arXiv subjects

Guoqiang Zhang

Publications and source records attributed to Guoqiang Zhang.

At least 19 recordsLinked to original sources

arXiv2026

NovGauge: A Fine-Grained Benchmark for Diagnosing LLMs' Capability in Paper Novelty Assessment

Large language models (LLMs) are increasingly used in peer review at major AI conferences, yet novelty remains a persistent weak point. Existing benchmarks assess novelty as a single holistic score, making it difficult to diagnose which dimension a model misjudges or whether its evidence is faithful. We present NovGauge, a human-anchored benchmark for fine-grained novelty assessment diagnosis. The benchmark contains 619 paper pairs and 50 multi-paper sets, drawn from two expert sources: ICLR reviewer overlap claims and survey co-citations. Instances are independently labeled along three dimensions: task, problem, and method, capturing application goals, technical challenges, and solution approaches. We propose a cascading diagnostic pipeline that verifies per-dimension correctness, evidence grounding, and logical support. Evaluation of 18 LLMs shows hallucination rates ranging from 0% to 39% across dimensions, and among non-hallucinated correct-positive judgments, over 70% cite evidence fails to logically support the stated reason. The best-performing model, GPT-5.5, achieves 43-72% Verified F1 across dimensions, while most models retain less than half of their raw F1 after faithfulness verification. These results suggest that current LLMs remain far from reliable scientific novelty assessment, particularly when correctness is conditioned on faithful evidence grounding.

cs.AI↗

arXiv2026

Long-time asymptotics of the integrable defocusing Wadati-Konno-Ichikawa equation with a finite-genus algebro-geometric background

We study the finite-genus algebro-geometric solutions of the Wadati-Konno-Ichikawa (WKI) equation with the saturable nonlinearity and long-time asymptotic behaviors of their short-range perturbations. First, for both the focusing and defocusing reductions, we formulate the finite-genus Baker-Akhiezer functions as explicitly solvable the matrix Riemann-Hilbert (RH) problems on the complex spectral plane and obtain theta-function representations together with the reconstruction formulae for the WKI field and the reciprocal coordinate. We then consider the Cauchy problem of the defocusing WKI equation on a finite-genus algebro-geometric background. We construct the scattering data and RH problem, and perform a Deift-Zhou nonlinear steepest descent analysis. The space-time plane is divided into two transition regions, a Zakharov-Manakov (ZM) region, and a fast-decay region. The leading term is a phase-shifted finite-genus WKI solution. The transition corrections are governed by a Painlevé-XXXIV model, while the ZM radiation is described by parabolic-cylinder functions. The reciprocal-coordinate asymptotics are obtained simultaneously.

nlin.SI↗

arXiv2026

On the global well-posedness for the nonlocal Fokas-Lenells equation with the weighted Sobolev initial data on the line

We establish the global well-posedness of the Cauchy problem for the reverse space-time nonlocal Fokas-Lenells equation with the weighted Sobolev initial data $q_0(x)\in H^{3}(\mathbb{R}) \cap H^{2,1}(\mathbb{R})$ on the line. We develop the inverse scattering transform formulated via the associated Riemann-Hilbert problems to study this issue. A spectral uniformization transform is introduced to resolve the singular behavior inherent in the KN-type negative flow spectral problem. Owing to the reverse space-time reduction, reflection coefficients no longer satisfy the usual Hermitian conjugation symmetry, and the coercivity of the jump matrix is therefore not available a priori. The quantitative smallness condition on the initial data yields uniform bounds on the reflection coefficients and ensures the uniform positive definiteness of the Hermitian part of the associated jump matrix. The resulting coercivity allows us to establish the bounded invertibility of the associated singular integral operator through a Fredholm and vanishing-lemma argument. Under this condition, we prove an $L^{2}$-Sobolev bijective correspondence between the potential and scattering data, exclude spectral singularities on continuous spectra, and obtain the global existence and uniqueness of solutions. Moreover, the associated solution map is Lipschitz continuous on the admissible initial-data class.

nlin.SI↗

arXiv2026

Long-time asymptotic behavior for the defocusing Hirota equation on a finite-genus algebro-geometric background

In this paper, we investigate the long-time asymptotics for the solution of the Cauchy problem of the defocusing Hirota equation on a finite-genus algebro-geometric background in the whole $(x,t)$-half-plane, whose method is mainly based on a Riemann-Hilbert (RH) formulation and Deift-Zhou nonlinear steepest descent method. The critical values of the phase function in the associated RH problem divide the space-time plane into four regions, in which the leading-order term is given by a phase-shifted finite-genus algebro-geometric solution. The subleading behavior depends on the region: the correction is of order $t^{-1/3}$ and is governed by a Painlevé-XXXIV model RH problem in the transition regions; the leading radiation is of order $t^{-1/2}$ in the Zakharov--Manakov region; and the error is $O(t^{-1})$ in the fast-decay region. These results can also be extended to other higher-order members of the AKNS hierarchy.

nlin.SI↗

arXiv2026

VRPO: Rethinking Value Modeling for Robust RL under Noisy Supervision in LLM Post-Training

Reinforcement Learning (RL) in real-world environments often suffers from ambiguous or incomplete reward supervision, which undermines policy stability and generalization. Such noise may cause models to ignore key information or even collapse in advantage estimation. We find that a strong value model is essential for absorbing unstable signals and producing reliable advantages, offering denser and more robust supervision than the reward model. To better optimize noisy supervision, we propose VRPO, a framework that enhances value modeling for robust RL in LLM post-training. VRPO integrates (1) auxiliary losses guided by entropy and perplexity from a frozen language model, and (2) a variational information bottleneck, enabling the value model to filter noise and capture key words. This design allows the value model to correct noise rewards and generate more reliable advantage estimates, transforming it from a passive predictor into an active noise regulator. Experiments on multi-turn dialogue, math reasoning, and science QA with both rule-based and model-based rewards show that VRPO consistently outperforms baselines such as PPO and GRPO. Our work highlight the central role of the value model in Robust RL and provide a principled and practical approach to policy optimization under noisy supervision.

cs.LG↗

arXiv2026

Improved Belief-Attention in Vision Task

Recently, Belief-Attention \cite{Guoqiang25BeliefAttention} has been proposed by first performing an orthogonal projection of the softmax-based weighted summation of $V$ vectors with respect to the original $V$ vectors and then taking the perpendicular component as the residual signal in Transformer for performance improvement. In this paper, we first conduct an ablation study showing the projected component also carries information about the token correlation, which should not be ignored. We then propose to extend Belief-Attention by making use of both the perpendicular and projected components. In particular, the projected component goes through certain activation function and then a linear mapping before merging with the considered token. Conceptually speaking, the neural block for the projected component can be viewed as a two-layer feedforward network (FFN) within the new attention block. It is also noted that standard attention captures the token correlation via the inner-product matrix $QK^T$. We propose to introduce an additional inner-product matrix $ZZ^T$ to $QK^T$ to capture richer token correlation. We refer to the new module as Belief2-Attention. It can be easily shown that Belief2-Attention is more expressive than standard Attention. We then verify the effectiveness of Belief2-Attention for vision tasks of image classification and segmentation.

cs.CV↗

arXiv2026

AgentV-RL: Scaling Reward Modeling with Agentic Verifier

Verifiers have been demonstrated to enhance LLM reasoning via test-time scaling (TTS). Yet, they face significant challenges in complex domains. Error propagation from incorrect intermediate reasoning can lead to false positives for seemingly plausible solutions, while lacking external grounding makes verifiers unreliable on computation or knowledge-intensive tasks. To address these challenges, we propose Agentic Verifier, a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process. We introduce complementary forward and backward agents: one traces solutions from premises to conclusions, while the other re-checks conclusions against their underlying premises. This bidirectional process enables a comprehensive, reliable, and interpretable assessment of solutions. To facilitate practical deployment, we propose AgentV-RL. Through proactive exploration and reinforcement learning, the verifier autonomously interleaves tool-use with internal reasoning. Extensive experiments show that Agentic Verifier yields consistent performance gains under both parallel and sequential TTS. Notably, our 4B variant surpasses state-of-the-art ORMs by 25.2%, positioning it as a promising paradigm for agentic reward modeling.

cs.CL↗

arXiv2026

Lookahead Drifting Model

Recently, a new paradigm named \emph{drifting model} has been proposed for mapping distributions, which achieves the SOTA image generation performance over ImageNet via one-step neural functional evaluation (NFE). The basic idea is to compute a drifting term at each training iteration and then push the output of the model towards the direction of the drifting term. In this paper, we propose a \emph{lookahead drifting model}. At each training iteration, we compute a set of drifting terms sequentially. Each drifting term is calculated by making use of previously computed ones as well as the positive samples and the output of the model. %One key step is to properly scale the drifting terms so that their magnitudes are in a comparable range. In principle, the drifting terms obtained at a later stage capture higher order gradient information towards the positive samples. At each training iteration, the model is optimized by pushing its output towards the direction of the (weighted) summation of the drifting terms. Experimental results on toy examples and CIFAR10 demonstrate the better performance of the new method than the baseline.

cs.LG↗

arXiv2026

Enhanced electron injection for efficient proton acceleration and neutron production in femtosecond laser-driven nano-structured targets

Micro- or nano-structured targets are advantageous in enhancing and manipulating laser-proton acceleration, due to the increased absorption of laser energy and onset of direct laser acceleration for high-energy electrons. Here, we experimentally demonstrate that nano-wire-array printed on a flat substrate is an efficient nano-injector of relativistic electrons that leads to a significant boost of laser-driven proton acceleration and neutron production beyond normal geometry. By employing an ultra-intense (2*1021 W/cm2) femtosecond laser pulse to irradiate nano-wire-array targets, protons with cut-off energies of 62.8 MeV are generated, and notably, the energy conversion efficiency from laser to protons reaches up to 9% - 3.5 times higher than that of flat foils. After bombarding a beryllium converter, 1.1*1010 neutrons are produced. Full 3D particle-in-cell simulations have reproduced experimental results and reveal interference mechanisms between the nano-wires and substrate, leading to continuous pumping of electrons from the substrate and standing-wave enhanced re-injection from the wire tip. This efficient injection finally results in the large sheath field and thus high yield of energetic protons and neutrons. Dependence on the wire length and scaling with laser amplitude are further discussed. These results suggest that 3D-printed structures are promising in developing compact laser-driven high-flux proton and neutron sources for numerous applications.

physics.plasm-ph↗

arXiv2026

Confusion-Aware Spectral Regularizer for Long-Tailed Recognition

Long-tailed image classification remains a long-standing challenge, as real-world data typically follow highly imbalanced distributions where a few head classes dominate and many tail classes contain only limited samples. This imbalance biases feature learning toward head categories and leads to significant degradation on rare classes. Although recent studies have proposed re-sampling, re-weighting, and decoupled learning strategies, the improvement on the most underrepresented classes still remains marginal compared with overall accuracy. In this work, we present a confusion-centric perspective for long-tailed recognition that explicitly focuses on worst-class generalization. We first establish a new theoretical framework of class-specific error analysis, which shows that the worst-class error can be tightly upper-bounded by the spectral norm of the frequency-weighted confusion matrix and a model-dependent complexity term. Guided by this insight, we propose the Confusion-Aware Spectral Regularizer (CAR) that minimizes the spectral norm of the confusion matrix during training to reduce inter-class confusion and enhance tail-class generalization. To enable stable and efficient optimization, CAR integrates a Differentiable Confusion Matrix Surrogate and an EMA-based Confusion Estimator to maintain smooth and low-variance estimates across mini-batches. Extensive experiments across multiple long-tailed benchmarks demonstrates that CAR substantially improves both worst-class accuracy and overall performance. When combined with ConCutMix augmentation, CAR consistently surpasses exisiting state-of-the-art long-tailed learning methods under both the training-from-scratch setting (by 2.37% ~ 4.83%) and the fine-tuning-from-pretrained setting (by 2.42% ~ 4.17%) across ImageNet-LT, CIFAR100-LT, and iNaturalist datasets.

cs.CE↗

arXiv2026

CCTU: A Benchmark for Tool Use under Complex Constraints

Solving problems through tool use under explicit constraints constitutes a highly challenging yet unavoidable scenario for large language models (LLMs), requiring capabilities such as function calling, instruction following, and self-refinement. However, progress has been hindered by the absence of dedicated evaluations. To address this, we introduce CCTU, a benchmark for evaluating LLM tool use under complex constraints. CCTU is grounded in a taxonomy of 12 constraint categories spanning four dimensions (i.e., resource, behavior, toolset, and response). The benchmark comprises 200 carefully curated and challenging test cases across diverse tool-use scenarios, each involving an average of seven constraint types and an average prompt length exceeding 4,700 tokens. To enable reliable evaluation, we develop an executable constraint validation module that performs step-level validation and enforces compliance during multi-turn interactions between models and their environments. We evaluate nine state-of-the-art LLMs in both thinking and non-thinking modes. Results indicate that when strict adherence to all constraints is required, no model achieves a task completion rate above 20%. Further analysis reveals that models violate constraints in over 50% of cases, particularly in the resource and response dimensions. Moreover, LLMs demonstrate limited capacity for self-refinement even after receiving detailed feedback on constraint violations, highlighting a critical bottleneck in the development of robust tool-use agents. To facilitate future research, we release the data and code.

cs.CL↗

arXiv2025

Intrinsic Fracture Nonreciprocity at the Nanoscale

We reveal intrinsic fracture nonreciprocity, manifesting as directional asymmetry in crack resistance, in two-dimensional heterostructures engineered through lattice-mismatched interfaces. Density-functional theory combined with machine-learning molecular dynamics show that intrinsic lattice mismatch between bonded component crystals imprints asymmetric prestrain states at crack tips, governing bond-breaking thresholds through charge redistribution. The failure criterion obeys a universal exponential scaling law between normalized charge density and bond strain, insensitive to bonding chemistry and local atomic environment. The magnitude of nonreciprocity scales systematically with lattice mismatch, reaching 49% at 10% mismatch. Validation across hexagonal, square, rectangular, and oblique two-dimensional lattices confirms universality, establishing interface strain engineering as a general design principle that bridges electronic structure to nanoscale failure, enabling rational design of damage-tolerant nanostructures.

cond-mat.mtrl-sci↗

arXiv2025

Breather gas and shielding for the focusing nonlinear Schrödinger equation with nonzero backgrounds

Breathers have been experimentally and theoretically found in many physical systems -- in particular, in integrable nonlinear-wave models. A relevant problem is to study the \textit{breather gas}, which is the limit, for $N\rightarrow \infty $, of $N$-breather solutions. In this paper, we investigate the breather gas in the framework of the focusing nonlinear Schrödinger (NLS) equation with nonzero boundary conditions, using the inverse scattering transform and Riemann-Hilbert problem. We address aggregate states in the form of $N$-breather solutions, when the respective discrete spectra are concentrated in specific domains. We show that the breather gas coagulates into a single-breather solution whose spectral eigenvalue is located at the center of the circle domain, and a multi-breather solution for the higher-degree quadrature concentration domain. These coagulation phenomena in the breather gas are called \textit{breather shielding}. In particular, when the nonzero boundary conditions vanish, the breather gas reduces to an $n$-soliton solution. When the discrete eigenvalues are concentrated on a line, we derive the corresponding Riemann-Hilbert problem. When the discrete spectrum is uniformly distributed within an ellipse, it is equivalent to the case of the line domain. These results may be useful to design experiments with breathers in physical settings.

nlin.SI↗

arXiv2025

Ultrashort Time-Integrated Diagnosis of Laser-Heated Deuterium Ions in Dense Plasma via Fusion Neutron Spectra

The ultrashort time-integrated diagnosis of ions plays a vital role in high energy density physics research. However, it is extremely challenging to measure in experiment. Here, we demonstrate a reliable approach for investigating the dynamics of deuterium ions in dense plasma. By irradiating a heavy water stream with the hundred Hertz repetitive intense femtosecond laser pulses, the neutrons from D(D,n)3He reaction can be detected via a single Time-of-Flight detector to accumulate the spectrum with a fine energy-resolution. This spectrum has been utilized to calculate the temperature and angular distribution of deuterium ions transported in plasma. And the calculated results are well verified by particle-in-cell simulations of deuterium ions dynamics. Our method paves a new way for diagnosing ions picoseconds time-integrated dynamics in plasma and holds great potential for understanding the ions transport process in high-energy density matters and studying laser plasma ion acceleration.

physics.plasm-ph↗

arXiv2025

Production of Iodine Isotopes via Ultra-intense Laser Driven Photonuclear Reactions

The investigation and production of proton-rich iodine isotopes predominantly rely on conventional accelerator-based methods, typically requiring prolonged irradiation periods to measure or achieve quantifiable yields for isotopic isolation. Bremsstrahlung radiation sources generated by high-power laser-plasma-accelerated electron beams with ultrahigh charge (tens of nanocoulombs) bombarding high-Z targets demonstrate extraordinary photon flux characteristics. An electron beam with a total charge of approximately 47.7 nC (E$_e$ $\gt$ 10.4 MeV) was generated in our experiment by focusing a ultra-intense laser pulse onto a deuterium gas jet. Laser-driven bremsstrahlung was employed to induce $^{127}I$$(γ,xn)$ ($x$ = 1,3,4,6-8), and the product yields and the corresponding flux-weighted average cross sections are reported. Our results demonstrate production of medical isotopes, with average yields of $^{124}$I and $^{123}$I at approximately $9.83\pm0.45\times10^{5}$/shot and $2.81\pm0.11\times10^{5}$/shot, respectively. This method, utilizing high-power lasers to generate bremsstrahlung radiation, shows significant potential for medical applications and opens new avenues for studying photonuclear processes in astrophysical contexts.

nucl-ex↗

arXiv2025

GeV-level $γ$-ray and positron beams produced by collisions of ultra-intense ultra-short laser on high-energy electron beam

Based on collisions between the 100 PW laser and 8 GeV superconducting linear accelerator constructing at the Shanghai hard X-ray free electron laser system (SHINE), the building of GeV-level $γ$-ray as well as positron beams are proposed according to particle-in-cell simulations. Key processes are considered involving the nonlinear inverse Compton scattering for $γ$-ray generation and the multiphoton Breit-Wheeler process for electron-positron pair production. Regardless of laser polarization, the simulations indicate that $γ$-ray beams achieve energy up to 8 GeV, brilliance around 10$^{27}$ photons/(s mm$^{2}$ mrad$^{2}$), and emittance as low as 0.1 mm mrad, while positron beams reach energy up to 7 GeV, brilliance around 4 $\times$ 10$^{24}$ positrons/(s mm$^{2}$ mrad$^{2}$), and emittance as low as 0.1 mm mrad. Various applications could benefit from the possible high-energy $γ$-ray and positron beams built at the SHINE facility, including fundamental physics of strong-field quantum electrodynamics theory validation, nuclear physics, radiopharmaceutical preparation, and imaging, etc.

physics.acc-ph↗

arXiv2025

High Quality Diffusion Distillation on a Single GPU with Relative and Absolute Position Matching

We introduce relative and absolute position matching (RAPM), a diffusion distillation method resulting in high quality generation that can be trained efficiently on a single GPU. Recent diffusion distillation research has achieved excellent results for high-resolution text-to-image generation with methods such as phased consistency models (PCM) and improved distribution matching distillation (DMD2). However, these methods generally require many GPUs (e.g.~8-64) and significant batchsizes (e.g.~128-2048) during training, resulting in memory and compute requirements that are beyond the resources of some researchers. RAPM provides effective single-GPU diffusion distillation training with a batchsize of 1. The new method attempts to mimic the sampling trajectories of the teacher model by matching the relative and absolute positions. The design of relative positions is inspired by PCM. Two discriminators are introduced accordingly in RAPM, one for matching relative positions and the other for absolute positions. Experimental results on StableDiffusion (SD) V1.5 and SDXL indicate that RAPM with 4 timesteps produces comparable FID scores as the best method with 1 timestep under very limited computational resources.

cs.CV↗

arXiv2025

Rigorous analysis of large-space and long-time asymptotics for the short-pulse soliton gases

We rigorously analyze the asymptotics of soliton gases to the short-pulse (SP) equation. The soliton gas is formulated in terms of a RH problem, which is derived from the RH problems of the $N$-soliton solutions with $N \to \infty$. Building on prior work in the study of the KdV soliton gas and orthogonal polynomials with Jacobi-type weights, we extend the reflection coefficient to two generalized forms on the interval $\left[η_1, η_2\right]$: $r_0(λ) = \left(λ- η_1\right)^{β_1}\left(η_2 - λ\right)^{β_2}|λ- η_0|^{β_0}γ(λ)$, $r_c(λ) = \left(λ- η_1\right)^{β_1}\left(η_2 - λ\right)^{β_2}χ_c(λ)γ(λ)$, where $0 < η_1 < η_0 < η_2$ and $β_j > -1$ ($j = 0, 1, 2$), $γ(λ)$ is continuous and positive on $\left[η_1, η_2\right]$, with an analytic extension to a neighborhood of this interval, $χ_c(λ) = 1$ for $λ\in \left[η_1, η_0\right)$ and $χ_c(λ) = c^2$ for $λ\in \left(η_0, η_2\right]$, where $c>0$ with $c \neq 1$. The asymptotic analysis is performed using the steepest descent method. A key aspect of the analysis is the construction of the $g$-function. To address the singularity at the origin, we introduce an innovative piecewise definition of $g$-function. To establish the order of the error term, we construct local parametrices near $η_j$ for $j = 1, 2$, and singularity $η_0$. At the endpoints, we employ the Airy parametrix and the first type of modified Bessel parametrix. At the singularity $η_0$, we use the second type of modified Bessel parametrix for $r_0$ and confluent hypergeometric parametrix for $r_c(λ)$.

nlin.SI↗