arXiv subjects

Peng Wang

Publications and source records attributed to Peng Wang.

At least 307 records · Page 17Linked to original sources

arXiv2025

Gravitational Lensing by Black Holes in Einstein-nonlinear Electrodynamic Theories with Multiple Photon Spheres

In this paper, we study the gravitational lensing effects of non-linear electrodynamic black holes. Non-linear electrodynamic black holes serve as typical models for multi-event horizon black holes. Depending on the choice of metric parameters, these black holes can possess more than five event horizons. Consequently, within certain parameter ranges, black holes can have more than three photon spheres of varying sizes outside the event horizon. Specifically, we focus on the strong gravitational lensing effects near the triple photon spheres, particularly the formation of higher-order images of point sources and celestial spheres. The presence of one, two, or three or more photon spheres significantly increases the number of higher-order images of a point source. When a black hole is illuminated by a celestial sphere, the three photon spheres generate three critical curves in the black hole image, with the smallest critical curve coinciding with the shadow's edge. Additionally, since non-linear electrodynamic black holes are models of multi-event horizon black holes, we can infer the gravitational lensing effects and the changes in celestial images for black holes with more than three photon spheres by analyzing the distinctions and patterns between the gravitational lensing effects of one, two, and three photon spheres.

gr-qc↗

arXiv2025

Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy

A prevalent approach in Parameter-Efficient Fine-Tuning (PEFT) of pre-trained Vision Transformers (ViT) involves freezing the majority of the backbone parameters and solely learning low-rank adaptation weight matrices to accommodate downstream tasks. These low-rank matrices are commonly derived through the multiplication structure of down-projection and up-projection matrices, exemplified by methods such as LoRA and Adapter. In this work, we observe an approximate orthogonality among any two row or column vectors within any weight matrix of the backbone parameters; however, this property is absent in the vectors of the down/up-projection matrices. Approximate orthogonality implies a reduction in the upper bound of the model's generalization error, signifying that the model possesses enhanced generalization capability. If the fine-tuned down/up-projection matrices were to exhibit this same property as the pre-trained backbone matrices, could the generalization capability of fine-tuned ViTs be further augmented? To address this question, we propose an Approximately Orthogonal Fine-Tuning (AOFT) strategy for representing the low-rank weight matrices. This strategy employs a single learnable vector to generate a set of approximately orthogonal vectors, which form the down/up-projection matrices, thereby aligning the properties of these matrices with those of the backbone. Extensive experimental results demonstrate that our method achieves competitive performance across a range of downstream image classification tasks, confirming the efficacy of the enhanced generalization capability embedded in the down/up-projection matrices.

cs.CV↗

arXiv2025

RT-OVAD: Real-Time Open-Vocabulary Aerial Object Detection via Image-Text Collaboration

Aerial object detection plays a crucial role in numerous applications. However, most existing methods focus on detecting predefined object categories, limiting their applicability in real-world open scenarios. In this paper, we extend aerial object detection to open scenarios through image-text collaboration and propose RT-OVAD, the first real-time open-vocabulary detector for aerial scenes. Specifically, we first introduce an image-to-text alignment loss to replace the conventional category regression loss, thereby eliminating category constraints. Next, we propose a lightweight image-text collaboration strategy comprising an image-text collaboration encoder and a text-guided decoder. The encoder simultaneously enhances visual features and refines textual embeddings, while the decoder guides object queries to focus on class-relevant image features. This design further improves detection accuracy without incurring significant computational overhead. Extensive experiments demonstrate that RT-OVAD consistently outperforms existing state-of-the-art methods across open-vocabulary, zero-shot, and traditional closed-set detection tasks. For instance, on the open-vocabulary aerial detection benchmarks DIOR, DOTA-v2.0, and LAE-80C, RT-OVAD achieves 87.7 AP$_{50}$, 53.8 mAP, and 23.7 mAP, respectively, surpassing the previous state-of-the-art (LAE-DINO) by 2.2, 7.0, and 3.5 points. In addition, RT-OVAD achieves an inference speed of 34 FPS on an RTX 4090 GPU, approximately three times faster than LAE-DINO (10 FPS), meeting the real-time detection requirements of diverse applications. The code will be released at https://github.com/GT-Wei/RT-OVAD.

cs.CV↗

arXiv2025

Observed Anti-parallel Correlation Between Spiral Galaxy and Cosmic Filament Spins

Understanding the origin of galactic angular momentum and its connection to the cosmic web remains a pivotal issue in galaxy formation. Using kinematic data from the MaNGA survey, we investigate the alignment between the spin directions of spiral galaxies and their host cosmic filaments. By incorporating filament spin measurements derived from redshift asymmetry across filament spines, we reveal a mass-dependent anti-parallel correlation: low-mass spiral galaxies ($\log_{10}(M_*/M_\odot) \lesssim 10$) exhibit a statistically significant anti-parallel alignment between their stellar/gas spins and filament spins, while high-mass spirals show no such trend. Spatial analysis further indicates that high-mass spirals preferentially reside near filament spines, whereas low-mass spirals occupy filament outskirts. These findings extend previous alignment studies that neglected directional spin correlations and provide new insights into how cosmic environments shape galactic angular momentum. The observed anti-parallel trend suggests a critical role for filament spin in regulating the angular momentum acquisition of low-mass spirals. This anti-parallel alignment is significantly enhanced for low-mass spirals residing in dynamically cold filaments, highlighting the importance of filament properties in shaping galaxy spin.

astro-ph.GA↗

arXiv2025

Compressed Ultrafast Photography of Plasmas Formed from Laser Breakdown of Dense Gases Reveals that Internal Processes Dominate Evolution at Early Times

Compressed ultrafast photography (CUP) is applied to laser breakdown in argon and xenon under pressures up to 40atm to obtain 2D images of the plasma dynamics of single events with a spatial resolution of 250x100 pixels and an equivalent frame rate of 500 GHz. Light emission as a function of position and time is measured through red, green, blue, and broad-band filters. The spatially encoded and temporally sheared image normally used in CUP is now enhanced by the introduction of a constraint given by a spatially integrated and temporally sheared unencoded signal. The data yield insights into the temperature, opacity, the plasma formation process, and heat flow within the plasma and to the surrounding ambient gas. Contours of constant emission indicate that plasmas formed from sufficiently dense gas contract rather than expand despite having a temperature of a few eV. Plasmas formed from relatively low pressure gases such as 7atm argon can radiate with emissivity near unity. Modeling transport and opacity as arising from inverse Bremsstrahlung requires a degree of ionization that strongly exceeds expectations based on Saha's equation even as customarily modified to include density and screening. According to this model, both electrons and ions are strongly coupled with a plasma coefficient >1. During the first few nanoseconds after formation, Stefan-Boltzmann radiation and thermal conduction to ambient gas are too weak to explain the observed cooling rates, suggesting that transport within the plasma dominates its evolution. Yet, thermal conduction within the plasma itself is also small as indicated by the persistence of thermal inhomogeneities for far longer timescales. The fact that plasma is isolated from the surroundings makes it an excellent system for the study of the equation of state and hydrodynamics of such dense plasmas via the systems and techniques described.

physics.plasm-ph↗

arXiv2025

Schwarzschild Black Holes Immersed in Born-Infeld Magnetic Fields and Their Observational Signatures

We investigate the influence of Born-Infeld (BI) nonlinear electrodynamics on magnetic field configurations, photon orbits, and black hole shadows for Schwarzschild black holes immersed in magnetic fields. Assuming that the BI magnetic fields are asymptotically uniform and aligned with the polar axis, we solve the nonlinear magnetic field equations numerically using pseudospectral methods. Our analysis shows that nonlinear electromagnetic effects become prominent near the event horizon, particularly in the polar regions, where the magnetic field strength is significantly enhanced. This enhancement leads to closed photon orbits on the meridional plane becoming prolate, with noticeable stretching along the polar axis. Simulations of black hole images reveal that, at high observer inclinations, the shadow, which is circular in the Maxwell limit, becomes increasingly elongated along the polar direction as the nonlinear effects increase.

gr-qc↗

arXiv2025

Rotating Black Holes in Einstein-Born-Infeld Theory

We numerically construct a family of stationary, axisymmetric black hole solutions in Einstein-Born-Infeld theory, incorporating both electric charge and rotation. Our results indicate that when nonlinear electromagnetic effects are weak, rotating BI black holes with fixed spin approach the extremal limit as the electric charge increases. In contrast, strong nonlinear effects lead to the termination of solutions at configurations corresponding to naked singularities. We demonstrate that nonlinear electrodynamics enhances the gyromagnetic ratio relative to that of Kerr-Newman (KN) black holes. Additionally, we analyze the Innermost Stable Circular Orbits (ISCOs) and find that both prograde and retrograde ISCO radii are consistently smaller than those found in KN black holes.

gr-qc↗

arXiv2025

Hybrid Character Sums From Vectorial Dual-Bent Functions and Asymptotically Optimal Complex Codebooks With Small Alphabet Sizes

Hybrid character sums are an important class of exponential sums which have nice applications in coding theory and sequence design. Let $\gf_{p^m}$ be the finite field with $p^m$ elements for a prime $p$ and a positive integer $m$. Let $V_n^{(p)}$ be an $n$-dimensional vector space over $\gf_p$ for a prime $p$. In this paper, we study the hybrid character sums of the form \begin{eqnarray*} \sum_{x \in V_n^{(p)}}ψ\left(F(x)\right)χ_1\left(a x\right), \end{eqnarray*} where $F$ is a function from $V_n^{(p)}$ to $\gf_{p^m}$ and $a \in V_n^{(p)}$, $ψ$ is a nontrivial multiplicative character of $\gf_{p^m}$ and $χ_1$ is the canonical additive character of $V_n^{(p)}$. If $F(x)$ is a vectorial dual-bent function and $a \in V_n^{(p)}\setminus \{0\}$, we determine their complex modulus or explicit values under certain conditions, which generalizes some known results as special cases. It is concluded that the hybrid character sums from vectorial dual-bent functions have very small complex modulus. As applications, three families of asymptotically optimal complex codebooks are constructed from vectorial dual-bent functions and their maximal cross-correlation amplitude are determined based on the hybrid character sums. The constructed codebooks have very small alphabet sizes, which enhances their appeal for implementation. Besides, all of the three families of codebooks have only two-valued or three-valued cross-correlation amplitudes.

cs.IT↗

arXiv2025

Seedream 3.0 Technical Report

We present Seedream 3.0, a high-performance Chinese-English bilingual image generation foundation model. We develop several technical improvements to address existing challenges in Seedream 2.0, including alignment with complicated prompts, fine-grained typography generation, suboptimal visual aesthetics and fidelity, and limited image resolutions. Specifically, the advancements of Seedream 3.0 stem from improvements across the entire pipeline, from data construction to model deployment. At the data stratum, we double the dataset using a defect-aware training paradigm and a dual-axis collaborative data-sampling framework. Furthermore, we adopt several effective techniques such as mixed-resolution training, cross-modality RoPE, representation alignment loss, and resolution-aware timestep sampling in the pre-training phase. During the post-training stage, we utilize diversified aesthetic captions in SFT, and a VLM-based reward model with scaling, thereby achieving outputs that well align with human preferences. Furthermore, Seedream 3.0 pioneers a novel acceleration paradigm. By employing consistent noise expectation and importance-aware timestep sampling, we achieve a 4 to 8 times speedup while maintaining image quality. Seedream 3.0 demonstrates significant improvements over Seedream 2.0: it enhances overall capabilities, in particular for text-rendering in complicated Chinese characters which is important to professional typography generation. In addition, it provides native high-resolution output (up to 2K), allowing it to generate images with high visual quality.

cs.CV↗

arXiv2025

Chain-of-Thought Enhanced Shallow Transformers for Wireless Symbol Detection

Transformers have shown potential in solving wireless communication problems, particularly via in-context learning (ICL), where models adapt to new tasks through prompts without requiring model updates. However, prior ICL-based Transformer models rely on deep architectures with many layers to achieve satisfactory performance, resulting in substantial storage and computational costs. In this work, we propose CHain Of thOught Symbol dEtection (CHOOSE), a CoT-enhanced shallow Transformer framework for wireless symbol detection. By introducing autoregressive latent reasoning steps within the hidden space, CHOOSE significantly improves the reasoning capacity of shallow models (1-2 layers) without increasing model depth. This design enables lightweight Transformers to achieve detection performance comparable to much deeper models, making them well-suited for deployment on resource-constrained mobile devices. Experimental results demonstrate that our approach outperforms conventional shallow Transformers and achieves performance comparable to that of deep Transformers, while maintaining storage and computational efficiency. This represents a promising direction for implementing Transformer-based algorithms in wireless receivers with limited computational resources.

cs.LG↗

arXiv2025

Analyzing the Impact of Strategic Bidding on the Reserve Capacity via a Bi-Level Model

The growing integration of renewable energy sources necessitates adequate reserve capacity to maintain power balance. However, in market clearing, power companies with flexible resources may submit strategic bids to maximize profits, potentially compromising system reserves. This paper examines the effects of such strategic behavior by modeling the market as a bi-level problem. The upper level represents a strategic company aiming to maximize profit, while the lower level simulates the system operator clearing the market based on submitted offers. To enable duality-based solution methods, we approximate unit commitments with a continuous reserve capacity calculation. Case studies indicate that, in an imperfectly competitive market, more units are incentivized to operate,enhancing system reserves. However, some units go online mainly for profit, ultimately raising electricity costs for consumers. These findings highlight the importance of market design in managing the trade-off between reserve adequacy and economic efficiency in the presence of strategic bidding behavior.

eess.SY↗

arXiv2025

Spin-induced Scalarized Black Holes in Einstein-Maxwell-scalar Models

We construct spin-induced scalarized black hole solutions in a class of Einstein-Maxwell-scalar models, where a scalar field is non-minimally coupled to the electromagnetic field. Our results show that scalar hair develops only for rapidly rotating black holes, while slowly spinning ones remain well described by the Kerr-Newman (KN) metric. The scalar field contributes only a small fraction of the total mass, indicating suppressed nonlinear effects. This suppression may account for the narrow existence domains of scalarized black holes and the similarities observed in their existence domains across different coupling functions. Moreover, scalarized black holes are found to coexist with linearly stable, entropically favored KN black holes. These results motivate further investigations into the nonlinear dynamics and stability of scalarized black holes in these models.

gr-qc↗

arXiv2025

A Spatial Relationship Aware Dataset for Robotics

Robotic task planning in real-world environments requires not only object recognition but also a nuanced understanding of spatial relationships between objects. We present a spatial-relationship-aware dataset of nearly 1,000 robot-acquired indoor images, annotated with object attributes, positions, and detailed spatial relationships. Captured using a Boston Dynamics Spot robot and labelled with a custom annotation tool, the dataset reflects complex scenarios with similar or identical objects and intricate spatial arrangements. We benchmark six state-of-the-art scene-graph generation models on this dataset, analysing their inference speed and relational accuracy. Our results highlight significant differences in model performance and demonstrate that integrating explicit spatial relationships into foundation models, such as ChatGPT 4o, substantially improves their ability to generate executable, spatially-aware plans for robotics. The dataset and annotation tool are publicly available at https://github.com/PengPaulWang/SpatialAwareRobotDataset, supporting further research in spatial reasoning for robotics.

cs.RO↗

arXiv2025

New Dataset and Methods for Fine-Grained Compositional Referring Expression Comprehension via Specialist-MLLM Collaboration

Referring Expression Comprehension (REC) is a foundational cross-modal task that evaluates the interplay of language understanding, image comprehension, and language-to-image grounding. It serves as an essential testing ground for Multimodal Large Language Models (MLLMs). To advance this field, we introduced a new REC dataset in our previous conference paper, characterized by two key features. First, it is designed with controllable difficulty levels, requiring multi-level fine-grained reasoning across object categories, attributes, and multi-hop relationships. Second, it incorporates negative text and images generated through fine-grained editing and augmentation, explicitly testing a model's ability to reject scenarios where the target object is absent, an often overlooked yet critical challenge in existing datasets. In this extended work, we propose two new methods to tackle the challenges of fine-grained REC by combining the strengths of Specialist Models and MLLMs. The first method adaptively assigns simple cases to faster, lightweight models and reserves complex ones for powerful MLLMs, balancing accuracy and efficiency. The second method lets a specialist generate a set of possible object regions, and the MLLM selects the most plausible one using its reasoning ability. These collaborative strategies lead to significant improvements on our dataset and other challenging benchmarks. Our results show that combining specialized and general-purpose models offers a practical path toward solving complex real-world vision-language tasks. Our dataset and code are available at https://github.com/sleepyshep/FineCops-Ref.

cs.CV↗

arXiv2025

The Throughput Gain of Hypercycle-level Resource Reservation for Time-Triggered Ethernet

Time-Triggered Communication is a key technology for many safety-critical systems, with applications spanning the areas of aerospace and industrial control. Such communication relies on time-triggered flows, with each flow consisting of periodic packets originating from a source and destined for a destination node. Each packet needs to reach its destination before its deadline. Different flows can have different cycle lengths. To achieve assured transmission of time-triggered flows, existing efforts constrain the packets of a flow to be cyclically transmitted along the same path. Under such Fixed Cyclic Scheduling (FCS), reservation for flows with different cycle lengths can become incompatible over a shared link, limiting the total number of admissible flows. Considering the cycle lengths of different flows, a hyper-cycle has length equal to their least common multiple (LCM). It determines the time duration over which the scheduling compatibility of the different flows can be checked. In this work, we propose a more flexible schedule scheme called the Hypercycle-level Flexible Schedule (HFS) scheme, where a flow's resource reservation can change across cycles within a hypercycle. HFS can significantly increase the number of admitted flows by providing more scheduling options while remaining perfectly compatible with existing Time-Triggered Ethernet system. We show that, theoretically the possible capacity gain provided by HFS over FCS can be unbounded. We formulate the joint pathfinding and scheduling problem under HFS as an ILP problem which we prove to be NP-Hard. To solve HFS efficiently, we further propose a least-load-first heuristic (HFS-LLF), solving HFS as a sequence of shortest path problems. Extensive study shows that HFS admits up to 6 times the number of flows achieved by FCS. Moreover, our proposed HFS-LLF can run 104 times faster than solving HFS using a generic solver.

cs.NI↗

arXiv2025

Improving exponential-family random graph models for bipartite networks

Bipartite graphs, representing two-mode networks, arise in many research fields. These networks have two disjoint node sets representing distinct entity types, for example persons and groups, with edges representing associations between the two entity types. In bipartite graphs, the smallest possible cycle is a cycle of length four, and hence four-cycles are the smallest structure to model closure in such networks. Exponential-family random graph models (ERGMs) are a widely used model for social, and other, networks, including specifically bipartite networks. Existing ERGM terms to model four-cycles in bipartite networks, however, are relatively rarely used. In this work we demonstrate some problems with these existing terms to model four-cycles, and define new ERGM terms to help overcome these problems. The position of the new terms in the ERGM dependence hierarchy, and their interpretation, is discussed. The new terms are demonstrated in simulation experiments, and their application illustrated on a canonical example of an empirical two-mode network.

stat.ME↗

arXiv2025

ByteMorph: Benchmarking Instruction-Guided Image Editing with Non-Rigid Motions

Editing images with instructions to reflect non-rigid motions, camera viewpoint shifts, object deformations, human articulations, and complex interactions, poses a challenging yet underexplored problem in computer vision. Existing approaches and datasets predominantly focus on static scenes or rigid transformations, limiting their capacity to handle expressive edits involving dynamic motion. To address this gap, we introduce ByteMorph, a comprehensive framework for instruction-based image editing with an emphasis on non-rigid motions. ByteMorph comprises a large-scale dataset, ByteMorph-6M, and a strong baseline model built upon the Diffusion Transformer (DiT), named ByteMorpher. ByteMorph-6M includes over 6 million high-resolution image editing pairs for training, along with a carefully curated evaluation benchmark ByteMorph-Bench. Both capture a wide variety of non-rigid motion types across diverse environments, human figures, and object categories. The dataset is constructed using motion-guided data generation, layered compositing techniques, and automated captioning to ensure diversity, realism, and semantic coherence. We further conduct a comprehensive evaluation of recent instruction-based image editing methods from both academic and commercial domains.

cs.CV↗

arXiv2025

Language Embedding Meets Dynamic Graph: A New Exploration for Neural Architecture Representation Learning

Neural Architecture Representation Learning aims to transform network models into feature representations for predicting network attributes, playing a crucial role in deploying and designing networks for real-world applications. Recently, inspired by the success of transformers, transformer-based models integrated with Graph Neural Networks (GNNs) have achieved significant progress in representation learning. However, current methods still have some limitations. First, existing methods overlook hardware attribute information, which conflicts with the current trend of diversified deep learning hardware and limits the practical applicability of models. Second, current encoding approaches rely on static adjacency matrices to represent topological structures, failing to capture the structural differences between computational nodes, which ultimately compromises encoding effectiveness. In this paper, we introduce LeDG-Former, an innovative framework that addresses these limitations through the synergistic integration of language-based semantic embedding and dynamic graph representation learning. Specifically, inspired by large language models (LLMs), we propose a language embedding framework where both neural architectures and hardware platform specifications are projected into a unified semantic space through tokenization and LLM processing, enabling zero-shot prediction across different hardware platforms for the first time. Then, we propose a dynamic graph-based transformer for modeling neural architectures, resulting in improved neural architecture modeling performance. On the NNLQP benchmark, LeDG-Former surpasses previous methods, establishing a new SOTA while demonstrating the first successful cross-hardware latency prediction capability. Furthermore, our framework achieves superior performance on the cell-structured NAS-Bench-101 and NAS-Bench-201 datasets.

cs.LG↗