arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 433 records · Page 24Linked to original sources

arXiv2025

NLO Corrections to Dimuonium Production in Photon-Photon Collision

Dimuonium ($μ^+ μ^-$) is one of the pure QED bound systems of leptons, together with positronium and ditauonium. The former had been observed in experiment in 1951, while the search for dimuonium and ditauonium ($τ^+ τ^-$) are still in vain. The ditauonium is thought to be hard to measure due to its short life time, whereas the dimuonium is very likely to be observed in current running experiments. We calculate in this work the para and ortho dimuonium production in photon-photon collision at the next-to-leading order (NLO) in QED. To handle the Coulomb divergence within the bound system, the non-relativistic QED description is adopted for consistency. The results indicate that the NLO corrections are negative for both para and ortho states, at the normal order of perturbative QED corrections. The measurement of the dimuonium in Belle II experiment, especially at the forthcoming facility of STCF, is tenable.

hep-ph↗

arXiv2025

Mitigating Easy Option Bias in Multiple-Choice Question Answering

In this early study, we observe an Easy-Options Bias (EOB) issue in some multiple-choice Visual Question Answering (VQA) benchmarks such as MMStar, RealWorldQA, SEED-Bench, Next-QA, STAR benchmark and Video-MME. This bias allows vision-language models (VLMs) to select the correct answer using only the vision (V) and options (O) as inputs, without the need for the question (Q). Through grounding experiments, we attribute the bias to an imbalance in visual relevance: the correct answer typically aligns more closely with the visual contents than the negative options in feature space, creating a shortcut for VLMs to infer the answer via simply vision-option similarity matching. To fix this, we introduce GroundAttack, a toolkit that automatically generates hard negative options as visually plausible as the correct answer. We apply it to the NExT-QA and MMStar datasets, creating new EOB-free annotations. On these EOB-free annotations, current VLMs approach to random accuracies under (V+O) settings, and drop to non-saturated accuracies under (V+Q+O) settings, providing a more realistic evaluation of VLMs' QA ability. Codes and new annotations will be released soon.

cs.CV↗

arXiv2025

Hybrid Deep Reconstruction for Vignetting-Free Upconversion Imaging through Scattering in ENZ Materials

Optical imaging through turbid or heterogeneous environments (collectively referred to as complex media) is fundamentally challenged by scattering, which scrambles structured spatial and phase information. To address this, we propose a hybrid-supervised deep learning framework to reconstruct high-fidelity images from nonlinear scattering measurements acquired with a time-gated epsilon-near-zero (ENZ) imaging system. The system leverages four-wave mixing (FWM) in subwavelength indium tin oxide (ITO) films to temporally isolate ballistic photons, thus rejecting multiply scattered light and enhancing contrast. To recover structured features from these signals, we introduce DeepTimeGate, a U-Net-based supervised model that performs initial reconstruction, followed by a Deep Image Prior (DIP) refinement stage using self-supervised learning. Our approach demonstrates strong performance across different imaging scenarios, including binary resolution patterns and complex vortex-phase masks, under varied scattering conditions. Compared to raw scattering inputs, it boosts average PSNR by 124%, SSIM by 231%, and achieves a 10 times improvement in intersection-over-union (IoU). Beyond enhancing fidelity, our method removes the vignetting effect and expands the effective field-of-view compared to the ENZ-based optical time gate output. These results suggest broad applicability in biomedical imaging, in-solution diagnostics, and other scenarios where conventional optical imaging fails due to scattering.

physics.optics↗

arXiv2025

Efficient High-Resolution Visual Representation Learning with State Space Model for Human Pose Estimation

Capturing long-range dependencies while preserving high-resolution visual representations is crucial for dense prediction tasks such as human pose estimation. Vision Transformers (ViTs) have advanced global modeling through self-attention but suffer from quadratic computational complexity with respect to token count, limiting their efficiency and scalability to high-resolution inputs, especially on mobile and resource-constrained devices. State Space Models (SSMs), exemplified by Mamba, offer an efficient alternative by combining global receptive fields with linear computational complexity, enabling scalable and resource-friendly sequence modeling. However, when applied to dense prediction tasks, existing visual SSMs face key limitations: weak spatial inductive bias, long-range forgetting from hidden state decay, and low-resolution outputs that hinder fine-grained localization. To address these issues, we propose the Dynamic Visual State Space (DVSS) block, which augments visual state space models with multi-scale convolutional operations to enhance local spatial representations and strengthen spatial inductive biases. Through architectural exploration and theoretical analysis, we incorporate deformable operation into the DVSS block, identifying it as an efficient and effective mechanism to enhance semantic aggregation and mitigate long-range forgetting via input-dependent, adaptive spatial sampling. We embed DVSS into a multi-branch high-resolution architecture to build HRVMamba, a novel model for efficient high-resolution representation learning. Extensive experiments on human pose estimation, image classification, and semantic segmentation show that HRVMamba performs competitively against leading CNN-, ViT-, and SSM-based baselines. Code is available at https://github.com/zhanghao5201/PoseVMamba.

cs.CV↗

arXiv2025

HunyuanWorld 1.0: Generating Immersive, Explorable, and Interactive 3D Worlds from Words or Pixels

Creating immersive and playable 3D worlds from texts or images remains a fundamental challenge in computer vision and graphics. Existing world generation approaches typically fall into two categories: video-based methods that offer rich diversity but lack 3D consistency and rendering efficiency, and 3D-based methods that provide geometric consistency but struggle with limited training data and memory-inefficient representations. To address these limitations, we present HunyuanWorld 1.0, a novel framework that combines the best of both worlds for generating immersive, explorable, and interactive 3D scenes from text and image conditions. Our approach features three key advantages: 1) 360° immersive experiences via panoramic world proxies; 2) mesh export capabilities for seamless compatibility with existing computer graphics pipelines; 3) disentangled object representations for augmented interactivity. The core of our framework is a semantically layered 3D mesh representation that leverages panoramic images as 360° world proxies for semantic-aware world decomposition and reconstruction, enabling the generation of diverse 3D worlds. Extensive experiments demonstrate that our method achieves state-of-the-art performance in generating coherent, explorable, and interactive 3D worlds while enabling versatile applications in virtual reality, physical simulation, game development, and interactive content creation.

cs.CV↗

arXiv2025

AUCAD: Automated Construction of Alignment Dataset from Log-Related Issues for Enhancing LLM-based Log Generation

Log statements have become an integral part of modern software systems. Prior research efforts have focused on supporting the decisions of placing log statements, such as where/what to log. With the increasing adoption of Large Language Models (LLMs) for code-related tasks such as code completion or generation, automated approaches for generating log statements have gained much momentum. However, the performance of these approaches still has a long way to go. This paper explores enhancing the performance of LLM-based solutions for automated log statement generation by post-training LLMs with a purpose-built dataset. Thus the primary contribution is a novel approach called AUCAD, which automatically constructs such a dataset with information extracting from log-related issues. Researchers have long noticed that a significant portion of the issues in the open-source community are related to log statements. However, distilling this portion of data requires manual efforts, which is labor-intensive and costly, rendering it impractical. Utilizing our approach, we automatically extract log-related issues from 1,537 entries of log data across 88 projects and identify 808 code snippets (i.e., methods) with retrievable source code both before and after modification of each issue (including log statements) to construct a dataset. Each entry in the dataset consists of a data pair representing high-quality and problematic log statements, respectively. With this dataset, we proceed to post-train multiple LLMs (primarily from the Llama series) for automated log statement generation. Both human and experimental evaluations indicate that these models significantly outperform existing LLM-based solutions, thereby validating the efficacy of our method for constructing a post-training dataset to enhance LLM-based log statement generation.

cs.SE↗

arXiv2025

SpectrumFM: Redefining Spectrum Cognition via Foundation Modeling

The enhancement of spectrum efficiency and the realization of secure spectrum utilization are critically dependent on spectrum cognition. However, existing spectrum cognition methods often exhibit limited generalization and suboptimal accuracy when deployed across diverse spectrum environments and tasks. To overcome these challenges, we propose a spectrum foundation model, termed SpectrumFM, which provides a new paradigm for spectrum cognition. An innovative spectrum encoder that exploits the convolutional neural networks and the multi-head self attention mechanisms is proposed to effectively capture both fine-grained local signal structures and high-level global dependencies in the spectrum data. To enhance its adaptability, two novel self-supervised learning tasks, namely masked reconstruction and next-slot signal prediction, are developed for pre-training SpectrumFM, enabling the model to learn rich and transferable representations. Furthermore, low-rank adaptation (LoRA) parameter-efficient fine-tuning is exploited to enable SpectrumFM to seamlessly adapt to various downstream spectrum cognition tasks, including spectrum sensing (SS), anomaly detection (AD), and wireless technology classification (WTC). Extensive experiments demonstrate the superiority of SpectrumFM over state-of-the-art methods. Specifically, it improves detection probability in the SS task by 30% at -4 dB signal-to-noise ratio (SNR), boosts the area under the curve (AUC) in the AD task by over 10%, and enhances WTC accuracy by 9.6%.

eess.SP↗

arXiv2025

Diffusion LLMs Can Do Faster-Than-AR Inference via Discrete Diffusion Forcing

Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to autoregressive (AR) LLMs for text generation, with the potential to decode multiple tokens in a single iteration. However, none of the existing open-source dLLMs have achieved superior inference speed over AR LLMs of similar size. This paper breaks this barrier based on a simple and effective strategy named discrete diffusion forcing (D2F). D2F equips dLLMs with two key capabilities: (1) block-wise autoregressive generation to enable KV cache utilization; (2) prediction of following tokens without requiring completion of prior blocks for inter-block parallel decoding. In this way, the vanilla dLLMs are refurbished into an AR-diffusion hybrid paradigm for efficient inference. D2F can be implemented with an asymmetric distillation process based on pre-trained dLLMs. We further propose a pipelined parallel decoding algorithm, which enables a trade-off between efficiency and efficacy. Empirically, D2F dLLMs achieve more than $\mathbf{2.5\times}$ inference speed than LLaMA3 and Qwen2.5 on GSM8K. Compared to vanilla dLLMs like LLaDA and Dream, the acceleration can be more than $\mathbf{50\times}$ while maintaining comparable output quality. The code is available at https://github.com/zhijie-group/Discrete-Diffusion-Forcing.

cs.LG↗

arXiv2025

A Design for an Early Quantum Network

With the rapid advancement of quantum information technology, quantum networks have become essential for supporting diverse applications, which often have stringent demands for key metrics such as fidelity and request completion time. In this work, we propose a design for early-stage quantum networks that is compatible with the three existing quantum repeater technologies. The design aims to maximize the ability of the network to accommodate the diverse needs of quantum applications, even under conditions of limited quantum resources and suboptimal network performance. We have also described the required identifiers in the quantum network and the specific process for implementing quantum requests. To assess the feasibility of our design, we conduct simulations based on discrete-event modeling of quantum networks. The simulations consider various types of noise and imperfect parameters that might exist in early-stage networks. We analyze the impact of these parameters on the fidelity of the generated entangled states and the request completion time. Furthermore, we investigated additional decisions that the central controller can make beyond path selection, such as the choice of cutoff time and the allocation of network resources to requests.

quant-ph↗

arXiv2025

How are pseudo-$q$-traces related to (co)ends?

Let $\mathbb V$ be an $\mathbb N$-graded $C_2$-cofinite vertex operator algebra (VOA), not necessarily rational or self-dual. Using a special case of the sewing-factorization theorem from [GZ25a], we show that the end $\mathbb E=\int_{\mathbb M\in\mathrm{Mod}(\mathbb V)}\mathbb M\otimes_{\mathbb C}\mathbb M'$ in $\mathrm{Mod}(\mathbb{V}^{\otimes2})$ (where $\mathbb{M}'$ is the contragredient module of $\mathbb{M}$) admits a natural structure of associative $\mathbb C$-algebra compatible with its $\mathbb{V}^{\otimes2}$-module structure. Moreover, we show that a suitable category $\mathrm{Coh}_{\mathrm{L}}(\mathbb E)$ of left $\mathbb E$-modules is isomorphic, as a linear category, to $\mathrm{Mod}(\mathbb V)$, and that the space of vacuum torus conformal blocks is isomorphic to the space $\mathrm{SLF}(\mathbb E)$ of symmetric linear functionals on $\mathbb E$. Combining these results with the main theorem of [GZ25b], we prove a conjecture of Gainutdinov-Runkel: For any projective generator $\mathbb G$ in $\mathrm{Mod}(\mathbb V)$, the pseudo-$q$-trace construction yields a linear isomorphism from $\mathrm{SLF}(\mathrm{End}_{\mathbb V}(\mathbb{G})^{\mathrm{opp}})$ to the space of vacuum torus conformal blocks of $\mathbb V$. In particular, if $A$ is a unital finite-dimensional $\mathbb C$-algebra such that the category of finite-dimensional left $A$-modules is equivalent to $\mathrm{Mod}(\mathbb V)$, then $\mathrm{SLF}(A)$ is linearly isomorphic to the space of vacuum torus conformal blocks of $\mathbb V$. This confirms a conjecture of Arike-Nagatomo.

math.QA↗

arXiv2025

SDMatte: Grafting Diffusion Models for Interactive Matting

Recent interactive matting methods have shown satisfactory performance in capturing the primary regions of objects, but they fall short in extracting fine-grained details in edge regions. Diffusion models trained on billions of image-text pairs, demonstrate exceptional capability in modeling highly complex data distributions and synthesizing realistic texture details, while exhibiting robust text-driven interaction capabilities, making them an attractive solution for interactive matting. To this end, we propose SDMatte, a diffusion-driven interactive matting model, with three key contributions. First, we exploit the powerful priors of diffusion models and transform the text-driven interaction capability into visual prompt-driven interaction capability to enable interactive matting. Second, we integrate coordinate embeddings of visual prompts and opacity embeddings of target objects into U-Net, enhancing SDMatte's sensitivity to spatial position information and opacity information. Third, we propose a masked self-attention mechanism that enables the model to focus on areas specified by visual prompts, leading to better performance. Extensive experiments on multiple datasets demonstrate the superior performance of our method, validating its effectiveness in interactive matting. Our code and model are available at https://github.com/vivoCameraResearch/SDMatte.

cs.CV↗

arXiv2025

IMoRe: Implicit Program-Guided Reasoning for Human Motion Q&A

Existing human motion Q\&A methods rely on explicit program execution, where the requirement for manually defined functional modules may limit the scalability and adaptability. To overcome this, we propose an implicit program-guided motion reasoning (IMoRe) framework that unifies reasoning across multiple query types without manually designed modules. Unlike existing implicit reasoning approaches that infer reasoning operations from question words, our model directly conditions on structured program functions, ensuring a more precise execution of reasoning steps. Additionally, we introduce a program-guided reading mechanism, which dynamically selects multi-level motion representations from a pretrained motion Vision Transformer (ViT), capturing both high-level semantics and fine-grained motion cues. The reasoning module iteratively refines memory representations, leveraging structured program functions to extract relevant information for different query types. Our model achieves state-of-the-art performance on Babel-QA and generalizes to a newly constructed motion Q\&A dataset based on HuMMan, demonstrating its adaptability across different motion reasoning datasets. Code and dataset are available at: https://github.com/LUNAProject22/IMoRe.

cs.CV↗

arXiv2025

VL-Cogito: Progressive Curriculum Reinforcement Learning for Advanced Multimodal Reasoning

Reinforcement learning has proven its effectiveness in enhancing the reasoning capabilities of large language models. Recent research efforts have progressively extended this paradigm to multimodal reasoning tasks. Due to the inherent complexity and diversity of multimodal tasks, especially in semantic content and problem formulations, existing models often exhibit unstable performance across various domains and difficulty levels. To address these limitations, we propose VL-Cogito, an advanced multimodal reasoning model trained via a novel multi-stage Progressive Curriculum Reinforcement Learning (PCuRL) framework. PCuRL systematically guides the model through tasks of gradually increasing difficulty, substantially improving its reasoning abilities across diverse multimodal contexts. The framework introduces two key innovations: (1) an online difficulty soft weighting mechanism, dynamically adjusting training difficulty across successive RL training stages; and (2) a dynamic length reward mechanism, which encourages the model to adaptively regulate its reasoning path length according to task complexity, thus balancing reasoning efficiency with correctness. Experimental evaluations demonstrate that VL-Cogito consistently matches or surpasses existing reasoning-oriented models across mainstream multimodal benchmarks spanning mathematics, science, logic, and general understanding, validating the effectiveness of our approach.

cs.CV↗

arXiv2025

100s TeV/m-Level Particle Accelerators Driven by High-density Electron Beams in Micro Structured Carbon Nanotube Forest Channel

Solid-state materials, such as carbon nanotubes (CNTs), have the potential to support ultra-high accelerating fields in the TV/m range for charged particle acceleration. In this study, we explore the feasibility of using nanostructured CNTs forest to develop plasma-based accelerators at the 100 TeV/m-level, driven by high-density, ultra-relativistic electron beams, using fully three-dimensional particle-in-cell simulations. Two different acceleration mechanisms are proposed and investigated: the surface plasmon leakage field and the bubble wakefield. The leakage field, driven by a relatively low-density beam, can achieve an acceleration field up to TV/m, capable of accelerating both electron and positron beams. In particular, due to the direct acceleration by the driver beam, the positron acceleration is highly efficient with an average acceleration gradient of 2.3 TeV/m. In contrast, the bubble wakefield mechanism allows significantly higher acceleration fields, e.g. beyond 400 TV/m, with a much higher energy transfer efficiency of $66.7\%$. In principle, electrons can be accelerated to PeV energies over distances of several meters. If the beam density is sufficiently high, the CNT target will be completely blown out, where no accelerating field is generated. Its threshold has been estimated. Two major challenges in these schemes are recognised and investigated. Leveraging the ultra-high energy and charge pumping rate of the driver beam, the nanostructured CNTs also offer significant potential for a wide range of advanced applications. This work represents a promising avenue for the development of ultra-compact, high-energy particle accelerators. We also outline conceptual experiments using currently available facilities, demonstrating that this approach is experimentally accessible.

physics.acc-ph↗

arXiv2025

Interactive Adversarial Testing of Autonomous Vehicles with Adjustable Confrontation Intensity

Scientific testing techniques are essential for ensuring the safe operation of autonomous vehicles (AVs), with high-risk, highly interactive scenarios being a primary focus. To address the limitations of existing testing methods, such as their heavy reliance on high-quality test data, weak interaction capabilities, and low adversarial robustness, this paper proposes ExamPPO, an interactive adversarial testing framework that enables scenario-adaptive and intensity-controllable evaluation of autonomous vehicles. The framework models the Surrounding Vehicle (SV) as an intelligent examiner, equipped with a multi-head attention-enhanced policy network, enabling context-sensitive and sustained behavioral interventions. A scalar confrontation factor is introduced to modulate the intensity of adversarial behaviors, allowing continuous, fine-grained adjustment of test difficulty. Coupled with structured evaluation metrics, ExamPPO systematically probes AV's robustness across diverse scenarios and strategies. Extensive experiments across multiple scenarios and AV strategies demonstrate that ExamPPO can effectively modulate adversarial behavior, expose decision-making weaknesses in tested AVs, and generalize across heterogeneous environments, thereby offering a unified and reproducible solution for evaluating the safety and intelligence of autonomous decision-making systems.

cs.RO↗

arXiv2025

A single-loop proximal-conditional-gradient penalty method

We consider the problem of minimizing a convex separable objective (as a separable sum of two proper closed convex functions $f$ and $g$) over a linear coupling constraint. We assume that $f$ can be decomposed as the sum of a smooth part having Hölder continuous gradient (with exponent $μ\in(0,1]$) and a nonsmooth part that admits efficient proximal mapping computations, while $g$ can be decomposed as the sum of a smooth part having Hölder continuous gradient (with exponent $ν\in(0,1]$) and a nonsmooth part that admits efficient linear oracles. Motivated by the recent works [1,49], we propose a single-loop variant of the standard penalty method, which we call a single-loop proximal-conditional-gradient penalty method (proxCG$^{\rm pen}_{1\ell}$), for this problem. In each iteration of proxCG$^{\rm pen}_{1\ell}$, we successively perform one proximal-gradient step involving $f$ and one conditional-gradient step involving $g$ on the quadratic penalty function, followed by an update of the penalty parameter. We present explicit rules for updating the penalty parameter and the stepsize in the conditional-gradient step in each iteration. Under a standard constraint qualification and domain boundedness assumption, we show that the objective value deviations (from the optimal value) along the sequence generated decay in the order of $t^{-\min\{μ,ν,1/2\}}$ with the associated feasibility violations decaying in the order of $t^{-1/2}$. Moreover, if the nonsmooth parts are indicator functions and the extended objective is a KL function with exponent $α\in[0,1)$, then the distances to the optimal solution set along the sequence generated by proxCG$^{\rm pen}_{1\ell}$ decay asymptotically at a rate of $t^{-(1-α)\min\{μ,ν,1/2\}}$. Finally, we illustrate numerically the behavior of proxCG$^{\rm pen}_{1\ell}$ on solving low rank Hankel matrix completion problems.

math.OC↗

arXiv2025

Computational complexity of three-dimensional Ising spin glass: Lessons from D-Wave annealer

Finding an exact ground state of a three-dimensional (3D) Ising spin glass is proven to be an NP-hard problem (i.e., at least as hard as any problem in the nondeterministic polynomial-time (NP) class). Given validity of the exponential time hypothesis, its computational complexity was proven to be no less than $2^{N^{2/3}}$, where $N$ is the total number of spins. Here, we report results of extensive experimentation with D-Wave 3D annealer with $N\le 5627$. We found exact ground states (in a probabilistic sense) for typical realizations of 3D spin glasses with the efficiency, which scales as $2^{N/ β}$ with $β\approx 10^3$. Based on statistical analysis of low-energy states, we argue that with an improvement of annealing protocols and device noise reduction, $β$ can be increased even further. This suggests that, for $N<β^3$, annealing devices provide most efficient way to find an exact ground state.

cond-mat.dis-nn↗

arXiv2025

TurboSpec: Closed-loop Speculation Control System for Optimizing LLM Serving Goodput

Large Language Model (LLM) serving systems batch concurrent user requests to achieve efficient serving. However, in real-world deployments, such inter-request parallelism from batching is often limited by external factors such as low request rates or memory constraints. Recent works focus on intra-request parallelism from speculative decoding as a solution to this problem. Unfortunately, benefits from intra-request parallelism are often fragile, as speculative decoding causes overhead, and speculated tokens may miss. We observe that speculative decoding may degrade LLM serving performance if added naively without tuning to the incoming requests and the speculation method. To alleviate the need for expert tuning and make speculative decoding more robust, we present TurboSpec, a speculation control system that automatically profiles the execution environment and utilizes a feedback-based algorithm to dynamically adjust the amount of intra-request parallelism in LLM serving. TurboSpec predicts "goodput" - the amount of successfully generated tokens - to evaluate and adjust intra-request parallelism amount to that with the highest goodput in runtime. We implement TurboSpec on a real-world LLM serving system vLLM and demonstrate its effectiveness across diverse workloads and hardware configurations, providing consistent performance improvements across all test scenarios.

cs.AI↗