arXiv subjects

Shuo Zhang

Publications and source records attributed to Shuo Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Shallow to Deep: Aligning Token Pruning with Stage-wise Roles in LVLMs

Large Vision-Language Models (LVLMs) incur high computational costs from redundant visual tokens. Although training-free attention-based multi-layer pruning in the vision encoder stage has been explored as an effective strategy, we find that pruning in shallow layers consistently degrades performance. In this paper, we aim to understand this problem and seek a solution. By analyzing attention patterns across network depth, we find that shallow layers primarily function as edge detectors with chaotic attention maps, while deeper layers transition through local subject recognition and unstable semantic aggregation. To address the misalignment between pruning strategies and network stages, we propose STD, a hierarchical token pruning framework that adapts token selection mechanisms to the functional role of each network stage. STD employs High-Frequency Spectral Analysis in shallow layers to deterministically preserve structural edges, uses Gaussian-Smoothed Attention in intermediate layers to maintain spatial coherence, and introduces a Stability-Adaptive Trigger in deep layers to execute pruning only during semantically stable phases. Extensive experiments show that STD outperforms state-of-the-art pruning methods by 1.1% on LLaVA-1.5-7B with 88.9% token reduction, while also being plug-and-play and highly effective when combined with other methods, and by 2.1% on LLaVA-NeXT-7B with 94.4% reduction, delivering a 3.9x speed-up in the prefilling stage. Our code will be released at https://github.com/Twilight03/STD.

cs.CV↗

arXiv2026

Global flows for the cubic nonlinear Schrödinger equation without decay assumptions

For each positive integer $N$, we construct global spectral group actions for the defocusing and focusing nonlinear Schrödinger hierarchies on initial-data classes containing $W^{2N+1,\infty}(\mathbb R)$. For $N\ge2$, the quadratic flow gives global classical cubic NLS solutions. The construction extends the Sato--Segal--Wilson and Kotani frameworks to matrix Dirac systems; a homogeneous tau-function identity supplies the focusing positivity. No decay, smallness, periodicity, arithmetic condition, or prescribed spectral background is required for the initial-data inclusion. As a principal application, we prove global well-posedness on $M_{\infty,1}^{5}(\mathbb R)$ for both signs, with finite-time bounds and locally Lipschitz dependence on initial-data norm balls. The passage to this Banach-space flow uses near-real-axis Weyl asymptotics and continuity of normalized Toeplitz operators on compact initial-data sets to obtain uniform spatial bounds from five bounded derivatives. In the defocusing case, the flow extends to $M_{\infty,1}^{5}(\mathbb R)+H^1(\mathbb R)$. Further consequences include quantitative approximation, preservation of spatial Bohr almost periodicity and its frequency module, conjugacy of spatial translation hulls, and preservation of the full complex Dirac spectrum and spatial transfer growth rates.

math.AP↗

arXiv2026

G-ray: Ray-Level Relative Geometric Position Encoding in Multi-View Vision Transformers under Camera Heterogeneity

We study relative position encoding for multi-view vision Transformers under camera heterogeneity, including varying fields of view (FoVs) or projection models. Existing rotary relative position encodings commonly use image-plane positional coordinates, producing projection-dependent relative phases and inconsistent geometric cues for cross-projection attention. We introduce G-ray, a ray-level relative position encoding whose rotary phases are parameterized by camera-local ray angles. The same camera-local ray pair induces the same relative phase across projections, providing projection-invariant positional consistency. G-ray can be used directly or integrated with existing encodings, retaining complementary geometric cues without additional learned parameters. We validate G-ray in three host encodings, RoPE, GTA, and RayRoPE, across 3D reconstruction and novel-view synthesis (NVS). Across three heterogeneous 3D reconstruction benchmarks at 50 views, G-ray leads all six averaged metrics and reduces mean pointmap relative error by 45.8% over MapAnything, with calibration supplied to both. Trained exclusively on homogeneous pinhole images, the 3D reconstruction model handles mixed pinhole and non-pinhole inputs without retraining and remains competitive on homogeneous pinhole 3D reconstruction protocols. For NVS, GTA and RayRoPE improve with G-ray under joint viewpoint and FoV variation. The project's webpage is available at https://g-ray-project.github.io/.

cs.CV↗

arXiv2026

GPCR Ligand Bioactivity Prediction with Physics-Informed Dual-State Query Learning

Predicting the bioactivity profiles of small molecules against G protein-coupled receptors (GPCRs) is a challenge in drug discovery. Although deep learning has accelerated the prediction of binding affinities, existing approaches often struggle to distinguish between functional efficacies because they neglect dynamic conformational equilibria. Furthermore, structure-based methods are frequently limited by the scarcity of high-resolution active-state crystal structures and the indistinguishability of conformational states in static representations. To bridge the gap between black-box prediction and biophysical reality, we propose Dual-State Query (DSQ), a physics-informed multimodal architecture that explicitly embeds the Monod-Wyman-Changeux (MWC) model of allostery within a neural network. Unlike conventional models that rely on explicit 3D structures, DSQ utilizes learnable orthogonal queries to extract disentangled representations of active and inactive receptor states. These latent representations are governed by a novel neural MWC gating module, which mathematically derives the probability of receptor activation from thermodynamic competition between ligand-state affinities and the receptor's intrinsic conformational energy barrier. A contrastive ranking objective is also introduced to enforce differential affinity constraints, ensuring physical consistency. Extensive experiments demonstrate that DSQ outperforms other baselines, particularly for the agonist subset. Additional homology-stratified, temperature-sensitivity, perturbation, clustering, and efficiency analyses show that DSQ provides useful thermodynamic inductive bias, while also exposing a clear limitation on low-homology receptors. The code is available at https://github.com/jiankliu/DSQ.

q-bio.QM↗

arXiv2026

An Optimal IPDG Scheme for the Biharmonic Equation

This paper presents an optimal interior penalty discontinuous Galerkin (IPDG) scheme for the planar biharmonic equation using piecewise polynomials of degree $k=3$ or $4$. In standard IPDG methods, large penalty parameters force the discrete solution into an overconstrained space, severely degrading accuracy---phenomenon known as numerical locking. To overcome this, our method enforces only vertex continuity and projects the jumps of the function and its normal derivative onto $\mathcal{P}^{k-3}$ and $\mathcal{P}^{k-2}$, respectively. Consequently, as the penalty parameters tend to infinity, the discrete solution is forced to lie in a constrained subspace $V_{h,\infty}^k$, which we identify as the optimal nonconforming finite element space $B_h^k$. This intrinsic connection fundamentally eliminates numerical locking. We prove optimal error estimates of $\mathcal{O}(h^{k-1})$ in a mesh-dependent energy norm and $\mathcal{O}(h^{k+1})$ in the $L^2$ norm. Numerical experiments on both convex and L-shaped domains confirm that the proposed scheme is robust and entirely locking-free, even for extremely large penalty parameters.

math.NA↗

arXiv2026

FAST Observations of Filamentary and Compact H I Structure in a Magellanic Stream IV Field

The Magellanic Stream (MS) is believed to have formed from gas removed from the Large and Small Magellanic Clouds through tidal forces and hydrodynamic interactions with the Milky Way's gaseous halo. It provides an important laboratory for studying how stripped gas fragments, mixes, and evolves in a circumgalactic environment. In this work, we present HI observations of a $3.8^\circ\times2.2^\circ$ field in the MS IV region, using the data from the Commensal Radio Astronomy FasT Survey (CRAFTS). The total HI mass in the analyzed field is $\simeq 5.3 \times 10^{6} (d/120\,{\rm kpc})^{2}\,M_{\odot}$, where $d$ is the distance to the MS IV gas. The data resolve the emission into three coherent filamentary HI structures with related but distinct velocity trends. To characterize the HI structures and study potential multiphase gas, we adopt a Gaussian decomposition procedure to identify and reconstruct sources. Our results indicate that the field is dominated by one major filamentary HI complex, together with several smaller kinematic clump-like components. Among these identified sources, only one source likely shows multiple velocity components. Because several sources appear spatially overlapped in projection, we further examine their apparent overlap regions using position-velocity (P-V) diagrams. P-V diagrams across the apparent overlap region show no clear intermediate-velocity bridge or V-shaped structure, favoring line-of-sight projection over direct cloud-cloud collision. These results demonstrate the value of deep, high-angular-resolution HI observations for resolving faint emission, compact morphologies, and kinematic structure in the MS.

astro-ph.GA↗

arXiv2026

Dense Process Supervision for Search Agents via Fact Utility Estimation

Reinforcement learning (RL) for search agents typically relies on outcome rewards. However, it often fails to achieve effective credit assignment, due to the unclear value of intermediate steps. It is hard to separate their contributions from the final result. In this paper, we propose a dense process supervision method based on fact utility estimation, which models the reasoning process as the accumulation of discrete evidence facts. We first extract structured facts from raw observations and organize them into an explicit fact store. To support credit assignment, we then cluster semantically equivalent facts and infer the posterior utility of each fact cluster using Bayesian estimation over group rollouts. Finally, we convert the estimated fact utilities into dense step-level rewards to guide RL training. Experiments on seven single-hop and multi-hop QA benchmarks show that our method consistently outperforms existing baselines. Ablation studies validate clear relative improvements on multi-hop QA compared to outcome reward-only training.

cs.CL↗

arXiv2026

MRMAD: A Multi-Round Multi-Audio Benchmark for Evaluating Acoustic Degradation Perception in Large Audio-Language Models

Large audio-language models (LALMs) have shown promising progress in understanding speech, music, and general sound events, yet their ability to reason about how audio signals are degraded remains underexplored. Existing benchmarks primarily evaluate semantic understanding, event recognition, or high-level audio reasoning, leaving a basic question unanswered: Do LALMs understand the differences in audio quality? We introduce MRMAD, a Multi-Round Multi-Audio Degradation benchmark for evaluating audio degradation perception and understanding in LALMs. MRMAD spans speech, music, and sound, and frames evaluation as multi-turn dialogues across multiple audio inputs, requiring models to identify types of degradation, compare severity, and perceive corruption changes across turns. Unlike current single-turn audio-language benchmarks, MRMAD evaluates whether LALMs can maintain consistent degradation hypotheses with new evidence and comprehend low-level acoustic phenomena over multi-turn dialogues. Through a systematic evaluation of 18 representative LALMs from non-thinking to reasoning and Omni models, we find that current models often recognize coarse content while failing to diagnose, compare, or reason about degradations reliably. Human evaluations further reveal a significant perception gap between LALMs and human listeners. MRMAD thus exposes a critical yet overlooked aspect of audio-language understanding and provides a diagnostic foundation for building future LALMs that are robust to real-world acoustic conditions.

cs.SD↗

arXiv2026

BTF-PINN: Enforcing Dirichlet Boundary Conditions Without Boundary Training

The homogeneous Dirichlet boundary value problem captures the core difficulty of solving Dirichlet problems with non-interpolatory methods. We propose BTF-PINN (Boundary-Training-Free Physics-Informed Neural Network), an interior-only strategy for solving homogeneous Dirichlet boundary value problems that requires no boundary training, boundary penalties, or boundary-conforming parametrizations. The key idea is to embed the essential boundary condition into a newly designed boundary-free loss function. We prove the equivalence between the proposed interior-only variational formulation and the original boundary value problem, establish a sharp threshold condition for the residual weight, and develop a convergence analysis based on the coercivity of the functional. Numerical experiments on high-dimensional problems, irregular geometries, and anisotropic elliptic equations demonstrate the effectiveness of BTF-PINN. Comparisons with standard boundary-penalty PINNs further show that BTF-PINN achieves superior boundary trace accuracy.

math.NA↗

arXiv2026

Asymptotics of Titchmarsh--Weyl functions near the real axis andan application to the KdV hierarchy

We establish high-energy asymptotic expansions of Titchmarsh--Weyl functions for one-dimensional Schrödinger and Dirac operators in regions whose boundaries approach the spectrum at a prescribed polynomial rate. For bounded potentials with bounded derivatives, the expansions remain uniform up to these boundaries, with an explicit loss in the remainder determined by the rate of approach. We treat both self-adjoint Dirac operators and non-self-adjoint Dirac operators with skew-adjoint potential matrices, keeping track of the distinct half-planes in which the two scalar Weyl coordinates are naturally defined. As an application of the Schrödinger expansion, we verify the high-energy hypothesis in Kotani's construction of KdV flows: for every odd integer $p\geq3$, each real-valued $q\in W^{2p-1,\infty}(\mathbb{R})$ generates a global classical solution of the member of the KdV hierarchy indexed by $(p+1)/2$.

math.SP↗

arXiv2026

Infrared Line Diagnostics Fail to Constrain Sgr A*'s UV Output

Sgr A*, the 4 x $10^6 M_{sun}$ supermassive black hole at the Galactic Center, exhibits frequent flaring with X-ray luminosities of $L_X \sim 10^{35}-10^{36}$ erg s$^{-1}$, while its ultraviolet (UV) emission remains unconstrained due to extreme extinction ($A_V \sim 30$ mag). We use JWST/MIRI time-resolved spectroscopy of the central 0.3" region to search for mid-infrared emission line variability driven by Sgr A* flares, comparing the results to CLOUDY photoionization models spanning flare luminosities of $L_{UV} = 10^{32}$-$10^{39}$ erg s$^{-1}$. We detect no continuum-correlated variability in any mid-infrared line, including [Fe II] (5.34 $μ$m), [Ne II] (12.813 $μ$m), [Fe II] (17.936 $μ$m), and [S III] (18.713 $μ$m) over lags of $\sim1$-8~hr. Despite expectations of a flare-driven response, we show that the lack of variability is consistent with the physical conditions in the spatially extended line-emitting gas, where light-crossing timescales of $\sim0.1$-$10$ days and recombination and cooling timescales much longer than the variability timescales suppress any observable response to individual flares. We further find that the modeled infrared spectra are dominated by continuum emission rather than isolated line emission. The brightest predicted lines are intrinsically weak (lower than 10$^{-5}$ mJy), and their detectability is further reduced by the large kinematic broadening expected at the CLOUDY-predicted emitting radii, reducing their contrast against the continuum. Extending the analysis to higher-ionization mid-infrared and near-infrared lines does not improve sensitivity. These results demonstrate that infrared emission lines trace a steady-state radiation field rather than individual flaring events, and therefore infrared line diagnostics cannot be used to constrain the instantaneous UV flux of Sgr A*.

astro-ph.HE↗

arXiv2026

Explicit Hamiltonian Classification in the $F_4(0)$ Toric Degeneration of $CP^2$

We give an explicit coordinate description of the Hamiltonian isotopy classes of the regular Lagrangian torus fibers of the smoothing $\widehat{F}_4(0)$ of the $F_4(0)$ toric degeneration, expressed in the explicit Oakley--Usher coordinates on $\CP^2(\sqrt2)$. For the wall fibers, they are not Hamiltonian isotopic to standard toric fibers, and no two distinct wall fibers are Hamiltonian isotopic. For the off-wall fibers, we find the standard toric fibers they are Hamiltonian isotopic to.

math.SG↗

arXiv2026

Localization phase diagram of the Hexagonal Lattice with irrational magnetic flux

We study the Hofstadter model on a hexagonal lattice with irrational magnetic flux in this work. The Hofstadter model of the square lattice with irrational flux has been solved mathematically by Avila and his collaborators in his Fields medal work. However, this theory is usually not applicable to lattices with internal degrees of freedom, such as spin or sublattice. In this work, we show that for the hexagonal lattice with only nearest neighbor hopping, the system can still be characterized by a two by two transfer matrix and solved exactly by the Avila global theory although this lattice has two sublattices. We obtained the exact localization phase diagram of the hexagonal lattice with irrational flux by this theory, which reveals three pure phases, i.e., the extended, localized and critical states but no mobility edge due to the chiral symmetry. We used the renormalization group (RG) theory to verify these results, which can determine part of the phase diagram. We then computed the fractal dimension of the remaining part numerically. The results from both the RG theory and numerical analysis confirmed the phase diagram we get from the Avila global theory. Our results can be tested in various hexagonal Moire lattices and artificial superlattices in recent experiments.

cond-mat.mes-hall↗

arXiv2026

FinVault: Benchmarking Financial Agent Safety in Execution-Grounded Environments

Financial agents powered by large language models (LLMs) are increasingly deployed for investment analysis, risk assessment, and automated decision-making, where their abilities to plan, invoke tools, and manipulate mutable state introduce new security risks in high-stakes and highly regulated financial environments. However, existing safety evaluations largely focus on language-model-level content compliance or abstract agent settings, failing to capture execution-grounded risks arising from real operational workflows and state-changing actions. To bridge this gap, we propose FinVault, the first execution-grounded security benchmark for financial agents, comprising 31 regulatory case-driven sandbox scenarios with state-writable databases and explicit compliance constraints, together with 107 real-world vulnerabilities and 963 test cases that systematically cover prompt injection, jailbreaking, financially adapted attacks, as well as benign inputs for false-positive evaluation. Experimental results reveal that existing defense mechanisms remain ineffective in realistic financial agent settings, with average attack success rates (ASR) still reaching up to 50.0\% on state-of-the-art models and remaining non-negligible even for the most robust systems (ASR 6.7\%), highlighting the limited transferability of current safety designs and the need for stronger financial-specific defenses. Our code can be found at https://github.com/aifinlab/FinVault.

cs.CR↗

arXiv2026

Isolating to Harness: Cross-Division Distillation for Fully Unsupervised Anomaly Detection

Fully Unsupervised Anomaly Detection (FUAD) addresses the practical scenario where training data is contaminated with unlabeled anomalies. This setting critically challenges conventional Unsupervised Anomaly Detection (UAD) methods, as they tend to misinterpret training anomalies as normal patterns, leading to false negatives. Although filtering anomalies from the training set is a common countermeasure, it inevitably discards valuable data and degrades the model's representation of normality. To overcome this dilemma, we propose an "isolating to harness" strategy, which isolates the influence of anomalies within specialized divisions and then leverages cross-division collaboration to generate robust pseudo supervision. We materialize this idea via a novel Cross-Division Distillation framework based on the widely studied Reverse Distillation paradigm. CDD first partitions the data into divisions with reduced anomaly ratios to train division-specific students. It then aggregates pseudo-normal features generated by each division-specific student for samples from other data divisions to guide a global student towards a robust anomaly-free representation. Experimental results on noisy versions of multiple AD datasets demonstrate that our method achieves significant performance improvements over the baseline. Code is available at https://github.com/hito2448/CDD.

cs.CV↗

arXiv2026

Private Private Information in Second-Price Auction

Classic results show that even an arbitrarily small correlation across bidders' information can enable full surplus extraction in auctions and related mechanism design settings. Motivated by this fragility, we study the information independence in a second-price auction when the seller commits to a private private information structure, meaning bidders' signals are independent ex ante, while bidders share a symmetric and arbitrarily correlated prior distribution over their valuations. We first show that the seller optimal efficient outcome with full surplus extraction can always be implemented by a private private information structure that admits a Bayes Nash equilibrium. However, this equilibrium may not be stable. We then further construct a private private information structure that achieves revenue arbitrarily close to maximum welfare while admitting a strict equilibrium. At the same time, we establish an impossibility result: under private private information, in general, bidder surplus cannot achieve maximal welfare exactly, and we characterize necessary and sufficient conditions on the prior distribution under which bidder surplus can be made arbitrarily close to maximal welfare. We finally explore which other efficient outcomes are achievable under private private information.

econ.TH↗

arXiv2026

HCPG-Flow:Hierarchical Contact-Progress Guidance for Flow-Policy Robot Manipulation

Flow policies can represent multimodal action distributions for robot manipulation, yet a robot must execute one action at each control step. When several proposals are sampled, critic-based ranking makes data collection depend on value estimates over candidate actions that may be weakly represented in replay. We introduce HCPG-Flow, an analytic rollout-time selector that augments SAC-Flow with hierarchical, object-centric contact-progress guidance while preserving its actor and critic objectives. HCPG switches from end-effector approach to task progress after contact, scores each proposal by the first-order reduction of a task-relevant distance, standardizes scores within the candidate set, and executes a temperature-controlled action embedding. Across ten simulated tasks, HCPG improves mean success over SAC-Flow on both benchmarks, including a 9.5 percentage-point gain on Maniskill. Four physical tasks further show high success with a 17.4% reduction in successful completion steps.Project page: https://hitxraz.github.io/HCPG-Flow/

cs.RO↗

arXiv2026

Convolution for Large Language Models

Large language models (LLMs) largely rely on Transformers, where self-attention provides global token interaction but does not explicitly encode the locality of natural language. We study whether lightweight depthwise convolutions can supply this local inductive bias without materially increasing model size. Our macro-level ablation compares convolution at 17 locations in a Qwen3 Transformer block and finds the best results when convolution is applied to the projected queries, keys, and values before attention. A subsequent micro-level study favors a residual depthwise convolution with kernel size $k=3$, without additional normalization or activation. Across Qwen3 models and several pre-training data budgets, this design improves the average accuracy on seven downstream benchmarks while adding less than $0.01\%$ parameters. A representation-level case study further suggests that the convolution makes repeated token IDs more sensitive to their immediate context. These results support depthwise convolution as a lightweight complement to self-attention for modeling short-range token interactions.

cs.CL↗