arXiv subjects

Zhe Wang

Publications and source records attributed to Zhe Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Spotter: Let the Embodied Model Lead, and the VLM Reflect for It

Current embodied models do not respond to their own failures, although what just went wrong could inform a small adjustment on the next attempt, the kind of reflection behind the gains of thinking in language models. We test whether they can repair a known error, which requires producing a correction and judging whether it is right. Stopped at a failure and allowed to retry, they seldom repair it through their own randomness or from a language description of the error, and best-of-N selection cannot pick the successful candidate after a failure. We attribute this to training only on successful demonstrations and to inputs too narrow to show what went wrong, and conclude that reflection must come from a vision-language model (VLM), which takes in far more information, such as the episode history and text, and is more general. Prior VLM-led work has the VLM plan every step and invoke the embodied model as a tool, placing the VLM on the critical path. We propose Spotter, which reverses the roles: the embodied model leads and executes continuously, while the VLM runs in parallel, monitors through a lightweight local screener, intervenes only when an error is detected, reflects on and corrects it, and returns control. We run Spotter with Qwen and with GPT as the VLM, and both improve the embodied models; with GPT, Spotter improves Cosmos Policy and $π_{0.5}$ by 5.6 and 7.5 percentage points on RoboCasa, and raises $π_{0.5}$ from 47.2% to 57.0% on the Hard setting of RoboTwin 2.0 and from 53% to 83% on a real robot. Because the VLM steps in only when an error is confirmed, a successful episode with Qwen takes only 13 to 16 s longer than with the embodied model alone and about 70% less time than with a VLM-led baseline using the same model. Our code is available at https://github.com/zqc3117/Spotter.

cs.RO↗

arXiv2026

Adaptive Reward Routing: Dynamic Multi-Reward Optimization for Joint Audio-Video Diffusion via Forward-Process RL

Multi-reward guided reinforcement learning (i.e., RL) offers a promising way to improve joint audio-video diffusion models along several complementary objectives, including modality-specific quality, cross-modal semantic alignment, and temporal synchronization. Its effectiveness, however, depends on two quantities that change during training: where reward-driven updates should act, and how competing rewards should be combined. Existing methods tend to rely on fixed routing and reward weights, failing to track evolving model functions. To address these limitations, we propose Adaptive Reward Routing to jointly adapt update locations and reward coordination during forward-process RL (i.e., DiffusionNFT) of joint audio-video diffusion models. Our method consists of two components. (i) Cross-Modal Influence-Guided Routing (Localizing Updates): We use bidirectional cross-attention responses as an efficient proxy for evolving cross-modal influence, dynamically reweighting token-aware losses and scaling gradients across cross-modal layers without additional model interventions. (ii) Preference-Preserving Modality-Aware Reweighting (Coordinating Rewards): We preserve predefined weights as preference priors and use branch-specific reward-gradient interactions as residual corrections after warm-up. This resolves evolving conflicts without letting dominant rewards suppress weak but essential objectives. Extensive experiments demonstrate consistent improvements in modality quality, semantic consistency, and audio-video synchronization over strong RL baselines. Ablations and mechanism analyses further validate the complementary benefits of adaptive update routing and reward coordination.

cs.CV↗

arXiv2026

Muon Detection and Direction Reconstruction with the Upgraded 1-ton Water Cherenkov Prototype Detector at CJPL-I

The 1-ton prototype detector for Jinping neutrino experiment (JNE-1ton) has completed its hardware upgrade and has been successfully operated in both water and liquid-scintillator modes at CJPL-I, which is situated under a 2400-m rock overburden. During the upgrade, the detector's mechanical support structure and PMT layout were redesigned. The original Hamamatsu PMTs were replaced with 8-inch MCP-PMTs from North Night Vision, increasing the PMT count from 30 to 60. The detector was then operated in water mode for 85 days (Water-I) and 79 days (Water-II). From the accumulated water-mode data, we derive a muon detection efficiency (Water-II) that is approximately 59% higher than that of the pre-upgrade liquid scintillator detector. The measured muon flux is $ϕ_{\text{I+II}} = (3.55 \pm 0.43_{\mathrm{stat}}\pm 0.28_{\mathrm{syst}}) \times 10^{-10}~\mathrm{cm}^{-2}\mathrm{s}^{-1}$, which is consistent with the previous measurement. Owing to the characteristic angular dependence of Cherenkov radiation and the increased PMT coverage, the uncertainty in muon direction reconstruction is approximately 6$^\circ$, which corresponds to a reduction to 30% of its previous value. Notably, one rare up-going muon event was clearly identified. It is excluded from the cosmic-ray muon flux sample and opens a new window for neutrino-induced event studies in the deepest underground laboratory in China. This work marks the first time that CJPL has employed a cost-effective water detector for muon flux measurement.

hep-ex↗

arXiv2026

No infinite spin in the $n$-body problem in all dimensions

In the planar $n$-body problem, Moeckel and Montgomery \cite{MM25} showed that there is no infinite spin for total collision solutions, when the reduced and normalized configuration converges to an isolated central configuration. Following their approach, generalizations have been obtained for partial collision solutions and parabolic solutions in the planar case, and for total collision solutions in the spatial cases by various authors, all under the same assumption that the reduced and normalized configuration converges to an isolated central configuration. Here we prove there is no infinite spin for any collision or parabolic solution in all dimensions without such an assumption.

math.DS↗

arXiv2026

When Does the Concept of "Dog" Emerge in an Audio LLM?

Multimodal large language models answer audio questions, but how they represent auditory semantics and use them in decisions remains unclear, limiting our understanding of response formation. We study dog barking in Qwen2.5-Omni-7B using Jacobian lens (J-lens) readout and directional interventions. We define the dog direction as a J-lens-derived hidden-state vector associated with dog; adding or removing its component modulates dog-related information. We find this information decodable without dog/bark prompt cues or animal-identification requirements. Directional interventions change response tendencies and some final answers, with effects concentrated in late-layer states immediately before generation across species classification, vocalization classification, and sound description. The dog direction shows no comparable advantage over controls in animal/other classification. These results provide causal-intervention evidence that the dog direction affects output scores in a task-dependent manner, most consistently at L22 and L24 immediately before generation.

cs.AI↗

arXiv2026

LIGE-GR: A Smooth Leap from Ranking to Generative Recommendation in the LLM Era

The remarkable success of large language models (LLMs) has provided important inspiration for the next generation of recommender systems. Structurally, recommendation and language generation share a similarity: both aim to produce an ordered sequence that optimizes the user's experience. However, how to precisely absorb the essence of the LLM paradigm into mature industrial recommender systems remains an open problem. There are two challenges. First, it is unclear how to incorporate the LLM paradigm -- sequence-level generation and optimization -- into recommendation. Second, real-world recommender systems are mature systems that have been iteratively customized for years around specific products, business constraints, serving infrastructure, and organizational ownership. Replacing such systems wholesale is often technically risky and organizationally disruptive. In this paper, we propose LIGE-GR, a listwise generation and evaluation recommendation framework that upgrades from a traditional ranking system (itemwise recommendation) toward a generative recommendation paradigm. Instead of rebuilding the entire recommendation stack from scratch, LIGE-GR generalizes the existing pointwise recommendation system into a listwise generation system. This allows mature recommender systems to benefit from listwise optimization while preserving compatibility with existing models, value functions, and serving infrastructure. We validate LIGE-GR in short-video recommendation on Instagram Reels and Facebook Video. On these recommendation surfaces, LIGE-GR improves time spent by 1.14 percent on Instagram Reels and 0.72 percent on Facebook Video, while requiring only modest additional inference resources.

cs.LG↗

arXiv2026

CDFCI: High-Performance Parallel Software for Quantum Many-Body Eigenvalue Problems

CDFCI is a shared-memory parallel numerical program for computing low-lying eigenpairs of large-scale, non-relativistic fermionic Hamiltonians. The software is designed to handle a broad class of many-body quantum models, including both ab initio electronic structure Hamiltonians and lattice-based Hamiltonians arising in condensed matter physics. CDFCI combines an efficient coordinate-descent-based selected configuration interaction algorithm with dedicated parallelization strategies, achieving high performance on modern multi-core architectures. Benchmark results on representative quantum chemistry and condensed matter test cases demonstrate that CDFCI attains state-of-the-art accuracy with competitive performance compared to established selected configuration interaction (such as CIPSI or SHCI) and DMRG implementations. The software is open-source, extensively documented, and provides a Python interface for seamless integration with PySCF and other many-body simulation workflows.

physics.comp-ph↗

arXiv2026

Holomorphic Maps from $\mathbb{C}^p$ into Semi-Abelian Varieties

We prove the Second Main Theorem with truncation level one for holomorphic maps with Zariski-dense image intersecting a reduced effective Cartier divisor: Let $A$ be a semi-abelian variety with an equivariant compactification $A\subset\overline A$, and let $f:\mathbb C^p\to A$ be a holomorphic map with Zariski-dense image. Let $D$ be a reduced effective divisor on $A$ extending to a divisor $\overline D$ on $\overline A$. After possibly replacing $\overline A$ by another equivariant compactification depending only on $D$ and independent of $f$, for every $\varepsilon>0$, \[T_f(r,\overline D)\leq_{\mathrm{exc}}N_f^{[1]}(r,D)+\varepsilon T_f(r,\overline D).\]

math.CV↗

arXiv2026

FootprintRAG: Visual Analytics for Evidence Context Refinement in RAG-based Scientific Literature Exploration

Retrieval-Augmented Generation (RAG) is increasingly used to ground large language model (LLM) outputs in scientific literature. However, in open-ended literature exploration, the evidence context used for generation is often produced through hidden retrieval, reranking, assessment, and filtering steps. Users may receive retrieval summaries without knowing how the system constructed the evidence context, which evidence units were retained or discarded, or whether potentially useful evidence was excluded before synthesis. We present FootprintRAG, an LLM-agent-powered visual analytics system for evidence context refinement in RAG-based scientific literature exploration. The core idea is to treat the RAG evidence context as an explicit, inspectable, and revisable analytical object before generation. FootprintRAG parses scientific literature into text and figure evidence units, expands an initial query into parallel query variants, retrieves and assesses evidence across iterative rounds, and surfaces ERS-ranked supplementary candidates from the corpus-level evidence space. Through coordinated views, the system connects retrieval trajectories, evidence-state revision, and provenance-aware summary generation into a user-steerable workflow. We evaluate FootprintRAG through two case studies, a user study, and a workflow-level comparison with representative RAG systems. The results show that FootprintRAG helps users compare retrieval directions, revise candidate evidence, recover potentially overlooked evidence, and trace generated summaries back to supporting evidence units. FootprintRAG is available at https://github.com/meteorshowering/FootprintRAGVA.git.

cs.GR↗

arXiv2026

Collective Symmetry and Criticality from Single-Component Fluctuations

Symmetry plays a fundamental role in modern physics by guiding the classification of phases and the construction of effective theories of critical systems. Identifying collective symmetry remains challenging because tests based on preselected order parameters can overlook hidden components. A central goal is therefore to extract symmetry and critical information from a single accessible component. We pursue this goal using Lee--Yang zero ratios generated by a single ordering field. Unmeasured components shape the projected fluctuation distribution and hence the relative zero positions. Deep in a phase with spontaneous symmetry breaking, we show analytically that the zero ratios for an isotropic order parameter approach a parameter-free Bessel sequence fixed by the number of collective components. Near quantum criticality, fluctuations deform this distribution. The resulting ratio deviations and their finite-size coupling dependence carry additional critical information. We demonstrate these results with large-scale quantum Monte Carlo simulations of many-body systems with conventional and enlarged continuous symmetries. This approach provides quantitative access to collective symmetry and criticality without measuring every competing order.

cond-mat.stat-mech↗

arXiv2026

Visual Graph Reasoning via Knowledge Compilation

Visual graph reasoning requires answering graph-theoretic questions directly from graph images, where graph topology and state are conveyed visually rather than given in symbolic form. Despite recent progress of vision-language models (VLMs), current approaches to visual graph reasoning still fail on simple visual graph problems. This reveals a fundamental limitation of existing approaches: they prioritize final-answer supervision over the intermediate recovery of an explicit graph representation that preserves graph topology and state from visual input. To address this limitation, we propose VGCompiler, a compilation-centric paradigm for visual graph reasoning via knowledge compilation. VGCompiler organizes reasoning around two compilers: a representation compiler that recovers a structure-preserving intermediate graph representation from visual input, and an operation compiler that compiles query intent under the recovered graph state into an executable graph operation. Specifically, we build VGCompiler on Qwen3-VL-8B and train it with reinforcement learning guided by a layered reward over executability, compiled graph validity, representation quality, and operation quality. VGCompiler uses a frozen observer to summarize graph and question conditions into lightweight signatures, enabling archive retrieval and code reuse across similar regimes. Experiments on three benchmarks GVLQA, VisionGraph, and VGCURE, show that Qwen-VGCompiler, built on an 8B backbone, surpasses the strongest closed-source VLM baseline by 28.9% and the strongest code-based baseline by 23.7%, while maintaining high efficiency. We further evaluate VGCompiler on three real-world domains, including metro routing, logistics delivery, and network fault assessment, where it generalizes across heterogeneous visual graphs and domain-grounded tasks.

cs.MM↗

arXiv2026

CATVis: A Collaborative Multi-Agent Workflow for Turbomachinery Simulation Data Visualization

Recent advances in AI for Science have enabled natural language (NL) interfaces for scientific data analysis. In turbomachinery CFD post-processing, translating ambiguous high-level analytical goals (e.g., vortex identification) into precise visualization procedures supporting complex domain-specific analysis is challenging. We present CATVis, a Collaborative multi-agent workflow system that bridges this gap by transforming NL intents into structured middle representation for visualization. Our approach reformulates domain-specific visualization procedures as composable workflow representations, and use multi agent to generate workflow representations via intent planning, template generation, and error-aware refinement, where each stage incrementally updates a shared structured representation. We evaluate the impact of external knowledge and workflow structuring on generation accuracy, demonstrating that the proposed approach significantly improves complex workflow generation correctness while reducing prompt complexity.

cs.HC↗

arXiv2026

Magnetization processes and spin dynamics across field-induced phase transitions in the quasi-two-dimensional quantum magnet Cu$_2$(OH)$_3$Br

We present magnetic-field-dependent evolution of magnetization and spin dynamics in the quasi-two-dimensional spin-$1/2$ magnet $\mathrm{Cu_2(OH)_3Br}$, consisting of alternately coupled ferromagnetic Cu1 and antiferromagnetic Cu2 spin chains. Terahertz spectroscopy reveals a pronounced field-direction dependence of the low-energy magnetic excitation spectrum. For magnetic fields applied perpendicular to the spin chains, $B\parallel a$ and $B\parallel c^*$, the spectra undergo abrupt reconstructions at the spin-flop transitions identified independently by high-field magnetization measurements. For $B\parallel b$, by contrast, no spin-flop occurs; instead, the excitation spectrum evolves continuously with field and exhibits a strong terahertz radiation polarization dependence as the ferromagnetic Cu1 subsystem becomes progressively polarized. At higher fields, the complex low-field spectrum is replaced by a reduced set of broad excitations, consistent with a weakening of the coupling between the ferromagnetic and antiferromagnetic chain subsystems. Complementary Raman spectroscopy resolves magnon and spinon excitations alongside several phonon modes and traces the characteristic temperature- and magnetic-field-dependent evolution of the magnetic excitations. The combined spectroscopic and magnetization results map out how a magnetic field reorganizes the coupled ferromagnetic and antiferromagnetic subsystems in $\mathrm{Cu_2(OH)_3Br}$ across field-induced phase transitions.

cond-mat.str-el↗

arXiv2026

Brillouin Spectroscopy Reveals Mechanical Properties Beyond Hydration

Brillouin light scattering is a noncontact technique for probing the micromechanical properties of cells and tissues through the GHz-frequency longitudinal modulus. Interpreting Brillouin spectra in hydrated biological materials remains challenging because the Brillouin shift is influenced not only by water content but also by polymer network mechanics, relaxation processes, and viscous dissipation. Here, we investigated two model hydrogels with distinct network chemistry, together with a binary solvent mixture, to disentangle these contributions. Time- and space-resolved measurements of a hydrating hydrogel further show that Brillouin microscopy captures local mechanical changes that bulk rheometry only partially resolves. Experiments on ethanol-water mixtures further demonstrate that the Brillouin shift does not vary monotonically with hydration but instead reflects changes in the mixture's mechanical properties. In addition, by comparing longitudinal, bulk, and shear viscosities, we identify bulk viscous dissipation as an important contributor to the Brillouin response. Together, these results show that hydration modulates Brillouin spectra but does not fully determine them. Our findings provide a mechanical framework for interpreting Brillouin measurements in hydrated and biomolecular systems.

physics.optics↗

arXiv2026

InfScene-SR: Seamless Super-Resolution of Arbitrarily Large Remote-Sensing Scenes via Variance-Preserving Joint Denoising

Diffusion models now give the best perceptual quality in super-resolution (SR), but their architecture and training confine them to small fixed crops. Operational remote sensing needs seamless scenes orders of magnitude larger. Joint denoising fuses overlapping tiles at every reverse step and lets text-to-image diffusion generate beyond its training crop, but it assumes deterministic ODE samplers. With the stochastic sampler of SR models such as SR3, the averaging also partly cancels independent per-tile noise. This known variance erosion blurs the detail SR should recover and must be corrected. We carry variance-corrected fusion to conditional SR and derive Spatially-Decoupled Variance Correction (SDVC), an exact reformulation that replaces per-step global normalization with independent per-tile contributions accumulated in one additive pass. SDVC turns the coupled per-step computation into independent tile-local work, so the resulting pipeline, InfScene-SR, runs in parallel across GPUs and makes SR of arbitrarily large scenes feasible. On a 5$\times$ SR task built from NAIP aerial imagery, we evaluate whole scenes with fidelity, perceptual, seam-continuity, and faithfulness metrics. Under one backbone, InfScene-SR is the only fusion strategy that is seamless and sharp at once, the closest to the low-resolution observation among those that synthesize detail, and within 0.003 IoU of native high-resolution imagery on downstream invasive-plant segmentation. Code is available at https://github.com/TitorX/infscene-sr.

cs.CV↗

arXiv2026

Scaling Muon for Diffusion Transformers

The matrix-aware optimizer Muon improves large model training by balancing updates across singular directions, yet its scaling behavior and end-to-end efficiency on large Diffusion Transformers (DiTs) remain unclear. We first establish Muon's scaling behavior on DiTs from 1.3B to 15B parameters, showing that its optimization and generative quality advantages over AdamW persist across model scales. However, at scale, the 5-step Newton--Schulz iteration (NS5) performed at every optimization step, together with full-momentum materialization, introduces substantial computation and communication overhead that can offset Muon's step-efficiency advantage. We introduce \emph{Periodic Row-wise Muon}, which performs a full NS5 spectral update once every $K$ steps and applies a low compute and communication cost row-wise constrained update based on the current momentum at the remaining steps. We further co-design a distributed implementation that operates directly on sharded momentum during non-refresh steps and accelerates spectral refreshes through bucketed all-gather and communication--computation overlap. Across all scales, Muon improves the best observed generative quality over AdamW by 12.9--19.1\%. Compared with vanilla Muon, Periodic Row-wise Muon remains within 0.5\% in best generative quality on the 1.3B--4B models and improves it by 4.5\% at 9B. It reduces optimizer time by 46.9--54.3\%, end-to-end step time by 15.7--24.3\%, and logical communication volume by 66.7\%, while reaching its respective best generative quality with 33.7--64.8\% less active training time. These results show that Periodic Row-wise Muon preserves Muon's generative quality advantage while translating it into end-to-end training efficiency for large DiTs.

cs.LG↗

arXiv2026

ShotVerse: Advancing Cinematic Camera Control for Text-Driven Multi-Shot Video Creation

Text-driven video generation has democratized film creation, but camera control in cinematic multi-shot scenarios remains a significant block. Implicit textual prompts lack precision, while explicit trajectory conditioning imposes prohibitive manual overhead and often triggers execution failures in current models. To overcome this bottleneck, we propose a data-centric paradigm shift, positing that aligned (Caption, Trajectory, Video) triplets form an inherent joint distribution that can connect automated plotting and precise execution. Guided by this insight, we present ShotVerse, a ``Plan-then-Control'' framework that decouples generation into two collaborative agents: a VLM (Vision-Language Model)-based Planner that leverages spatial priors to obtain cinematic, globally aligned trajectories from text, and a Controller that renders these trajectories into multi-shot video content via a camera adapter. Central to our approach is the construction of a data foundation: we design an automated multi-shot camera calibration pipeline aligns disjoint single-shot trajectories into a unified global coordinate system. This facilitates the curation of ShotVerse-Bench, a high-fidelity cinematic dataset with a three-track evaluation protocol that serves as the bedrock for our framework. Extensive experiments demonstrate that ShotVerse effectively bridges the gap between unreliable textual control and labor-intensive manual plotting, achieving superior cinematic aesthetics and generating multi-shot videos that are both camera-accurate and cross-shot consistent.

cs.CV↗

arXiv2026

Self-Calibrating Dense Displacement Fields for Reliable Co-Registration of Large Optical Satellite Imagery

Co-registration underlies nearly every multi-temporal and multi-sensor use of optical satellite imagery, and operational products still carry documented offsets well above the fraction-of-a-pixel scale at which change detection, time series, and data fusion degrade. Real image pairs differ along several axes at once (sensor response, scene content, viewing geometry, resolution, mosaic seams), and the last of these is not a single global motion. Existing tools embed a motion model and constants tuned to their development data; a pair that fits is registered precisely, while one that does not either fails to match or returns a result wrong by tens of pixels with no failure reported. Learned matchers add a GPU requirement and carry no accuracy guarantee outside their training distribution. We present SCDF (self-calibrating displacement fields), a training-free, GPU-free estimator whose motion model is the dense per-pixel displacement field itself, so no scene motion falls outside the model. A single predict--measure--filter loop runs over a resolution pyramid: the accumulated field predicts where each patch of the moving image falls in the reference, RootSIFT matching and a correlation pass measure the displacement there to sub-pixel precision, and filters whose thresholds are all calibrated on the image pair itself decide what survives. One configuration, with no per-dataset tuning, processes full $8192^2$ scenes on a single CPU core. On 584 constructed-ground-truth pairs built from real Sentinel-2, Landsat-8/9, and NAIP imagery, against seven classical baselines and two zero-shot pretrained matchers, SCDF registers every pair with zero failures, reduces the best baseline's real-pair median end-point error from 6.83 to 4.17m, and cuts its 90th percentile from 17.8 to 7.77m.

cs.CV↗