CL↗

arXiv2026

Brain-to-Language Decoding: Paradigms, Signals, Methods, Evaluation, Practical Use and Beyond

Brain-to-language decoding translates neural activity associated with language production, internal speech and perception into linguistic or expressive outputs. It offers a route to restoring communication after speech loss and a means of studying how the brain represents language. Advances in neural recording and representation learning have expanded the field from constrained recognition and acoustic reconstruction to text generation, streaming personalised speech and facial animation. This survey synthesises these developments across invasive and non-invasive measurements. We connect Articulated, Inner and Perceived paradigms to the neural populations they engage, the representations available to decoders and the outputs those representations can support. We examine model development, public resources and the evolution of evaluation, and compare published performance and communication costs within their reported protocols. The synthesis identifies complementary routes to progress: phonetic, acoustic and semantic targets preserve different aspects of a message; shared representations support reuse across recording conditions and tasks; and online communication increasingly depends on calibration, feedback and user control alongside decoding accuracy. Shared benchmarks enable algorithmic comparisons, while longitudinal studies reveal the demands of sustained use. We discuss these developments and their remaining limitations, then outline a prospective five-level trajectory from commands and language to meaning, scenarios and bidirectional cognitive exchange.

cs.CL↗