arXiv subjects

Peng Xu

Publications and source records attributed to Peng Xu.

At least 109 records · Page 6Linked to original sources

arXiv2025

Primordial Black Holes Formed during Magneto-Hydrodynamic Turbulence in The Early Universe as Dominant Part of Dark Matter

Primordial black holes (PBHs) offer a compelling candidate for dark matter. The production of PBHs through well-tested and accepted physical processes is highly worthy of investigation. This work highlights the role of turbulences in the very early universe in sustaining intense and persistent fluctuations in energy or mass density, which could provide a natural mechanism for PBH formation in the primordial universe. We analyze the mass range and abundance of PBHs produced in the magnetohydrodynamic turbulence induced by the electroweak phase transition. Remarkably, we find that the mass range of the produced PBHs falls within the most viable ``asteroid mass'' window from the present-day observations, and within natural parameter regions their abundance can be sufficiently large. These findings suggest that PBHs produced during magnetohydrodynamic turbulence in the very early universe may comprise a dominant part of dark matter.

gr-qc↗

arXiv2025

AdsQA: Towards Advertisement Video Understanding

Large language models (LLMs) have taken a great step towards AGI. Meanwhile, an increasing number of domain-specific problems such as math and programming boost these general-purpose models to continuously evolve via learning deeper expertise. Now is thus the time further to extend the diversity of specialized applications for knowledgeable LLMs, though collecting high quality data with unexpected and informative tasks is challenging. In this paper, we propose to use advertisement (ad) videos as a challenging test-bed to probe the ability of LLMs in perceiving beyond the objective physical content of common visual domain. Our motivation is to take full advantage of the clue-rich and information-dense ad videos' traits, e.g., marketing logic, persuasive strategies, and audience engagement. Our contribution is three-fold: (1) To our knowledge, this is the first attempt to use ad videos with well-designed tasks to evaluate LLMs. We contribute AdsQA, a challenging ad Video QA benchmark derived from 1,544 ad videos with 10,962 clips, totaling 22.7 hours, providing 5 challenging tasks. (2) We propose ReAd-R, a Deepseek-R1 styled RL model that reflects on questions, and generates answers via reward-driven optimization. (3) We benchmark 14 top-tier LLMs on AdsQA, and our \texttt{ReAd-R}~achieves the state-of-the-art outperforming strong competitors equipped with long-chain reasoning capabilities by a clear margin.

cs.CV↗

arXiv2025

Peekaboo, I See Your Queries: Passive Attacks Against DSSE Via Intermittent Observations

Dynamic Searchable Symmetric Encryption (DSSE) allows secure searches over a dynamic encrypted database but suffers from inherent information leakage. Existing passive attacks against DSSE rely on persistent leakage monitoring to infer leakage patterns, whereas this work targets intermittent observation - a more practical threat model. We propose Peekaboo - a new universal attack framework - and the core design relies on inferring the search pattern and further combining it with auxiliary knowledge and other leakage. We instantiate Peekaboo over the SOTA attacks, Sap (USENIX' 21) and Jigsaw (USENIX' 24), to derive their "+" variants (Sap+ and Jigsaw+). Extensive experiments demonstrate that our design achieves >0.9 adjusted rand index for search pattern recovery and 90% query accuracy vs. FMA's 30% (CCS' 23). Peekaboo's accuracy scales with observation rounds and the number of observed queries but also it resists SOTA countermeasures, with >40% accuracy against file size padding and >80% against obfuscation.

cs.CR↗

arXiv2025

Bayesian Analysis of Wave-Optics Gravitationally Lensed Massive Black Hole Binaries with Space-Based Gravitational Wave Detector

Within a Bayesian statistical framework, we jointly estimate the source and lens parameters and evaluate the relative evidence between the lensed and unlensed models. This work focuses on the wave optics effects induced by a point mass (PM) lens on gravitational waves (GW) from equal-mass massive binary black holes (MBHB), and assesses the capability of the space-based GW detector Taiji to detect such effects. Specifically, we investigate the impact of the redshifted lens mass MLz in the range [3e5, 3e7] solar masses, impact parameter y in [10, 50], source redshift zs in [4, 6], and total source mass Ms in [1e5, 1e7] solar masses on parameter estimation and model selection. Our results show that, for the cases we studied, larger MLz increases the waveform mismatch MM, which directly enhances the waveform difference and the corresponding signal-to-noise ratio (SNR), thereby improving the ability to discriminate between the lensed and unlensed models. In contrast, for y > 50, both MM and SNR are too small to allow effective model discrimination in these cases. Parameter estimation further indicates that for y < 50, the degeneracy between the luminosity distance and MLz can be effectively broken. Although the Bayes factor decreases as zs increases, lensing signatures remain identifiable up to zs = 6. The role of Ms depends on the overlap of the GW signal with the detector sensitive band. Overall, effective model discrimination requires MM greater than or equal to 1e-7 (corresponding to SNR greater than 5).

astro-ph.HE↗

arXiv2025

MIDAS: Modeling Ground-Truth Distributions with Dark Knowledge for Domain Generalized Stereo Matching

Despite the significant advances in domain generalized stereo matching, existing methods still exhibit domain-specific preferences when transferring from synthetic to real domains, hindering their practical applications in complex and diverse scenarios. The probability distributions predicted by the stereo network naturally encode rich similarity and uncertainty information. Inspired by this observation, we propose to extract these two types of dark knowledge from the pre-trained network to model intuitive multi-modal ground-truth distributions for both edge and non-edge regions. To mitigate the inherent domain preferences of a single network, we adopt network ensemble and further distinguish between objective and biased knowledge in the Laplace parameter space. Finally, the objective knowledge and the original disparity labels are jointly modeled as a mixture of Laplacians to provide fine-grained supervision for the stereo network training. Extensive experiments demonstrate that: (1) Our method is generic and effectively improves the generalization of existing networks. (2) PCWNet with our method achieves the state-of-the-art generalization performance on both KITTI 2015 and 2012 datasets. (3) Our method outperforms existing methods in comprehensive ranking across four popular real-world datasets.

cs.CV↗

arXiv2025

Quantum Sensing with Topological-Paired Bound States

We present an efficient and robust protocol for quantum-enhanced sensing using a single qubit in the topological waveguide system. Our method relies on the topological-paired bound states, which are localized near the qubit and can be effectively regarded as a two-level system. Through the lens of Bayesian inference theory, we show that the sensitivity can reach the Heisenberg limit across a large field range. Inheriting from the topological robustness of the waveguide, our sensing protocol is robust against local perturbations. Besides, our sensing protocol utilizes a product state as the initial state, which can be easily prepared in experiments. We expect this approach would pave the way toward robust topological quantum sensors based on near-term quantum platforms such as superconducting qubits and Rydberg arrays.

quant-ph↗

arXiv2025

RankMixer: Scaling Up Ranking Models in Industrial Recommenders

Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMixer retains the transformer's high parallelism while replacing quadratic self-attention with multi-head token mixing module for higher efficiency. Besides, RankMixer maintains both the modeling for distinct feature subspaces and cross-feature-space interactions with Per-token FFNs. We further extend it to one billion parameters with a Sparse-MoE variant for higher ROI. A dynamic routing strategy is adapted to address the inadequacy and imbalance of experts training. Experiments show RankMixer's superior scaling abilities on a trillion-scale production dataset. By replacing previously diverse handcrafted low-MFU modules with RankMixer, we boost the model MFU from 4.5\% to 45\%, and scale our ranking model parameters by 100x while maintaining roughly the same inference latency. We verify RankMixer's universality with online A/B tests across two core application scenarios (Recommendation and Advertisement). Finally, we launch 1B Dense-Parameters RankMixer for full traffic serving without increasing the serving cost, which improves user active days by 0.3\% and total in-app usage duration by 1.08\%.

cs.IR↗

arXiv2025

LONGER: Scaling Up Long Sequence Modeling in Industrial Recommenders

Modeling ultra-long user behavior sequences is critical for capturing both long- and short-term preferences in industrial recommender systems. Existing solutions typically rely on two-stage retrieval or indirect modeling paradigms, incuring upstream-downstream inconsistency and computational inefficiency. In this paper, we present LONGER, a Long-sequence Optimized traNsformer for GPU-Efficient Recommenders. LONGER incorporates (i) a global token mechanism for stabilizing attention over long contexts, (ii) a token merge module with lightweight InnerTransformers and hybrid attention strategy to reduce quadratic complexity, and (iii) a series of engineering optimizations, including training with mixed-precision and activation recomputation, KV cache serving, and the fully synchronous model training and serving framework for unified GPU-based dense and sparse parameter updates. LONGER consistently outperforms strong baselines in both offline metrics and online A/B testing in both advertising and e-commerce services at ByteDance, validating its consistent effectiveness and industrial-level scaling laws. Currently, LONGER has been fully deployed at more than 10 influential scenarios at ByteDance, serving billion users.

cs.IR↗

arXiv2025

Multi-Site rs-fMRI Domain Alignment for Autism Spectrum Disorder Auxiliary Diagnosis Based on Hyperbolic Space

Increasing the volume of training data can enable the auxiliary diagnostic algorithms for Autism Spectrum Disorder (ASD) to learn more accurate and stable models. However, due to the significant heterogeneity and domain shift in rs-fMRI data across different sites, the accuracy of auxiliary diagnosis remains unsatisfactory. Moreover, there has been limited exploration of multi-source domain adaptation models on ASD recognition, and many existing models lack inherent interpretability, as they do not explicitly incorporate prior neurobiological knowledge such as the hierarchical structure of functional brain networks. To address these challenges, we proposed a domain-adaptive algorithm based on hyperbolic space embedding. Hyperbolic space is naturally suited for representing the topology of complex networks such as brain functional networks. Therefore, we embedded the brain functional network into hyperbolic space and constructed the corresponding hyperbolic space community network to effectively extract latent representations. To address the heterogeneity of data across different sites and the issue of domain shift, we introduce a constraint loss function, Hyperbolic Maximum Mean Discrepancy (HMMD), to align the marginal distributions in the hyperbolic space. Additionally, we employ class prototype alignment to mitigate discrepancies in conditional distributions across domains. Experimental results indicate that the proposed algorithm achieves superior classification performance for ASD compared to baseline models, with improved robustness to multi-site heterogeneity. Specifically, our method achieves an average accuracy improvement of 4.03%. Moreover, its generalization capability is further validated through experiments conducted on extra Major Depressive Disorder (MDD) datasets. The code is available at https://github.com/LYQbyte/H2MSDA.

q-bio.NC↗

arXiv2025

CVFusion: Cross-View Fusion of 4D Radar and Camera for 3D Object Detection

4D radar has received significant attention in autonomous driving thanks to its robustness under adverse weathers. Due to the sparse points and noisy measurements of the 4D radar, most of the research finish the 3D object detection task by integrating images from camera and perform modality fusion in BEV space. However, the potential of the radar and the fusion mechanism is still largely unexplored, hindering the performance improvement. In this study, we propose a cross-view two-stage fusion network called CVFusion. In the first stage, we design a radar guided iterative (RGIter) BEV fusion module to generate high-recall 3D proposal boxes. In the second stage, we aggregate features from multiple heterogeneous views including points, image, and BEV for each proposal. These comprehensive instance level features greatly help refine the proposals and generate high-quality predictions. Extensive experiments on public datasets show that our method outperforms the previous state-of-the-art methods by a large margin, with 9.10% and 3.68% mAP improvements on View-of-Delft (VoD) and TJ4DRadSet, respectively. Our code will be made publicly available.

cs.CV↗

arXiv2025

Holographic deep thermalization for secure and efficient quantum random state generation

Randomness is a cornerstone of science, underpinning fields such as statistics, information theory, dynamical systems, and thermodynamics. In quantum science, quantum randomness, especially random pure states, plays a pivotal role in fundamental questions like black hole physics and quantum complexity, as well as in practical applications such as quantum device benchmarking and quantum advantage certification. The conventional approach for generating genuine random states, called `deep thermalization', faces significant challenges, including scalability issues due to the need for a large ancilla system and susceptibility to attacks, as demonstrated in this work. We introduce holographic deep thermalization, a secure and hardware-efficient quantum random state generator. By adopting a sequential application of a scrambling-measure-reset process, it continuously trades space with time, and substantially reduces the required ancilla size to as small as a system-size independent constant; At the same time, it guarantees security by removing quantum correlation between the data system and attackers. Thanks to the resource reduction, our circuit-based implementation on IBM Quantum devices achieves genuine $5$-qubit random state generation utilizing only a total of $8$ qubits.

quant-ph↗

arXiv2025

Enhancing Taiji's Parameter Estimation under Non-Stationarity: a Time-Frequency Domain Framework for Galactic Binaries and Instrumental Noises

The data analysis of space-based gravitational wave detectors like Taiji faces significant challenges from non-stationary noise, which compromises the efficacy of traditional frequency-domain analysis. This work proposes a unified framework based on short-time Fourier transform (STFT) to enhance parameter estimation of Galactic binary and characterization of instrumental noise under non-stationarity. Segmenting data into locally stationary intervals, we derive STFT-based models for signals and noises, and implement Bayesian inference via the extended Whittle likelihood. Validated through the analysis of verification Galactic binaries and instrumental noises, our STFT approach outperforms frequency-domain methods by reducing the uncertainty and bias of estimation, successfully recovering low signal-to-noise ratio signals missed by frequency-domain analysis, and mitigating the degeneracy among noise parameters. The framework's robustness against noise drifts and computational efficiency highlight its potential for integration into future global analysis pipelines.

gr-qc↗

arXiv2025

AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning

Despite recent progress in large-scale reinforcement learning (RL) for reasoning, the training recipe for building high-performing reasoning models remains elusive. Key implementation details of frontier models, such as DeepSeek-R1, including data curation strategies and RL training recipe, are often omitted. Moreover, recent research indicates distillation remains more effective than RL for smaller models. In this work, we demonstrate that large-scale RL can significantly enhance the reasoning capabilities of strong, small- and mid-sized models, achieving results that surpass those of state-of-the-art distillation-based models. We systematically study the RL training process through extensive ablations and propose a simple yet effective approach: first training on math-only prompts, then on code-only prompts. Notably, we find that math-only RL not only significantly enhances the performance of strong distilled models on math benchmarks (e.g., +14.6% / +17.2% on AIME 2025 for the 7B / 14B models), but also code reasoning tasks (e.g., +6.8% / +5.8% on LiveCodeBench for the 7B / 14B models). In addition, extended code-only RL iterations further improve performance on code benchmarks with minimal or no degradation in math results. We develop a robust data curation pipeline to collect challenging prompts with high-quality, verifiable answers and test cases to enable verification-based RL across both domains. Finally, we identify key experimental insights, including curriculum learning with progressively increasing response lengths and the stabilizing effect of on-policy parameter updates. We find that RL not only elicits the foundational reasoning capabilities acquired during pretraining and supervised fine-tuning (e.g., distillation), but also pushes the limits of the model's reasoning ability, enabling it to solve problems that were previously unsolvable.

cs.LG↗

arXiv2025

Characterization of NbTi wires for the electron-ion collider project

The Electron-Ion Collider (EIC) is a proposed machine to explore the behaviour of the fundamental particles and forces that bind atomic nuclei together. The design and construction of the EIC are underway at Brookhaven National Laboratory in collaboration with Thomas Jefferson National Accelerator Facility. EIC will use several different types of superconducting strands for magnets near the interaction region (IR). At beam injection, the magnetic field is usually very low compared with its maximum operating field. This usually creates considerable field errors mainly generated from magnetization current in superconducting strands even using very fine filament. The accurate magnetization measurement results from those superconducting strands will be critical for the calculation and future correction of magnetic field for EIC. In this work, we characterized three billets of superconductor NbTi strands. The magnetization was measured at 4.2 K and 1.9 K in magnetic fields below 1.5 T. The critical current at 4.2 K and in magnetic field down to 5 T were also measured. Other properties that are important for the safety margin of superconducting magnet fabrication, operation, and quench protection such as residual-resistance-ratio (RRR), filament diameter, Cu to non-Cu ratio, twist pitch, and mechanical properties at 77 K will also be presented.

physics.acc-ph↗

arXiv2025

Whole-Body Constrained Learning for Legged Locomotion via Hierarchical Optimization

Reinforcement learning (RL) has demonstrated impressive performance in legged locomotion over various challenging environments. However, due to the sim-to-real gap and lack of explainability, unconstrained RL policies deployed in the real world still suffer from inevitable safety issues, such as joint collisions, excessive torque, or foot slippage in low-friction environments. These problems limit its usage in missions with strict safety requirements, such as planetary exploration, nuclear facility inspection, and deep-sea operations. In this paper, we design a hierarchical optimization-based whole-body follower, which integrates both hard and soft constraints into RL framework to make the robot move with better safety guarantees. Leveraging the advantages of model-based control, our approach allows for the definition of various types of hard and soft constraints during training or deployment, which allows for policy fine-tuning and mitigates the challenges of sim-to-real transfer. Meanwhile, it preserves the robustness of RL when dealing with locomotion in complex unstructured environments. The trained policy with introduced constraints was deployed in a hexapod robot and tested in various outdoor environments, including snow-covered slopes and stairs, demonstrating the great traversability and safety of our approach.

cs.RO↗

arXiv2025

Effective potential and scattering length of shielding polar molecules

We investigate the effective potential and scattering length of ultracold polar molecules under different shielding techniques. First, we derive the effective potential for two polar molecules in the presence of an elliptical polarization field, combined elliptical and linear polarization fields, and combined elliptical polarization and static fields. The effective potential is then expressed as a sum of a zero-range contact interaction and a long-range dipole-dipole interaction under the Born approximation. We find that the first two shielding methods only partially suppress attractive interactions, while the second method allows for the construction of bound states with different polarization shapes. The last shielding method can achieve complete cancellation of residual attractive forces, which is particularly significant for maintaining quantum degeneracy in ultracold dipolar systems. Our results provide a comprehensive understanding of the effective potential and scattering length of shielding polar molecules, which is crucial for studying many-body physics in ultracold dipolar systems.

cond-mat.quant-gas↗

arXiv2025

VTBench: Comprehensive Benchmark Suite Towards Real-World Virtual Try-on Models

While virtual try-on has achieved significant progress, evaluating these models towards real-world scenarios remains a challenge. A comprehensive benchmark is essential for three key reasons:(1) Current metrics inadequately reflect human perception, particularly in unpaired try-on settings;(2)Most existing test sets are limited to indoor scenarios, lacking complexity for real-world evaluation; and (3) An ideal system should guide future advancements in virtual try-on generation. To address these needs, we introduce VTBench, a hierarchical benchmark suite that systematically decomposes virtual image try-on into hierarchical, disentangled dimensions, each equipped with tailored test sets and evaluation criteria. VTBench exhibits three key advantages:1) Multi-Dimensional Evaluation Framework: The benchmark encompasses five critical dimensions for virtual try-on generation (e.g., overall image quality, texture preservation, complex background consistency, cross-category size adaptability, and hand-occlusion handling). Granular evaluation metrics of corresponding test sets pinpoint model capabilities and limitations across diverse, challenging scenarios.2) Human Alignment: Human preference annotations are provided for each test set, ensuring the benchmark's alignment with perceptual quality across all evaluation dimensions. (3) Valuable Insights: Beyond standard indoor settings, we analyze model performance variations across dimensions and investigate the disparity between indoor and real-world try-on scenarios. To foster the field of virtual try-on towards challenging real-world scenario, VTBench will be open-sourced, including all test sets, evaluation protocols, generated results, and human annotations.

cs.CV↗

arXiv2025

Vertical Profile Corrected Satellite NH3 Retrievals Enable Accurate Agricultural Emission Characterization in China

Ammonia (NH3) emissions significantly contribute to atmospheric pollution, yet discrepancies exist between bottom-up inventories and satellite-constrained top-down estimates, with the latter typically one-third higher. This study quantifies how assumptions about NH3 vertical distribution in satellite retrievals contribute to this gap. By implementing spatially and temporally resolved vertical profiles from the Community Multiscale Air Quality model to replace steep gradients in Infrared Atmospheric Sounding Interferometer (IASI) retrievals, we reduced satellite-model column discrepancies from 71% to 18%. We subsequently constrained NH3 emissions across China using a hybrid inversion framework combining iterative mass balance and four-dimensional variational methods. Our posterior emissions showed agreement with the a priori inventory (7.9% lower), suggesting that discrepancies between inventory approaches were amplified by overestimation of near-surface NH3 in baseline satellite retrievals, potentially causing a 43% overestimation of growing season emissions. Evaluation against ground-based measurements confirmed improved model performance, with normalized root-mean-square error reductions of 1-27% across six months. These findings demonstrate that accurate representation of vertical profiles in satellite retrievals is critical for robust NH3 emission estimates and can reconcile the long-standing discrepancy between bottom-up and top-down approaches. Our hybrid inversion methodology, leveraging profile-corrected satellite data, reveals that China's NH3 emissions exhibit greater spatial concentration than previously recognized, reflecting agricultural intensification. This advancement enables timely and accurate characterization of rapidly changing agricultural emission patterns, critical for implementing effective nitrogen pollution control measures.

physics.ao-ph↗