Search arXiv⌕ Search

arXiv subjects

Peng Wang

Publications and source records attributed to Peng Wang.

At least 55 records · Page 3Linked to original sources
arXiv2026

Rheotaxis of a Microswimmer in Poiseuille Flow

We investigate the motion of a spherical microswimmer in a planar Poiseuille flow using analytical calculations and numerical simulations. Our results show that the microswimmer's orientation undergoes periodic oscillations governed by pendulum-like dynamics. The oscillation period is proportional to the channel width and to the complete elliptic integral of the first kind evaluated at the maximum orientation angle, and inversely proportional to the square root of the product of the maximum flow speed and the self-propulsion speed. Based on the net displacement along the flow direction within one oscillation period and the signs of the maximum and minimum velocities, we identify five motion states: upstream, oscillatory upstream, zero-drift oscillatory, oscillatory downstream, and downstream motion. The net displacement is jointly determined by the ratio of the self-propulsion speed to the maximum flow speed and the complete elliptic integrals of the first and second kinds. The signs of the maximum and minimum velocities are determined by the speed ratio and the cosine of the orientation angle. Flow nonuniformity promotes upstream migration, whereas inertial lift produces negligible radial displacement during one oscillation period. These findings identify the key parameters governing the motion of spherical microswimmers and provide significant implications for understanding microorganism motility and designing microrobots with prescribed motion states.

physics.flu-dyn↗
arXiv2026

Beyond Attention Imbalance: Mitigating Hallucinations via Spectral Surgery

While Large Vision-Language Models (LVLMs) achieve remarkable success, hallucinations remain a significant barrier to their reliable deployment. Recent studies primarily attribute these issues to cross-modal attention imbalances; most solutions therefore focus on reweighting visual tokens or suppressing language priors. However, such approaches often overlook the spectral characteristics of the visual information flow and frequently rely on Contrastive Decoding (CD), which doubles inference time. Instead of following conventional approaches, we identify two distinct hallucination patterns-Perceptual-Semantic Dissociation and Localized Fixation-and propose FLASH (Frequency-Localized Attention SHaping), a training-free and CD-free framework. FLASH utilizes a Spectral Vortex Score to detect vision heads within multi-head attention layers and applies adaptive spectral modulation to rectify the visual information flow during decoding. Empirical results demonstrate that FLASH achieves a superior balance between performance and efficiency compared to SOTA methods.

cs.CV↗
arXiv2026

UniBYD: A Unified Framework for Learning Robotic Manipulation Across Embodiments Beyond Imitation of Human Demonstrations

In embodied intelligence, the embodiment gap between robotic and human hands brings significant challenges for learning from human demonstrations. Although some studies have attempted to bridge this gap using reinforcement learning, they remain confined to merely reproducing human manipulation, resulting in limited task performance. Moreover, current methods struggle to support diverse robotic hand configurations. In this paper, we propose UniBYD, a unified framework that uses a dynamic reinforcement learning algorithm to discover manipulation policies aligned with the robot's physical characteristics. To enable consistent modeling across diverse robotic hand morphologies, UniBYD incorporates a unified morphological representation (UMR). Building on UMR, we design a dynamic PPO with an annealed reward schedule, enabling reinforcement learning to transition from offline-informed imitation of human demonstrations to online-adaptive exploration of policies better adapted to diverse robotic morphologies, thereby going beyond mere imitation of human hands. To address the severe state drift caused by the incapacity of early-stage policies, we design a hybrid Markov-based shadow engine that provides fine-grained guidance to anchor the imitation within the expert's manifold. To evaluate UniBYD, we propose UniManip, the first benchmark for cross-embodiment manipulation spanning diverse robotic morphologies. Experiments demonstrate a 44.08% average improvement in success rate over the current state-of-the-art. Our project page is https://zhanheng-creator.github.io/UniBYD.

cs.RO↗
arXiv2026

P4Q: Co-designing Token Pruning and Quantization for Vision-Language Model Acceleration

Vision language models have achieved strong performance across a wide range of multimodal applications, yet their substantial computational and memory costs hinder efficient deployment. Visual token pruning and post-training quantization reduce inference overhead along two complementary dimensions, namely sequence length and numerical precision. Existing workflows typically optimize these techniques independently or apply them sequentially. Their distinct optimization objectives leave critical interactions unaddressed and constrain the achievable compression performance. We revisit these designs and present P4Q, a practical co-design framework that jointly optimizes visual token pruning and low-bit quantization for efficient VLM inference. First, P4Q introduces a quantization-aware visual token selection strategy before the LLM. It applies fake quantization to copies of the features produced by the projector and selects visual tokens using statistics computed from these fake-quantized features, thereby conditioning the selector's feature-based decisions on simulated low-bit perturbations. Second, P4Q introduces a pruning-aware quantization calibration strategy. It uses the same selection strategy as pruning to calibrate the quantized model on the retained-token distribution, thereby aligning the calibration process with the pruned execution path used during deployment. By coupling these two components, P4Q achieves substantial inference speedups while maintaining comparable task performance, resulting in a better efficiency-accuracy trade-off than independently optimized pipelines. For instance, on LLaVA-NeXT, P4Q achieves an average end-to-end inference speedup of 2.8x across eight distinct test sets, while retaining higher accuracy than prior compression and quantization methods.

cs.CV↗
arXiv2026

SubRot: Signed Gradient Subspace Calibration for VLM Rotation Quantization

Post-training quantization reduces the deployment cost of vision-language models (VLMs), but preserving multimodal capabilities at low bit widths remains challenging. Existing methods rely on modality- or token-level gradient statistics, which are susceptible to cross-sample variations in visual-to-textual token ratios and the positions of visual information, limiting statistical stability. Moreover, overly coarse aggregation through absolute values and averaging discards gradient signs and channel-wise differences, limiting the separation of modality-specific sensitivities. In contrast, the channel space provides a shared coordinate system across samples, making it a more natural basis for capturing stable task-sensitive structures. We therefore propose SubRot, a signed gradient subspace calibration method for VLM rotation quantization. Through eigendecomposition of the empirical Fisher matrix of activation gradients, SubRot identifies a sensitive channel subspace with three properties: cross-sample stability, clear sensitivity separation, and consistent signed effects on the autoregressive loss along certain directions. Guided by a local Taylor expansion, SubRot combines signed first-order guidance along sign-stable directions with second-order constraints along the remaining sensitive directions, while retaining MSE for overall reconstruction quality. This objective steers quantization errors toward loss-decreasing directions while controlling their magnitude. Experiments on five VLMs across five benchmarks show consistent average-score improvements over FlatQuant under W4A6 and W4A4, reaching 1.4 percentage points on LLaVA-NeXT-7B. Under W4A4, average accuracy degradation from FP16 remains within 1.4 percentage points across all evaluated models, while LLaVA-v1.5-13B exceeds its FP16 average score by 0.4 percentage points.

cs.CV↗
arXiv2026

Classical simulation of amplitude-damped bucket-brigade quantum random access memories via predictable branch evolution

Classical simulation of quantum random access memory (QRAM) under noise can be accelerated dramatically by branch pruning: noise histories are sampled as trajectory ensembles, and good branches, whose routing paths avoid every sampled fault, need not be evolved because their final states are analytically predictable. Under amplitude damping this predictability is not automatic, because the no-jump operator $K_0$ acts on every branch at every time slice. Here we give a complete account of damping-channel predictability in bucket-brigade QRAM simulation, for both qutrit and qubit encodings. For the qutrit encoding, $K_0$ is diagonal in the node basis, and the wait state $|W\rangle$ is its fixed point, so a good branch follows the noiseless orbit up to a scalar attenuation. In the qubit encoding, phase-kickback memory fetching between two Hadamard walls makes $K_0$ no longer diagonal, yet the resulting structure is exactly solvable in closed form. Conditional on no jump, the in-branch infidelity of a good branch is second order in the damping rate $γ$, with coherent leakage $\sim n^2γ^2/4$ per data qubit, and the closed form captures it without approximation. We also revise the pruning criterion for the qubit encoding. The resulting algorithm evolves only the bad branches plus one reference branch and shares every other cost with the full evolution it replaces. It is validated by end-to-end benchmarks with pruned-over-full speedups up to $285\times$ and by trajectory-level tests confirming every closed form to machine precision.

quant-ph↗
arXiv2026

Classical Bound on the Fisher Information Rate of a Dephasing-Enhanced Quantum Neuron

Dephasing can sharpen the activation of a dissipative quantum neuron, but a stationary response does not determine how much input information its output delivers per unit time. For a single qubit with partial-reset updates at rate $ν$, we derive the threshold response bandwidth and the exact local Fisher information rate of the complete record from repeated projective readouts, including their backaction. Near the activation threshold, the rate obeys $\mathcal R\leqν(1-s)(1+s)^2/2\leq16ν/27$, where $s$ is the retained amplitude of each update. For every fixed finite Markovian dephasing rate, optimization over the readout interval and $s$ gives the same supremum, approached at $s=1/3$ with arbitrarily rapid ideal readout. A finite measurement dead time instead selects a finite optimal interval, which we obtain by a dimensionless two-parameter optimization. The ideal bound is saturated by a classical two-state jump process: quantum coherence changes finite-cadence performance but cannot raise the optimized bound. These results separate noise-enhanced activation from the operational information throughput of a quantum neuron and provide a hardware-aware benchmark for finite-time quantum neural processing.

quant-ph↗
arXiv2026

WorldAgent: Verification-Guided Agentic Physical World Construction

Constructing complex physical worlds from language requires coordinating extensive 3D environments, detailed structures and objects at different spatial scales, and interacting physical processes under both stated goals and implicit physical constraints. We present WorldAgent, an agentic framework for verification-guided physical world construction from a single natural-language prompt, without iterative user debugging. A world construction layer expands the prompt into a structured world specification and uses physical knowledge to build scenes and run numerical simulations. After every step, a verification layer inspects scene geometry and simulation states alongside rendered views. Failed checks guide automatic revisions to the specification and re-execution of the affected steps. Accepted worlds pass the required checks and remain editable for further inspection and resimulation. We introduce AgenticSimBench, on which WorldAgent achieves the best scores among the evaluated agent-based methods on five of seven metrics. In a 26-participant user study, it receives the highest mean ratings across all four criteria.

cs.AI↗
arXiv2026

Self-constrained Magnetic Reconnection of a Solar Flare Co-spatial with a Non-ejected Filament

For flares associated with eruptive filaments, the filament normally rises above the post-flare loops. Here, we report an unusual event in which the post-flare loops instead overlie a filament that remains non-ejected throughout the flare. This event was simultaneously captured by the Chinese H$α$ Solar Explorer (CHASE), the Solar Extreme-UltraViolet Imager (SUVI) onboard Fengyun-4C, the Hard X-ray Imager (HXI) onboard the Advanced Space-based Solar Observatory (ASO-S) and the Solar Dynamics Observatory (SDO). Extreme Ultraviolet (EUV) observations provided by FY4C/SUVI and SDO show that two hot loop systems form at the flare onset. The western end of the northern loop system undergoes a southwestward drifting motion with gradually increasing shear, whereas the southern loop system progressively converges toward and eventually overlies the northern one. Meanwhile, three-dimensional magnetic field extrapolations reveal an increase in both the twist of the filament-supporting magnetic structure and its poloidal magnetic field after the flare. Taken together, these observations suggest that the flare is driven by a self-constraining zipper-type magnetic reconnection, in which highly sheared arcades reconnect to form a series of overlying flare arcades and an underlying flux rope. This reconnection continuously strengthens the magnetic confinement of the filament, preventing its outward eruption, while facilitating substantial plasma drainage along the reconnected magnetic field revealed by the CHASE H$α$ spectroscopic observations.

astro-ph.SR↗
arXiv2026

CFCH: Coarse-Fine Collaborative Hierarchical Learning for Anterior Segment Disease Analysis

Accurate classification of anterior segment diseases is crucial for ophthalmic screening and diagnosis. However, slit-lamp image analysis remains challenging due to substantial variability in imaging conditions and the intrinsic anatomical-disease hierarchy of ocular pathologies. Existing methods typically formulate this task as a flat multi-class classification problem, ignoring the structured dependency between anatomical regions (e.g., cornea, conjunctiva, and lens) and disease manifestations.To address these limitations, we propose CFCH, a Coarse-Fine Collaborative Hierarchical learning framework that explicitly models anatomical context and disease semantics through a dual-branch architecture. To enable effective cross-granularity collaboration, CFCH introduces semantic and cross-granularity attention consistency constraints, encouraging aligned yet complementary feature learning across branches. In addition, we construct AS-9K, a large-scale anterior segment dataset with 8975 images covering 12 common disease categories. To the best of our knowledge, AS-9K is the largest publicly available dataset for anterior segment image classification. Extensive experiments on two anterior segment datasets demonstrate that CFCH outperforms state-of-the-art methods. Qualitative visualizations further show more focused and lesion-relevant activation responses, validating the effectiveness of the proposed framework. Code will be available at https://github.com/ybupengwang/CFCH.

cs.CV↗
arXiv2026

CORDIAL: Calibrating Ordinal LLM Outputs from Few Labels

A large language model (LLM) can turn a text into a distribution over an ordered scale, but that distribution is a noisy measurement: saturated, compressed or exaggerated, and biased in a consistent direction. We propose CORDIAL, which treats the model's output as a noisy reading of the true label and corrects it with a channel of five interpretable parameters. The channel is small enough for its posterior to be averaged from a handful of labels, and we prove that the resulting calibration preserves first-order stochastic order. On Amazon reviews and CMU-MOSEI transcripts with four LLMs, CORDIAL has the lowest log loss among nine calibrators in 76 of 80 settings with 5 to 100 labels; with 20 labels and the main 7B reader, it matches the strongest baseline using 28-54 labels. The same posterior lets us learn priors from other tasks and fuse several LLMs. Unrestricted calibrators such as Dirichlet calibration overtake it only as the calibration set grows into the hundreds or thousands.

cs.CL↗
arXiv2026

Multi-tracer mass bias in matched cosmic voids from SDSS DR7 and the ELUCID constrained simulation

Cosmic voids provide a unique environment for studying the relationship between galaxies, subhaloes, and dark matter in the underdense Universe. Using the SDSS galaxy catalogue and the ELUCID constrained simulation, we establish an observationally anchored framework for measuring multi-tracer mass bias within matched cosmic voids. A sample of 102 matched void pairs is constructed to directly compare galaxy, subhalo, and dark matter mass distributions within an observationally constrained realisation of the local Universe. We find that both the galaxy-to-dark matter and subhalo-to-dark matter mass ratios decrease toward void centres, indicating that luminous and halo tracers become increasingly depleted relative to the underlying matter distribution in the deepest underdensities. In contrast, the galaxy-to-subhalo mass ratio exhibits substantially larger statistical uncertainties within the inner void regions ($r/R_{\rm v}\lesssim0.5$). By comparing measurements obtained using independent and common coordinate frameworks, we show that coordinate offsets contribute to the observed scatter but cannot fully account for the large uncertainties. The remaining uncertainty primarily arises from the severe scarcity of massive subhaloes ($\log_{10}(M_{\rm sub}/h^{-1}M_\odot)\ge11.8$) within void interiors, which greatly reduces the number of statistically valid measurements near void centres. Our results provide an empirical characterization of multi-tracer mass bias in observationally constrained cosmic environments and demonstrate the importance of accounting for coordinate consistency and tracer scarcity when interpreting such measurements in extreme underdense regions.

astro-ph.CO↗
arXiv2026

Precision and resource scaling of real-time flux distortion compensation for superconducting quantum control

Real-time waveform generation supports dynamic quantum circuits without pre-storing complete waveforms for every execution path. However, long-lived distortions in flux-control lines degrade gate fidelity, requiring compensation to account for the actual pulse history. A frequency-domain inversion and time-domain fitting method is proposed for resource-efficient real-time flux distortion compensation. The method fits the reconstructed compensation impulse response with a compact hybrid infinite impulse response (IIR) and finite impulse response (FIR) filter. Look-ahead parallelization enables this filter to process synthesized waveforms at 1.2GSa/s on a field-programmable gate array (FPGA). Two-qubit cross-entropy benchmarking shows that real-time IIR filtering achieves a median controlled-Z Pauli fidelity close to the software-reference value of 99.57%. Numerical analysis and FPGA synthesis indicate approximately logarithmic growth in hardware resource use with compensation timescale. Extending compensation from microsecond to hundred-microsecond timescales increases look-up table (LUT) and digital signal processing (DSP) resource use by only about 14% and 4%, respectively, while maintaining a relative arithmetic error below $10^{-4}$. This work provides a scalable hardware foundation for high-fidelity flux control in dynamic superconducting quantum circuits.

quant-ph↗
arXiv2026

A proof of the Willmore-type conjecture in $\mathbb{C}P^2$

In 2002, Montiel and Urbano conjectured that the Clifford torus in $\mathbb{C}P^2$ minimizes the Willmore-type functional $ \mathcal{W}^-=\int_{T^2}(2+|H|^2)\,dA,$ either among all tori or among all Lagrangian tori. In this paper, we confirm this conjecture in the Lagrangian setting and disprove it in the general setting. We establish that every oriented closed Lagrangian surface of genus $g\geq 1$ in $\mathbb{C}P^2$ has $\mathcal{W}^-$-energy no less than that of the Clifford torus. Moreover, we construct non-Lagrangian deformations of the Clifford torus along which $\mathcal{W}^-$ strictly decreases.

math.DG↗
arXiv2026

LEAP-NBV: Lightweight Edge Active-Perception for Foundation-Model Next-Best-View Planning

Foundation models are endowing autonomous systems with greater intelligence, enabling a more comprehensive understanding of the environment through visual perception. A representative example is Human Mesh Recovery (HMR), which provides useful estimates of a target's 3D pose and shape that can benefit tactical missions. However, the size and power demands of such models make them difficult to run on edge platforms and limit their real-time performance, undermining the requirements of tactical edge deployment - especially for active perception, where a mobile robot must plan its next-best view on-board and cannot offload computation under contested communications. We present LEAP-NBV, a lightweight active-perception framework that runs foundation-model-driven Next-Best-View (NBV) planning on-board an edge device. To this end, we distill a family of large HMR teachers, each into a compact 32M student, with an offline mesh objective, then quantize the vision encoder to FP16 and characterize its on-device accuracy and latency. Within an occlusion-aware active perception loop, we evaluate all configurations on the same held-out benchmark and deploy the end-to-end pipeline on an NVIDIA Jetson Xavier NX, reporting measured on-device latency and energy. Distillation recovers 6-7 mm of Procrustes-aligned mean per-vertex position error (PA-MPVPE) over the undistilled student on the test set. Selecting the edge-optimal compression model brings the HMR engine to ~12 ms at a small accuracy cost and runs the full closed loop at 3.6 FPS and 2.6 J per frame, achieving a 2.0x speedup and 3.0x lower energy than the uncompressed model while nearly matching downstream task quality.

cs.AI↗
arXiv2026

Robotic Multiphase Interaction: Manipulating Coupled Liquid and Solid Dynamics with a World Model

This work presents \textit{Robotic Multiphase Interaction (RMI)}, a setting in which liquid enters a porous material and interacts mechanically with its deforming solid skeleton. Manipulation can therefore change pore volume, expel or redistribute retained liquid, and alter grasp stability at the same time. Spilled liquid can also create safety risks in domestic and manufacturing settings. This differs from most manipulation of solid objects and from tasks that involve both liquid and solid while keeping the phases spatially separate. We study a sponge filled with water as the first RMI example. We use implicit incompressible porous flow with smoothed particle hydrodynamics as the dynamics engine and enable robotic manipulation by adding Coulomb contact memory, hybrid velocity and force regulation, and a stability gate for lifting. The resulting environment connects robot commands to changes in the coupled liquid and solid state. A world model conditioned on actions predicts how this state evolves under candidate commands, while a temporal UNet generates actions using either Diffusion Policy or rectified flow matching. Our world model reduces retained water prediction error by more than $60\%$ compared with the baseline. The best action sequence selected by the world model from policy proposals further reduces the predicted terminal water error by about half. These improvements show that modelling the coupled liquid and solid state helps the robot predict how its actions affect both the porous object and the liquid held inside.

cs.RO↗
arXiv2026

Dephasing-Enhanced Response and Phase Transitions in Quantum Boltzmann Samplers

Dephasing can drive a quantum Boltzmann sampler into an ordered phase while its programmed probability distribution remains disordered. We trace this transition to an enhanced conditional response: coherence loss changes how each spin responds to its neighbors, and interactions amplify the change. For all-to-all coupling, we establish the stationary transition at finite dephasing. Near criticality, the magnetization distribution depends on the ratio of dephasing strength to the square root of system size, retaining a finite imprint of coherence even as the local coherent correction vanishes. In the strong-dephasing limit on a square lattice, the dynamics generates effective multispin interactions and probability currents, with Ising-like critical statistics. Adjusting the local dissipative update restores the target probability process on arbitrary graphs.

quant-ph↗
arXiv2026

The isoperimetric inequality and CMC hypersurfaces in Cartan-Hadamard manifolds

We show that the sharp Euclidean isoperimetric inequality holds for domains in complete simply connected Riemannian $n$-manifolds of nonpositive sectional curvature, $3\leq n\leq9$, which establishes the Cartan-Hadamard conjecture in these dimensions. The main step is a sharp inequality for constant-mean-curvature hypersurfaces, proved via integrals over pairs of boundary points, together with estimates for Jacobi fields along geodesic chords of the boundary. The weights in these integrals depend on the length of the chord and its angles with the boundary, and are chosen dimension by dimension, with a Green function pole. The inequality persists for boundaries of isoperimetric regions trapped in geodesic balls, whose mean curvature is constant only on the free part and which may have singularities. The isoperimetric-profile argument of Kleiner and Ghomi-Spruck completes the proof.

math.DG↗