arXiv subjects

Hao Zhang

Publications and source records attributed to Hao Zhang.

At least 361 records · Page 20Linked to original sources

arXiv2025

Evaluation-Driven Development and Operations of LLM Agents: A Process Model and Reference Architecture

Large Language Models (LLMs) have enabled the emergence of LLM agents, systems capable of pursuing under-specified goals and adapting after deployment. Evaluating such agents is challenging because their behavior is open ended, probabilistic, and shaped by system-level interactions over time. Traditional evaluation methods, built around fixed benchmarks and static test suites, fail to capture emergent behaviors or support continuous adaptation across the lifecycle. To ground a more systematic approach, we conduct a multivocal literature review (MLR) synthesizing academic and industrial evaluation practices. The findings directly inform two empirically derived artifacts: a process model and a reference architecture that embed evaluation as a continuous, governing function rather than a terminal checkpoint. Together they constitute the evaluation-driven development and operations (EDDOps) approach, which unifies offline (development-time) and online (runtime) evaluation within a closed feedback loop. By making evaluation evidence drive both runtime adaptation and governed redevelopment, EDDOps supports safer, more traceable evolution of LLM agents aligned with changing objectives, user needs, and governance constraints.

cs.SE↗

arXiv2025

Deep Learning Assisted Modeling for $χ^{(2)}$ Nonlinear Optics

Modeling second-order ($χ^{(2)}$) nonlinear optical processes remains computationally expensive due to the need to resolve fast field oscillations and simulate wave propagation using methods like the split-step Fourier method (SSFM). This can become a bottleneck in real-time applications, such as high-repetition-rate laser systems requiring rapid feedback and control. We present an LSTM-based surrogate model trained on SSFM simulations generated from a start-to-end model of the photocathode drive laser at SLAC National Accelerator Laboratory's Linac Coherent Light Source II. The model achieves over 250x speedup while maintaining high fidelity, enabling future real-time optimization and laying the foundation for data-integrated modeling frameworks and digital twins of laser systems.

physics.optics↗

arXiv2025

Charged-particle control via spatio-temporally tailored pulses from gas-based nonlinear optics

Gas-filled waveguides enable few-cycle, spatio-temporally coupled (STC) pulses with programmable structure, opening new routes to control charged particles with optical fields. This review maps the landscape of optical-field-driven photoemission, then surveys gas-based nonlinear drivers, photonic crystal fibers (PCFs) for low-energy, high-repetition operation and hollow-core capillaries (HCCs) for high-power, few-cycle synthesis. We highlight mechanisms for deterministic pulse shaping, including four-wave-mixing-based spectral-phase transfer in HCCs, and show how tailored STC waveforms steer emission dynamics from the multiphoton to tunneling regimes, enabling sub-cycle gating, momentum control, and brightness scaling. We conclude with open challenges: phase stability, mid-IR scalability, coupling to nanophotonic emitters, metrology of vectorial fields, and outline a path toward compact, ultrafast, phase-coherent electron sources and emerging quantum applications powered by nonlinear photonics.

physics.optics↗

arXiv2025

Coherent synchrotron radiation by excitation of surface plasmon polariton on near-critical solid microtube surface

Coherent synchrotron radiation (CSR) is crucial for the development of powerful ultrashort light sources. We present a mechanism for generating CSR in the form of generalised superradiance, based on surface plasmon polaritons (SPPs), which are resonantly excited on a solid, near-critical-density inner surface of a microtube. A high-intensity, circularly polarised laser pulse, propagating along the microtube axis, efficiently couples the cylindrical SPP modes. This process creates azimuthally structured, rotating electromagnetic fields. These rotating fields subsequently confine, modulate, and directly accelerate surface electrons to emit CSR in the Vavilov-Cherenkov angle. We further demonstrate that by improving the azimuthal symmetry of these electrons, the helical modulation enables CSR emission across all azimuthal directions in the form of isolated harmonics, significantly enhancing radiation intensity even when full coherence is imperfect. Our full 3D Particle-in-Cell simulations indicate this scheme can generate X-rays with coherence enhanced by up to two orders of magnitude compared to incoherent emission. The challenges to experimentally realise this scheme are discussed, including the need for high-contrast lasers to prevent pre-plasma formation and the demanding tolerances for microtube fabrication and alignment, while these challenges are not beyond the scope of existing or near-future experimental capabilities.

physics.plasm-ph↗

arXiv2025

Little red dots as embryos of active galactic nuclei

As an unprecedented large population in the early universe, the JWST-discovered little red dots (LRDs) have garnered much attention for formation of massive black holes and galaxies, but their nature remains a mystery. The LRDs appearing as ``Chimeras" like both active galactic nuclei (AGNs) and galaxies have stimulated renewed interest in the roadmap of central massive black hole (cMBH) formation in AGNs. In this paper, we suggest that the LRDs contain $M_{\bullet}\lesssim 10^6\,M_{\odot}$ cMBHs as demonstrated by the Sołtan argument and there is a large population of stellar-mass black holes (sMBHs with total mass of $\mathscr{M}_{m_{\bullet}}$) embedded inside cMBH accretion disks (cMBH-disk) as motivated by anomalous reverberations of broad H$β$ line in local AGNs. This embryo structure of LRDs ($M_{\bullet}<\mathscr{M}_{m_{\bullet}}$) is formed as a consequence of gravitational collapse of primordial clouds. In this Chimera, accretion onto sMBHs powers the rest-frame optical continuum of the LRDs but the UV continuum is jointly contributed by slim parts of the cMBH-disks and nuclear starbursts in the core of collapsing clouds governing the appearance of the observed V-shaped spectral energy distributions (SEDs). Outflowing clumped-envelopes are unavoidably formed by radiation pressure leading to absorption features of the Balmer lines. The present model works very well for LRDs' SEDs and avoids the issues of overly massive cMBHs. Evolution of LRDs is briefly discussed including gravitational waves.

astro-ph.GA↗

arXiv2025

Detection of unexpected leading delays in broad Hβ line reverberations in the quasar PHL 1092

Delayed reverberations of broad emission lines in response to optical continuum variations have been widely observed in active galactic nuclei (AGNs). They serve as a powerful tool for probing inner structures of AGNs and estimating the masses of supermassive black holes (SMBHs). The delays exhibit a strong correlation with approximately the square root of the optical luminosity - a relationship known as the "standard structure" of AGN broad-line regions (BLRs). Here, we report the discovery of leading delays in Hβ line reverberations (LDRs) in the quasar PHL 1092 preceding variations of the 5100 Å continuum by 17-57 days, based on our eight-year continuous campaign of reverberation mapping of super Eddington AGNs. The LDRs suggest that the 5100 Å continuum regions are so extensive that they are larger than the BLRs. This phenomenon not only fundamentally disrupts the well-established BLR size-luminosity relation but also violates the principle of causality. This unprecedented LDRs challenge the conventional methods for estimating SMBH mass as well as the standard model of AGNs. A preferred scenario to explain the LDRs is that the SMBH-disk contains a population of accreting stellar-mass black holes (sMBHs) as extra heating sources of the disk. Consequently, continuum regions of the disk are efficiently stretched so that the 5100 Å regions exceed the BLRs, yielding the observed LDRs. Generally, sMBH activities there could provide new physics of AGN phenomena, which can be tested by LIGO, LISA/Tianqin and ET detections of gravitational waves from sMBH mergers.

astro-ph.GA↗

arXiv2025

Reg-DPO: SFT-Regularized Direct Preference Optimization with GT-Pair for Improving Video Generation

Recent studies have identified Direct Preference Optimization (DPO) as an efficient and reward-free approach to improving video generation quality. However, existing methods largely follow image-domain paradigms and are mainly developed on small-scale models (approximately 2B parameters), limiting their ability to address the unique challenges of video tasks, such as costly data construction, unstable training, and heavy memory consumption. To overcome these limitations, we introduce a GT-Pair that automatically builds high-quality preference pairs by using real videos as positives and model-generated videos as negatives, eliminating the need for any external annotation. We further present Reg-DPO, which incorporates the SFT loss as a regularization term into the DPO loss to enhance training stability and generation fidelity. Additionally, by combining the FSDP framework with multiple memory optimization techniques, our approach achieves nearly three times higher training capacity than using FSDP alone. Extensive experiments on both I2V and T2V tasks across multiple datasets demonstrate that our method consistently outperforms existing approaches, delivering superior video generation quality.

cs.CV↗

arXiv2025

Scattering Induced Mode Chirality in Ring Resonators

Non-Hermitian physics can be used to break time reversal symmetry and is important for interactions in a wide range of systems, from active matter and neural networks to metamaterials and non-equilibrium thermodynamics. In integrated photonic devices, non-Hermitian physics can be used for direction-dependent light propagation, reconfigurable light paths, selective energy localization and optical isolators. In this work, we report previously unexplored direction-dependent mode splitting in ring microresonators, achieved by adding multiple scatterers around the cavity. Through experiments, simulations, and theoretical modeling, we unveil the underlying physics that changes the resonance shapes in resonant systems with backscattering. By engineering the spatial configuration of the scatterers, we can produce a predictable and repeatable direction-dependent mode splitting, enabling new ways to route light through optical resonators and photonic networks. In addition, the direction dependent mode-splitting can be used for precise near-field measurements, enhancing traditional sensing in integrated photonic chips.

physics.optics↗

arXiv2025

ACT-R: Adaptive Camera Trajectories for Single View 3D Reconstruction

We introduce the simple idea of adaptive view planning to multi-view synthesis, aiming to improve both occlusion revelation and 3D consistency for single-view 3D reconstruction. Instead of producing an unordered set of views independently or simultaneously, we generate a sequence of views, leveraging temporal consistency to enhance 3D coherence. More importantly, our view sequence is not determined by a pre-determined and fixed camera setup. Instead, we compute an adaptive camera trajectory (ACT), forming an orbit, which seeks to maximize the visibility of occluded regions of the 3D object to be reconstructed. Once the best orbit is found, we feed it to a video diffusion model to generate novel views around the orbit, which can then be passed to any multi-view 3D reconstruction model to obtain the final result. Our multi-view synthesis pipeline is quite efficient since it involves no run-time training/optimization, only forward inferences by applying pre-trained models for occlusion analysis and multi-view synthesis. Our method predicts camera trajectories that reveal occlusions effectively and produce consistent novel views, significantly improving 3D reconstruction over SOTA alternatives on the unseen GSO dataset. Project Page: https://mingrui-zhao.github.io/ACT-R/

cs.GR↗

arXiv2025

Audio-Thinker: Guiding Audio Language Model When and How to Think via Reinforcement Learning

Recent advancements in large language models, multimodal large language models, and large audio language models (LALMs) have significantly improved their reasoning capabilities through reinforcement learning with rule-based rewards. However, the explicit reasoning process has yet to show significant benefits for audio question answering, and effectively leveraging deep reasoning remains an open challenge, with LALMs still falling short of human-level auditory-language reasoning. To address these limitations, we propose Audio-Thinker, a reinforcement learning framework designed to enhance the reasoning capabilities of LALMs, with a focus on improving adaptability, consistency, and effectiveness. Our approach introduces an adaptive think accuracy reward, enabling the model to adjust its reasoning strategies based on task complexity dynamically. Furthermore, we incorporate an external reward model to evaluate the overall consistency and quality of the reasoning process, complemented by think-based rewards that help the model distinguish between valid and flawed reasoning paths during training. Experimental results demonstrate that our Audio-Thinker model outperforms existing reasoning-oriented LALMs across various benchmark tasks, exhibiting superior reasoning and generalization capabilities.

cs.SD↗

arXiv2025

Stable Part Diffusion 4D: Multi-View RGB and Kinematic Parts Video Generation

We present Stable Part Diffusion 4D (SP4D), a framework for generating paired RGB and kinematic part videos from monocular inputs. Unlike conventional part segmentation methods that rely on appearance-based semantic cues, SP4D learns to produce kinematic parts - structural components aligned with object articulation and consistent across views and time. SP4D adopts a dual-branch diffusion model that jointly synthesizes RGB frames and corresponding part segmentation maps. To simplify the architecture and flexibly enable different part counts, we introduce a spatial color encoding scheme that maps part masks to continuous RGB-like images. This encoding allows the segmentation branch to share the latent VAE from the RGB branch, while enabling part segmentation to be recovered via straightforward post-processing. A Bidirectional Diffusion Fusion (BiDiFuse) module enhances cross-branch consistency, supported by a contrastive part consistency loss to promote spatial and temporal alignment of part predictions. We demonstrate that the generated 2D part maps can be lifted to 3D to derive skeletal structures and harmonic skinning weights with few manual adjustments. To train and evaluate SP4D, we construct KinematicParts20K, a curated dataset of over 20K rigged objects selected and processed from Objaverse XL (Deitke et al., 2023), each paired with multi-view RGB and part video sequences. Experiments show that SP4D generalizes strongly to diverse scenarios, including real-world videos, novel generated objects, and rare articulated poses, producing kinematic-aware outputs suitable for downstream animation and motion-related tasks.

cs.CV↗

arXiv2025

FlexQ: Efficient Post-training INT6 Quantization for LLM Serving via Algorithm-System Co-Design

Large Language Models (LLMs) demonstrate exceptional performance but entail significant memory and computational costs, restricting their practical deployment. While existing INT4/INT8 quantization reduces these costs, they often degrade accuracy or lack optimal efficiency. INT6 quantization offers a superior trade-off between model accuracy and inference efficiency, but lacks hardware support in modern GPUs, forcing emulation via higher-precision arithmetic units that limit acceleration. In this paper, we propose FlexQ, a novel post-training INT6 quantization framework combining algorithmic innovation with system-level optimizations. FlexQ employs uniform 6-bit weight quantization across all layers, with adaptive retention of 8-bit activations in layers identified through layer-wise sensitivity analysis. To maximize hardware efficiency, we develop a specialized high-performance GPU kernel supporting matrix multiplication for W6A6 and W6A8 representations via Binary Tensor Core (BTC) equivalents, effectively bypassing the lack of native INT6 tensor cores. Evaluations on LLaMA family models show FlexQ maintains near-FP16 accuracy, with perplexity increases of no more than 0.1 on WikiText2. The proposed kernel achieves an average 1.39$\times$ speedup over ABQ-LLM on LLaMA-2-70B linear layers. End-to-end, FlexQ delivers 1.33$\times$ inference acceleration and 1.21$\times$ memory savings over SmoothQuant. Code is released at https://github.com/FlyFoxPlayer/FlexQ.

cs.LG↗

arXiv2025

VSA: Faster Video Diffusion with Trainable Sparse Attention

Scaling video diffusion transformers (DiTs) is limited by their quadratic 3D attention, even though most of the attention mass concentrates on a small subset of positions. We turn this observation into VSA, a trainable, hardware-efficient sparse attention that replaces full attention at \emph{both} training and inference. In VSA, a lightweight coarse stage pools tokens into tiles and identifies high-weight \emph{critical tokens}; a fine stage computes token-level attention only inside those tiles subjecting to block computing layout to ensure hard efficiency. This leads to a single differentiable kernel that trains end-to-end, requires no post-hoc profiling, and sustains 85\% of FlashAttention3 MFU. We perform a large sweep of ablation studies and scaling-law experiments by pretraining DiTs from 60M to 1.4B parameters. VSA reaches a Pareto point that cuts training FLOPS by 2.53$\times$ with no drop in diffusion loss. Retrofitting the open-source Wan-2.1 model speeds up attention time by 6$\times$ and lowers end-to-end generation time from 31s to 18s with comparable quality. These results establish trainable sparse attention as a practical alternative to full attention and a key enabler for further scaling of video diffusion models. Code will be available at https://github.com/hao-ai-lab/FastVideo.

cs.CV↗

arXiv2025

MoPFormer: Motion-Primitive Transformer for Wearable-Sensor Activity Recognition

Human Activity Recognition (HAR) with wearable sensors is challenged by limited interpretability, which significantly impacts cross-dataset generalization. To address this challenge, we propose Motion-Primitive Transformer (MoPFormer), a novel self-supervised framework that enhances interpretability by tokenizing inertial measurement unit signals into semantically meaningful motion primitives and leverages a Transformer architecture to learn rich temporal representations. MoPFormer comprises two stages. The first stage is to partition multi-channel sensor streams into short segments and quantize them into discrete ``motion primitive'' codewords, while the second stage enriches those tokenized sequences through a context-aware embedding module and then processes them with a Transformer encoder. The proposed MoPFormer can be pre-trained using a masked motion-modeling objective that reconstructs missing primitives, enabling it to develop robust representations across diverse sensor configurations. Experiments on six HAR benchmarks demonstrate that MoPFormer not only outperforms state-of-the-art methods but also successfully generalizes across multiple datasets. More importantly, the learned motion primitives significantly enhance both interpretability and cross-dataset performance by capturing fundamental movement patterns that remain consistent across similar activities, regardless of dataset origin.

cs.CV↗

arXiv2025

Local Learning for Covariate Selection in Nonparametric Causal Effect Estimation with Latent Variables

Estimating causal effects from nonexperimental data is a fundamental problem in many fields of science. A key component of this task is selecting an appropriate set of covariates for confounding adjustment to avoid bias. Most existing methods for covariate selection often assume the absence of latent variables and rely on learning the global network structure among variables. However, identifying the global structure can be unnecessary and inefficient, especially when our primary interest lies in estimating the effect of a treatment variable on an outcome variable. To address this limitation, we propose a novel local learning approach for covariate selection in nonparametric causal effect estimation, which accounts for the presence of latent variables. Our approach leverages testable independence and dependence relationships among observed variables to identify a valid adjustment set for a target causal relationship, ensuring both soundness and completeness under standard assumptions. We validate the effectiveness of our algorithm through extensive experiments on both synthetic and real-world data.

cs.LG↗

arXiv2025

Antisymmetric chirp transfer to high-energy ultraviolet pulses via gas-based chirped four-wave mixing

Spectro-temporal shaping of high-power femtosecond ultraviolet (UV) pulses remains a key challenge in ultrafast optics. Tailoring high-energy, ultrashort UV pulses underpins applications in ultrafast dynamics, high-precision spectroscopy, strong-field physics, charged-particle radiation sources, and industrial microfabrication. However, the transmission and damage threshold of the dielectric optics limit programmable shaping for high-power UV pulses. Towards overcoming this challenge, we demonstrate an antisymmetric dispersion transfer from near-infrared (NIR) pulses to UV pulses via chirped four-wave mixing (CFWM) in noble gas, whereby negatively chirped NIR pulses map quasi-linearly to positively chirped UV in a gas-filled hollow capillary fiber (HCF) at the expense of a moderate conversion efficiency of 13%. This antisymmetric chirp transfer approach broadens the basis for tailoring UV pulses by leveraging the high damage tolerance, large bandwidth acceptance, and intrinsically low dispersion of noble gases, rather than relying on conventional nonlinear crystals.

physics.optics↗

arXiv2025

SAMA: Towards Multi-Turn Referential Grounded Video Chat with Large Language Models

Achieving fine-grained spatio-temporal understanding in videos remains a major challenge for current Video Large Multimodal Models (Video LMMs). Addressing this challenge requires mastering two core capabilities: video referring understanding, which captures the semantics of video regions, and video grounding, which segments object regions based on natural language descriptions. However, most existing approaches tackle these tasks in isolation, limiting progress toward unified, referentially grounded video interaction. We identify a key bottleneck in the lack of high-quality, unified video instruction data and a comprehensive benchmark for evaluating referentially grounded video chat. To address these challenges, we contribute in three core aspects: dataset, model, and benchmark. First, we introduce SAMA-239K, a large-scale dataset comprising 15K videos specifically curated to enable joint learning of video referring understanding, grounding, and multi-turn video chat. Second, we propose the SAMA model, which incorporates a versatile spatio-temporal context aggregator and a Segment Anything Model to jointly enhance fine-grained video comprehension and precise grounding capabilities. Finally, we establish SAMA-Bench, a meticulously designed benchmark consisting of 5,067 questions from 522 videos, to comprehensively evaluate the integrated capabilities of Video LMMs in multi-turn, spatio-temporal referring understanding and grounded dialogue. Extensive experiments and benchmarking results show that SAMA not only achieves strong performance on SAMA-Bench but also sets a new state-of-the-art on general grounding benchmarks, while maintaining highly competitive performance on standard visual understanding benchmarks.

cs.CV↗

arXiv2025

Real Deep Research for AI, Robotics and Beyond

With the rapid growth of research in AI and robotics now producing over 10,000 papers annually it has become increasingly difficult for researchers to stay up to date. Fast evolving trends, the rise of interdisciplinary work, and the need to explore domains beyond one's expertise all contribute to this challenge. To address these issues, we propose a generalizable pipeline capable of systematically analyzing any research area: identifying emerging trends, uncovering cross domain opportunities, and offering concrete starting points for new inquiry. In this work, we present Real Deep Research (RDR) a comprehensive framework applied to the domains of AI and robotics, with a particular focus on foundation models and robotics advancements. We also briefly extend our analysis to other areas of science. The main paper details the construction of the RDR pipeline, while the appendix provides extensive results across each analyzed topic. We hope this work sheds light for researchers working in the field of AI and beyond.

cs.AI↗