arXiv subjects

Hong Zhang

Publications and source records attributed to Hong Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Breakdown of Aharonov-Bohm cage in Rydberg synthetic lattices: the roles of inhomogeneity and long-range exchange

While the interaction-induced breakdown of Aharonov-Bohm (AB) cage is typically attributed to uniform bound-pair transport, systems with inhomogeneous exchange interactions realized with Rydberg synthetic lattices exhibit more complex dynamics. Employing the evolution-path symmetry (EPS) framework developed recently, we analyze the two-particle dynamics via path interference in Fock space. We find that a homogeneous nearest-neighbor exchange interaction cannot break the AB cage, regardless of whether the long-range exchange interaction is present or not. In contrast, we demonstrate that inhomogeneous nearest-neighbor exchange interaction breaks the destructive-interference EPS, and lifts the degeneracy of many-body compact localized states, thereby generating non-local dispersive eigenstates. Consequently, the initial state gains a non-zero overlap with these dispersive states, enabling delocalized transport. Furthermore, while long-range exchange interaction alone preserves the AB cage, its coupling with nearest-neighbor inhomogeneous exchange interaction opens non-canceling pathways that alter the diffusion profile. Our work connects microscopic path interference with macroscopic spectral reorganization, offering an analytical understanding of the mechanism underlying exchange-interaction-induced transport in Rydberg synthetic lattices.

cond-mat.quant-gas↗

arXiv2026

VLAConf: Calibrated Task-Success Confidence for Vision-Language-Action Models

Task-success confidence estimation for Vision-Language-Action (VLA) models provides a crucial task-level signal for monitoring manipulation in open-world environments and supporting downstream decision-making. Existing methods typically construct task-success confidence from action-token probabilities. However, such probabilities are not naturally available in flow-matching policies, limiting their applicability to mainstream flow-matching VLAs. To address this issue, we propose VLAConf, a two-stage representation-level confidence framework that operates on frozen pretrained VLA representations. A step-conditioned Coin-Flip Network learns an uncalibrated inverse success-support score from successful demonstrations, while a low-capacity calibrator fitted on outcome-labeled successful and failed rollouts maps the aggregated score to task-success probability. Experimental results on the LIBERO benchmark demonstrate that VLAConf improves online task-success confidence estimation over alternative approaches. We further demonstrate its utility in selective expert assistance, where confidence-triggered handoffs improve task success over no intervention. Its applicability is also evaluated in real-robot experiments. To access the source code and supplementary videos, visit https://sites.google.com/view/vlaconf.

cs.RO↗

arXiv2026

A fully discrete LBRFD-IPDG method for linear fourth-order parabolic equations

We propose a fully discrete method for linear fourth-order parabolic equations with Dirichlet boundary conditions, combining an implicit LBRFD multistep scheme in time with a mixed interior penalty discontinuous Galerkin (IPDG) method in space. The temporal discretization employs equispaced linear barycentric rational interpolants and incorporates a startup procedure. To facilitate the spatial discretization, the original problem is reformulated through an auxiliary variable. For certain parameter pairs $(n,d)$, the LBRFD method is shown to be $A(α)$-stable and to possess a wider stability angle than the corresponding BDF$p$ method of the same order. Stability and a priori error estimates are established via a $G$-energy technique and the discrete Grönwall lemma. The theoretical analysis yields a total $L^2$ error estimate of order $h^{k-1}+τ^p$, where $p=d$ if $n-d$ is even and $p=d+1$ if $n-d$ is odd. The reduced spatial convergence rate is attributed to boundary contributions on $\partialΩ$. Despite this theoretical prediction, numerical experiments confirm the stability and demonstrate optimal convergence of order $h^{k+1}+τ^p$.

math.NA↗

arXiv2026

Matching Supervision to the Student's Learning Capacity: A Unified Framework for On-Policy Self-Distillation

On-policy self-distillation (OPSD) improves the reasoning abilities of LLMs by internalizing privileged context into model parameters through self-distillation. Two recent research lines promote vanilla OPSD by choosing which tokens to learn from and by controlling how much privileged information the teacher receives, respectively. However, we show that each line optimizes one variable while holding the other fixed, which leads to a suboptimal solution. We argue that the two variables are coupled through the student's learning capacity: the privileged information sets the per-token divergence the teacher prescribes, while token weighting selects which of these the student must absorb. We formalize the two lines of work into a unified optimization framework, which maximizes the aggregate teacher--student divergence, subject to a budget on the aggregate learning difficulty the student can absorb. Under this modelling, we propose Unified On-Policy Self-Distillation (USD), a lightweight online algorithm to solve the Lagrangian. USD reveals that a single dual variable governs both decisions: at one price for learning difficulty, it simultaneously sets the token-selection threshold and the direction of privileged-information adjustment, keeping supervision matched to the student's evolving capacity. Through extensive experiments, USD consistently demonstrates superior performance over OPSD and token- and PI-side baselines across various model scales on various reasoning benchmarks. Code is available at https://github.com/lauvlalala/USD.

cs.AI↗

arXiv2026

HAM-VLN: Harnessing Hierarchical Agentic Memory for Zero-Shot Vision-and-Language Navigation

Vision-and-language navigation (VLN) enables robots to follow instructions in previously unseen environments. Recently, a training-free paradigm has emerged: the robot queries a multimodal LLM to understand its observations and plan the next action. However, long-horizon navigation based on either image streams or dense map inevitably introduces a growing memory and reasoning bottleneck. We present HAM-VLN, a decision-coupled, agent-authored memory that equips the robot with a persistent, depth-grounded world graph. In the same model call used to select the next action, HAM-VLN also records semantic and reflective information---including room type, objects, navigation progress, and failure notes. Recent waypoints remain verbatim within a bounded window, while older history re-enters the context only through retrieval scored by relevance, recency, and salience, together with one-hop topological expansion. This design requires no additional LLM calls beyond the per-waypoint decision. Compared to previous methods, HAM-VLN not only improves various navigation metrics but also reduces the context length by more than 65%. Specifically, HAM-VLN achieves 61.0% Success Rate (SR) on VLN-CE R2R, 52.7% SR on VLN-CE RxR, and 79.7% SR on HM3D-v2 ObjectNav without any training.

cs.RO↗

arXiv2026

A Portable and Versatile Limited-Memory BFGS Implementation in PETSc/TAO

The limited-memory BFGS (L-BFGS) Hessian update scheme is the critical kernel in many quasi-Newton optimization algorithms. The most common approach to implementing L-BFGS uses $2m$ sequential rank-1 updates as part of solving a linear system when there are $m$ history steps. The performance of this approach suffers when the latency of synchronization is significant, and its poor temporal locality increases the memory traffic when vectors do not fit in cache. The compact dense representation of L-BFGS results in an approach that has minimal synchronization latency and better temporal locality, but it requires an additional pass over the basis vectors and an additional basis that must be recomputed when the $B_0$ matrix changes as in variable-metric methods. In the Portable Extensible Toolkit for Scientific Computation and the Toolkit for Advanced Optimization (PETSc/TAO), we have implemented an intermediate dense formulation of BFGS that retains most of the good characteristics of both the recursive and compact dense approaches. We report single-node performance tests of these implementations on the U.S. Department of Energy's Polaris and Frontier machines, testing both GPU-based and CPU-based computations.

cs.DC↗

arXiv2026

AnchorMark: Robust Diffusion Watermarking via Latent-Space Rotation Synchrony

Inversion-based watermarking embeds watermark payloads directly into the generative process, avoiding a separate post-hoc image-domain embedding stage while preserving the native visual fidelity of synthesized images. However, existing methods remain vulnerable to compound lossy post-processing, particularly when rotation is involved, as it disrupts the spatial correspondence required for latent-space decoding. To overcome this limitation, we introduce AnchorMark, a training-free, robust inversion-based watermarking. We uncover a latent-space property termed Rotation Synchrony: image-domain rotations and their counterparts in the recovered initial latent share the same angle. Building on this property, AnchorMark embeds a synchronization anchor in the central region of the initial latent, enabling accurate estimation and correction of the rotation angle during extraction. Experiments show that AnchorMark substantially improves bit accuracy under rotation and combined attacks, with limited impact on image quality.

cs.CR↗

arXiv2026

BlindPSNR: A No-Reference Fidelity Predictor for Low-Light Image Enhancement

Low-light image enhancement (LLIE) methods involve tunable parameters that are typically fixed, often leading to performance degradation when applied across scenes. Manually selecting the best configuration, however, can be time-consuming and not always practical. Peak signal-to-noise ratio (PSNR) is the natural fidelity criterion for automating parameter selection, yet it requires a ground-truth reference that is typically unavailable. To our knowledge, no learning-based method addresses no-reference PSNR prediction for low-light image enhancement; the natural surrogate, no-reference image quality assessment (NR-IQA), targets perceptual quality rather than signal fidelity, and all seven baselines we test achieve 0% top-1 selection accuracy on our benchmark. With paired training data, the ground-truth PSNR is analytically computable, providing exact supervision without a separate teacher network. Building on this, we propose BlindPSNR, a lightweight no-reference network that fuses the enhanced image with the degraded low-light input via windowed cross-attention and estimates PSNR through heteroscedastic regression. While a scalar-regression baseline achieves top-1 accuracy of 54.4%, BlindPSNR raises this to 89.5% with regret dropping from 1.62 dB to 0.026 dB, and generalizes to unseen datasets (SRCC = 0.61-0.67).

cs.CV↗

arXiv2026

Enhancing Glass Surface Reconstruction via Depth Prior for Robot Navigation

Indoor robot navigation is often compromised by glass surfaces, which severely corrupt depth sensor measurements. While foundation models like Depth Anything 3 provide excellent geometric priors, they lack an absolute metric scale. We propose a training-free framework that leverages depth foundation models as a structural prior, employing a robust local RANSAC-based alignment to fuse it with raw sensor depth. This naturally avoids contamination from erroneous glass measurements and recovers an accurate metric scale. Furthermore, we introduce \ti{GlassRecon}, a novel RGB-D dataset with geometrically derived ground truth for glass regions. Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art baselines, especially under severe sensor depth corruption. The dataset and related code will be released at https://github.com/jarvisyjw/GlassRecon.

cs.RO↗

arXiv2026

ReferTrack: Referring Then Tracking for Embodied Visual Tracking

Embodied visual tracking (EVT) requires a mobile agent to continuously follow a specific target described in natural language using only onboard vision. While recent vision-language-action (VLA) policies unify target identification and trajectory planning, their chain-of-thought (CoT) reasoning often operates in abstract spatial latents that are difficult to supervise and weakly aligned with explicit image-space detections. To address this, we introduce ReferTrack, a referring-then-tracking paradigm that grounds EVT using a single forward-facing camera. Our model first selects the target from an indexed set of bounding boxes, then decodes tracking waypoints conditioned on this image-grounded decision. To preserve target motion cues over time, ReferTrack maintains a sliding-window queue of previously selected bounding boxes, injecting their geometric features into the visual history via temporal-viewpoint-bbox indicator (TVBI) tokens. We further enhance target identification by co-training on a custom Refer-QA dataset. On EVT-Bench, ReferTrack achieves state-of-the-art single-view performance with success rates of 89.4%, 73.3%, and 74.1% on the single-target, distracted, and ambiguity tracking splits, respectively -- matching or even surpassing several multi-camera baselines on identification-heavy tasks. Finally, real-world deployments on legged and humanoid robots validate its robust sim-to-real transfer capabilities. Code is available at https://github.com/MedlarTea/referTrack.

cs.RO↗

arXiv2026

Quasi-bound states and late-time evolution of a massive fermion around a Reissner-Nordström black hole

A massive fermion around a charged black hole provides a gravitational analogue of atomic bound states and their relaxation. In this work, we study this system by formulating the radial equation as a coupled matrix system and constructing the Green's function with ingoing boundary conditions at the horizon and decaying boundary conditions at infinity. In the weak-coupling scenario $|qQ|\sim mM<1$, a matrix matching scheme gives an improved analytic expression of quasi-bound-state spectrum, including fine-structure corrections and more accurate decay widths. The extremal Reissner-Nordström case ($|Q|=M$) is treated separately and shown to be the smooth limiting result of the non-extremal spectrum. We further analyze the branch-cut contribution to the time-domain Green's function in the late-time limit. We confirm an oscillatory power-law behavior in intermediate late-time regime $1/m < t < 1/m^3M^2$. In the far late-time regime $t>1/m^3M^2$, the activation of the quasi-bound states produces an $t^{-5/6}\exp(-ηt^{1/3})$ suppression with a chirping phase before the asymptotic $t^{-5/6}$ tail previously found in the limit $t\to\infty$. Direct time-domain simulations support this distinction and show how the quasi-bound contribution coexists with the familiar power-law component.

gr-qc↗

arXiv2026

TRIDS: AI-native molecular docking framework for accelerating high-throughput virtual screening with physically valid binding poses

Molecular docking is a cornerstone of drug discovery for unveiling the mechanism of ligand-receptor interactions. With the recent advances of deep learning (DL), AI-powered molecular docking methods have achieved higher accuracy for binding pose prediction and virtual screening compared with classical physics-based methods. However, there is still a scarcity of approaches to strike a balance among accuracy, computational efficiency, and rigorous physical validity of the output conformations. In the previous two versions of DSDP, we demonstrated the effectiveness of guiding conformation sampling with the gradient of an analytic scoring function. As the third version, TRIDS was devised as an AI-native docking framework that expand the similar strategy to unify conformation sampling and docking processes with DL-based model for improving accuracy of docking and screening. Furthermore, it is tailored for seamless cooperation of AI and physics to guarantee the physical validity of predicted binding poses. Being user-friendly, TRIDS predicts the binding site, parses multiple file formats, and supports Python programming and PyMOL graphical interaction. It improves docking accuracy and passes physical validation with high computational efficiency, i.e. a single docking task is done in a fraction of second while maintaining a highly lightweight GPU memory footprint of merely hundreds of megabytes, facilitating high-throughput virtual screening in reality. As a proof of concept, TRIDS allowed us to obtain hit compounds with novel scaffolds for tumor necrosis factor-alpha (TNFα) inhibitor through a large-scale virtual screening.

physics.chem-ph↗

arXiv2026

Can Single-View Mesh Reconstruction Generalize to Robot Camera Rotation?

Single-view mesh reconstruction predicts object meshes and spatial layouts from a single observation, making it attractive for fast robot spatial reasoning and real-to-sim digital twins. However, robot-mounted cameras naturally rotate during manipulation and navigation, while learned single-view reconstruction models often rely on view-dependent priors and may generalize poorly to out-of-distribution camera rotations. Such rotations can introduce 3D inconsistencies, incorrect layouts, and violations of physical constraints, but this failure mode remains under-evaluated. We introduce an evaluation protocol with controlled axis-wise roll, pitch, and yaw sweeps to trace errors in monocular depth estimation (MDE), canonical object meshes, camera-space layout, and physical plausibility within a representative SAM3D-style pipeline. On the Aria Digital Twin dataset and a real Franka wrist-camera sequence, camera rotations induce MDE distortion, layout drift, and collision penetration, while canonical mesh predictions remain relatively stable. A two-stage SAM3D+FoundationPose pipeline is more robust than one-stage feed-forward layout prediction, and our Gravity-Aware Refinement reduces one-stage pairwise ICP-based layout-orientation error by 47.1$\%$. Our evaluation reveals that current single-view mesh reconstruction methods generalize poorly to robot camera rotation, and suggests that explicit gravity cues are important for reliable robotic single-view mesh reconstruction.

cs.CV↗

arXiv2026

Anisotropic 2D FUP and quantum open baker's map

We prove an essential spectral gap for 2D anisotropic quantum open baker's map. This extends the 1D results of Dyatlov--Jin 2017 and the isotropic 2D results of Cohen 2025a. The key ingredient is the anisotropic discrete fractal uncertainty principle (FUP) associated with a 2D anisotropic fractal set called the Bedford--McMullen carpet. We also study the relation between our anisotropic discrete FUP and its continuous counterpart in the spirit of Dyatlov--Jin 2018 and Cohen 2025a. In particular, we prove {continuous FUP} for 2D {anisotropic porous} sets, extending the (high-dimensional) isotropic results of Cohen 2025b. To the best of our knowledge, the anisotropic (line) porosity condition -- a variant of Cohen's line porosity and stronger than ball porosity -- appears to be new to the literature.

math.CA↗

arXiv2026

FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction

Recent indoor occupancy prediction methods adopt Gaussian primitives as a sparse 3D representation for computational efficiency. However, their training relies on voxel classification, which imposes only local constraints and lacks global supervision on the distribution of the primitives. Therefore, they inevitably predict spurious primitives in empty regions, undermining both representational and computational efficiency. To address this, we propose Feed-forward Likelihood Maximization (FLM), a novel framework that reformulates occupancy prediction as voxel distribution estimation. In FLM, a network is trained to predict a mixture model that maximizes the likelihood over ground-truth occupied voxels in a feed-forward manner. To enable end-to-end training of networks and voxelization of a standard mixture model, we define mixture weights as normalized primitive volumes to implicitly enforce simplex constraints and derive novel voxelization formulas. Based on FLM, our FLM-Occ, a novel method that is capable of relocating randomly initialized primitives over long distances to model a scene. On Occ-ScanNet, FLM-Occ achieves superior accuracy using only 32 superquadrics, 2.7% of the prior SoTA, while running 3.7 times faster.

cs.CV↗

arXiv2026

GCNGrasp-VP: Affordance-Guided View Planning for Efficient Task-Oriented Grasping

Task-oriented grasping performance degrades significantly when object views suffer from occlusions. Existing task-oriented grasping methods typically assume task-relevant regions are visible in the initial frame, while view planning approaches enable active perception but often ignore task semantics and rely on time-consuming scene reconstruction. To address these limitations, we present GCNGrasp-VP, an efficient framework integrating affordance field prediction with active view planning. Central to this framework is GCNGrasp-v2, a task-oriented grasp model that simultaneously supports grasp evaluation and affordance field prediction, achieving constant-time inference complexity. Leveraging this capability, our Affordance-guided View Planner (Affordance-VP) utilizes the affordance field as an information gain metric to guide camera observation of task-relevant regions without requiring scene reconstruction. View planning results show that our method significantly outperforms scene-uncertainty-driven baselines with only one view adjustment. Real-world validation further confirms substantial improvements in grasp success rates for single-object scenarios while maintaining millisecond-level computational latency. Code and models are available at https://github.com/Instinct323/GCNGrasp-VP.

cs.RO↗

arXiv2026

Scene-Adaptive Nonlinear Tone Curves for Pseudo Ground-Truth Generation in Low-Light 3D Gaussian Splatting

Low-light novel view synthesis is challenging because dark multi-view images contain noise, weak structural detail, and compressed dynamic range. Recent 3D Gaussian Splatting (3DGS) methods address these challenges by generating pseudo ground-truth (pseudo-GT) images as supervision targets when paired normal-light references are unavailable. Existing pseudo-GT methods apply a uniform linear gain to all pixels, which clips bright regions while providing insufficient enhancement in dark regions, limiting reconstruction quality. We observe that nonlinear tone mappings, long established in 2D low-light enhancement, have not been explored for pseudo-GT generation in 3D reconstruction. Accordingly, we propose a scene-adaptive nonlinear tone-curve framework that replaces linear pseudo-GT with nonlinear alternatives. The framework introduces percentile-based normalisation for scene-agnostic curve application, a scene-adaptive offset for automatic black-level adjustment, and two complementary curves: Adaptive SoftExp (ASE), a bounded exponential curve, and Adaptive Poly3 (AP3), a data-driven cubic polynomial. The module changes only the pseudo-GT computation and leaves the 3DGS backbone unchanged. Experiments on three benchmarks covering 21 scenes show that both curves consistently outperform the linear baseline with PSNR improvements up to +4.34 dB on LOM and +3.25 dB on RealX3D. Both curves achieve similar performance despite their different mathematical forms, suggesting the improvement is curve-agnostic. Code is available at https://github.com/lvmingzhe/adaptiveToneCurve

cs.CV↗

arXiv2026

Can Aerial VLA Models Cooperate? Evaluating Closed-Loop Air-Ground Coordination with CARLA-Air

Recent aerial vision-language-action (VLA) models show promising single-UAV capabilities, such as tracking moving objects and navigating to language-specified landmarks. However, it remains unclear whether these capabilities can transfer to air-ground cooperation, where a UAV and a UGV must act jointly in a shared, closed-loop physical world. We study this question with CARLA-Air, a single-process air-ground evaluation environment that unifies CARLA and AirSim inside one Unreal Engine runtime. By sharing the same world state, physics tick, and sensing pipeline, CARLA-Air enables physically consistent UAV--UGV interaction and precise measurement of simulation-timestamp alignment and effective coordination latency. Using CARLA-Air, we evaluate representative aerial VLA and planning baselines on two complementary diagnostic tasks: moving-platform landing and occlusion-recovery escort. The results show that current aerial VLA models can often track or follow a ground partner, but struggle to convert this single-agent competence into stable cooperative behavior. State prompting provides limited benefit, and naive bidirectional interaction fails to consistently improve performance and can amplify errors for most baselines. These findings suggest that, under the tested text-based cue interfaces, zero-shot cooperative air-ground VLA requires three components beyond the current paradigm: explicit partner-state grounding, low-latency action coordination, and team-level objective alignment. Our code is available at https://github.com/louiszengCN/CarlaAir.

cs.RO↗