arXiv subjects

Qi Sun

Publications and source records attributed to Qi Sun.

At least 73 records · Page 4Linked to original sources

arXiv2026

Atomic Information Flow: A Network Flow Model for Tool Attributions in RAG Systems

Many tool-based Retrieval Augmented Generation (RAG) systems lack precise mechanisms for tracing final responses back to specific tool components -- a critical gap as systems scale to complex multi-agent architectures. We present \textbf{Atomic Information Flow (AIF)}, a graph-based network flow model that decomposes tool outputs and LLM calls into atoms: indivisible, self-contained units of information. By modeling LLM orchestration as a directed flow of atoms from tool and LLM nodes to a response super-sink, AIF enables granular attribution metrics for AI explainability. Motivated by the max-flow min-cut theorem in network flow theory, we train a lightweight Gemma3 (4B parameter) language model as a context compressor to approximate the minimum cut of tool atoms using flow signals computed offline by AIF. We note that the base Gemma3-4B model struggles to identify critical information with \textbf{54.7\%} accuracy on HotpotQA, barely outperforming lexical baselines (BM25). However, post-training on AIF signals boosts accuracy to \textbf{82.71\%} (+28.01 points) while achieving \textbf{87.52\%} (+1.85\%) context token compression -- bridging the gap with the Gemma3-27B variant, a model nearly $7\times$ larger.

cs.IR↗

arXiv2026

Orchestrating Specialized Agents for Trustworthy Enterprise RAG

Retrieval-Augmented Generation (RAG) shows promise for enterprise knowledge work, yet it often underperforms in high-stakes decision settings that require deep synthesis, strict traceability, and recovery from underspecified prompts. One-pass retrieval-and-write pipelines frequently yield shallow summaries, inconsistent grounding, and weak mechanisms for completeness verification. We introduce ADORE (Adaptive Deep Orchestration for Research in Enterprise), an agentic framework that replaces linear retrieval with iterative, user-steered investigation coordinated by a central orchestrator and a set of specialized agents. ADORE's key insight is that a structured Memory Bank (a curated evidence store with explicit claim-evidence linkage and section-level admissible evidence) enables traceable report generation and systematic checks for evidence completeness. Our contributions are threefold: (1) Memory-locked synthesis - report generation is constrained to a structured Memory Bank (Claim-Evidence Graph) with section-level admissible evidence, enabling traceable claims and grounded citations; (2) Evidence-coverage-guided execution - a retrieval-reflection loop audits section-level evidence coverage to trigger targeted follow-up retrieval and terminates via an evidence-driven stopping criterion; (3) Section-packed long-context grounding - section-level packing, pruning, and citation-preserving compression make long-form synthesis feasible under context limits. Across our evaluation suite, ADORE ranks first on DeepResearch Bench (52.65) and achieves the highest head-to-head preference win rate on DeepConsult (77.2%) against commercial systems.

cs.IR↗

arXiv2026

Chemistry in a Cryogenic Buffer Gas Cell

Cryogenic buffer gas sources are ubiquitous for producing cold, collimated molecular beams for quantum science, chemistry, and precision measurements. The molecules are typically produced by laser ablating a metal target in the presence of a donor gas. The radical of interest emerges due to a barrier-free reaction or under thermal or optical excitation. High-barrier reactions, such as between Ca and H$_2$, should be precluded. We study chemical reactions between Ca and three hydrogen isotopologues H$_2$, D$_2$, and HD in a cryogenic cell with helium buffer gas. We observe that H$_2$ can serve as both a reactant and a buffer gas, outperforming D$_2$ and HD. We use a reaction network model to describe the chemical dynamics and find that the enhanced molecular yield can be attributed to rapid vibrational excitations of the reactant gas. Our results demonstrate a robust method for generating bright cold beams of alkaline-earth-metal hydrides for laser cooling and trapping.

physics.atom-ph↗

arXiv2026

Neural Green's Function Accelerated Iterative Methods for Solving Indefinite Boundary Value Problems

Neural operators, which learn mappings between the function spaces, have been applied to solve boundary value problems in various ways, including learning mappings from the space of the forcing terms to the space of the solutions with the substantial requirements of data pairs. In this work, we present a data-free neural operator integrated with physics, which learns the Green kernel directly. Our method proceeds in three steps: 1. The governing equations for the Green's function are reformulated into an interface problem, where the delta Dirac function is removed; 2. The interface problem is embedded in a lifted space of higher-dimension to handle the jump in the derivative, but still solved on a two-dimensional surface without additional sampling cost; 3. Deep neural networks are employed to address the curse of dimensionality caused by this lifting operation. The approximate Green's function obtained through our approach is then used to construct preconditioners for the linear systems allowed by its mathematical properties. Furthermore, the spectral bias of it revealed through both theoretical analysis and numerical validation contrasts with the smoothing effects of traditional iterative solvers, which motivates us to propose a hybrid iterative method that combines these two solvers. Numerical experiments demonstrate the effectiveness of our approximate Green's function in accelerating iterative methods, proving fast convergence for solving indefinite problems even involving discontinuous coefficients.

math.NA↗

arXiv2025

Generalized Three-Family Supersymmetric Pati-Salam Models from Type IIA Intersecting D6-Branes

Generalizing three-family chiral fermion conditions to $I_{ac}=-(3+h)$ and $I_{ac'}=h$, with positive integer $h$, we extend the landscape of three-family ${\cal N}=1$ supersymmetric Pati-Salam models in a broader region. Differing from the former investigation with $I_{ac}=-3$ and $I_{ac'}=0$, we do not restrict that the $a$ stack of D6-branes must be parallel to the orientifold image of the $c$-stack along one of the three two-tori. In this investigation, without the simple parallel construction, we find four new classes of supersymmetric Pati-Salam models that are allowed by the extended three generation condition with $I_{ac}=3, I_{ac'}=-6$ and $I_{ac}=-1, I_{ac'}=-2$ through the intersections of $a$- and $c/c'$-branes. Moreover, with the $SU(2)_{L'}$ gauge coupling realized from $SU(2)_{L_1}\times SU(2)_{L_2}$ symmetry breaking, the canonical normalization requirement of the gauge kinetic term provides an alternative approach that can be imposed before the renormalization group equation evolution for $SU(2)_{L'}$ gauge coupling. This turns out to be an effective mechanism to realize the string-scale gauge coupling relation, especially for the new supersymmetric Pati-Salam models with large $g_b/g_a$ ratio. We show that this symmetry-breaking modified renormalization group evolution can highly suppress $g_b/g_a$, and finally realizes string-scale gauge coupling relations for the extended supersymmetric Pati-Salam models as well.

hep-th↗

arXiv2025

Animus3D: Text-driven 3D Animation via Motion Score Distillation

We present Animus3D, a text-driven 3D animation framework that generates motion field given a static 3D asset and text prompt. Previous methods mostly leverage the vanilla Score Distillation Sampling (SDS) objective to distill motion from pretrained text-to-video diffusion, leading to animations with minimal movement or noticeable jitter. To address this, our approach introduces a novel SDS alternative, Motion Score Distillation (MSD). Specifically, we introduce a LoRA-enhanced video diffusion model that defines a static source distribution rather than pure noise as in SDS, while another inversion-based noise estimation technique ensures appearance preservation when guiding motion. To further improve motion fidelity, we incorporate explicit temporal and spatial regularization terms that mitigate geometric distortions across time and space. Additionally, we propose a motion refinement module to upscale the temporal resolution and enhance fine-grained details, overcoming the fixed-resolution constraints of the underlying video model. Extensive experiments demonstrate that Animus3D successfully animates static 3D assets from diverse text prompts, generating significantly more substantial and detailed motion than state-of-the-art baselines while maintaining high visual integrity. Code will be released at https://qiisun.github.io/animus3d_page.

cs.CV↗

arXiv2025

Foveation Improves Payload Capacity in Steganography

Steganography finds its use in visual medium such as providing metadata and watermarking. With support of efficient latent representations and foveated rendering, we trained models that improve existing capacity limits from 100 to 500 bits, while achieving better accuracy of up to 1 failure bit out of 2000, at 200K test bits. Finally, we achieve a comparable visual quality of 31.47 dB PSNR and 0.13 LPIPS, showing the effectiveness of novel perceptual design in creating multi-modal latent representations in steganography.

cs.CV↗

arXiv2025

The Bichromatic Two-Center Problem on Graphs

In this paper, we study the (weighted) bichromatic two-center problem on graphs. The input consists of a graph $G$ of $n$ (weighted) vertices and $m$ edges, and a set $\mathcal{P}$ of pairs of distinct vertices, where no vertex appears in more than one pair. The problem aims to find two points (i.e., centers) on $G$ by assigning vertices of each pair to different centers so as to minimize the maximum (weighted) distance of vertices to their assigned centers (so that the graph can be bi-colored with this goal). To the best of our knowledge, this problem has not been studied on graphs, including tree graphs. In this paper, we propose an $O(m^2n\log n\log mn)$ algorithm for solving the problem on an undirected graph provided with the distance matrix, an $O(n\log n)$-time algorithm for the problem on trees, and a linear-time approach for the unweighted tree version.

cs.DS↗

arXiv2025

Unitho: A Unified Multi-Task Framework for Computational Lithography

Reliable, generalizable data foundations are critical for enabling large-scale models in computational lithography. However, essential tasks-mask generation, rule violation detection, and layout optimization-are often handled in isolation, hindered by scarce datasets and limited modeling approaches. To address these challenges, we introduce Unitho, a unified multi-task large vision model built upon the Transformer architecture. Trained on a large-scale industrial lithography simulation dataset with hundreds of thousands of cases, Unitho supports end-to-end mask generation, lithography simulation, and rule violation detection. By enabling agile and high-fidelity lithography simulation, Unitho further facilitates the construction of robust data foundations for intelligent EDA. Experimental results validate its effectiveness and generalizability, with performance substantially surpassing academic baselines.

cs.LG↗

arXiv2025

Improving the robustness of the immersed interface method through regularized velocity reconstruction

Robust, broadly applicable fluid-structure interaction (FSI) algorithms remain a challenge for computational mechanics. In previous work, we introduced an immersed interface method (IIM) for discrete surfaces and an extension based on an immersed Lagrangia-Eulerian (ILE) coupling strategy for modeling FSI involving complex geometries. The ability of the method to sharply resolve stress discontinuities induced by singular immersed boundary forces in the presence of low-regularity geometrical representations makes it a compelling choice for three-dimensional modeling of complex geometries in diverse engineering applications. Although the IIM we previously introduced offers many desirable advantages, it also imposes a restrictive mesh factor ratio, requiring the surface mesh to be coarser than the fluid grid to ensure stability. This is because when the mesh factor ratio constraint is not satisfied, parts of the structure motion are not controlled by the discrete FSI system. This constraint can significantly increase computational costs, particularly in applications involving multiscale geometries with highly localized complexity or fine-scale features. To address this limitation, we devise a stabilization strategy for the velocity restriction operator inspired by Tikhonov regularization. This study demonstrates that using a stabilized velocity restriction operator in IIM enables a broader range of structure-to-fluid grid-size ratios without compromising accuracy or altering the flow dynamics. This advancement significantly broadens the applicability of the method to real-world FSI problems involving complex geometries and dynamic conditions, offering a robust and practical solution.

physics.comp-ph↗

arXiv2025

OregairuChar: A Benchmark Dataset for Character Appearance Frequency Analysis in My Teen Romantic Comedy SNAFU

The analysis of character appearance frequency is essential for understanding narrative structure, character prominence, and story progression in anime. In this work, we introduce OregairuChar, a benchmark dataset designed for appearance frequency analysis in the anime series My Teen Romantic Comedy SNAFU. The dataset comprises 1600 manually selected frames from the third season, annotated with 2860 bounding boxes across 11 main characters. OregairuChar captures diverse visual challenges, including occlusion, pose variation, and inter-character similarity, providing a realistic basis for appearance-based studies. To enable quantitative research, we benchmark several object detection models on the dataset and leverage their predictions for fine-grained, episode-level analysis of character presence over time. This approach reveals patterns of character prominence and their evolution within the narrative. By emphasizing appearance frequency, OregairuChar serves as a valuable resource for exploring computational narrative dynamics and character-centric storytelling in stylized media.

cs.CV↗

arXiv2025

Object-Centric 3D Gaussian Splatting for Strawberry Plant Reconstruction and Phenotyping

Strawberries are among the most economically significant fruits in the United States, generating over $2 billion in annual farm-gate sales and accounting for approximately 13% of the total fruit production value. Plant phenotyping plays a vital role in selecting superior cultivars by characterizing plant traits such as morphology, canopy structure, and growth dynamics. However, traditional plant phenotyping methods are time-consuming, labor-intensive, and often destructive. Recently, neural rendering techniques, notably Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS), have emerged as powerful frameworks for high-fidelity 3D reconstruction. By capturing a sequence of multi-view images or videos around a target plant, these methods enable non-destructive reconstruction of complex plant architectures. Despite their promise, most current applications of 3DGS in agricultural domains reconstruct the entire scene, including background elements, which introduces noise, increases computational costs, and complicates downstream trait analysis. To address this limitation, we propose a novel object-centric 3D reconstruction framework incorporating a preprocessing pipeline that leverages the Segment Anything Model v2 (SAM-2) and alpha channel background masking to achieve clean strawberry plant reconstructions. This approach produces more accurate geometric representations while substantially reducing computational time. With a background-free reconstruction, our algorithm can automatically estimate important plant traits, such as plant height and canopy width, using DBSCAN clustering and Principal Component Analysis (PCA). Experimental results show that our method outperforms conventional pipelines in both accuracy and efficiency, offering a scalable and non-destructive solution for strawberry plant phenotyping.

cs.CV↗

arXiv2025

From Grounding to Manipulation: Case Studies of Foundation Model Integration in Embodied Robotic Systems

Foundation models (FMs) are increasingly used to bridge language and action in embodied agents, yet the operational characteristics of different FM integration strategies remain under-explored -- particularly for complex instruction following and versatile action generation in changing environments. This paper examines three paradigms for building robotic systems: end-to-end vision-language-action (VLA) models that implicitly integrate perception and planning, and modular pipelines incorporating either vision-language models (VLMs) or multimodal large language models (LLMs). We evaluate these paradigms through two focused case studies: a complex instruction grounding task assessing fine-grained instruction understanding and cross-modal disambiguation, and an object manipulation task targeting skill transfer via VLA finetuning. Our experiments in zero-shot and few-shot settings reveal trade-offs in generalization and data efficiency. By exploring performance limits, we distill design implications for developing language-driven physical agents and outline emerging challenges and opportunities for FM-powered robotics in real-world conditions.

cs.RO↗

arXiv2025

Neutron capture measurement of the 165Ho at the CSNS Backn facility in the resonance energy region

The neutron capture yield of 165Ho have been measured at the Back-streaming White neutron beam line (Back-n) of the China Spallation Neutron Source (CSNS) using a 4π BaF2 Gamma Total Absorption Facility (GTAF). The resonance shapes in the 1eV to 1.0keV region were analyzed with the Bayesian R-matrix code SAMMY. For 18 s-wave resonances below 100eV, the resonance energy ER, neutron width Γn, and radiative width Γγ were extracted. The statistical analyses of the resonance parameters show that the nearest-neighbour level-spacing distribution follows a Wigner-Dyson form with mean spacing D0 = 4.53(3)eV,indicating chaotic compound-nucleus behaviour; Using the extracted parameters, the s-wave neutron strength function for 165Ho was derived to be 10-4S0 = 2.01(1), in excellent agreement with the values reported in both the Atlas of Neutron Resonances and ENDF/B-VIII.0 data.

nucl-ex↗

arXiv2025

Learning Function-to-Function Mappings: A Fourier Neural Operator for Next-Generation MIMO Systems

Next-generation multiple-input multiple-output (MIMO) systems, characterized by extremely large-scale arrays, holographic surfaces, three-dimensional architectures, and flexible antennas, are poised to deliver unprecedented data rates, spectral efficiency and stability. However, these advancements introduce significant challenges for physical layer signal processing, stemming from complex near-field propagation, continuous aperture modeling, sub-wavelength antenna coupling effects, and dynamic channel conditions. Conventional model-based and deep learning approaches often struggle with the immense computational complexity and model inaccuracies inherent in these new regimes. This article proposes a Fourier neural operator (FNO) as a powerful and promising tool to address these challenges. The FNO learns function-to-function mappings between infinite-dimensional function spaces, making them exceptionally well-suited for modeling complex physical systems governed by partial differential equations based on electromagnetic wave propagation. We first present the fundamental principles of FNO, demonstrating its mesh-free nature and function-to-function ability to efficiently capture global dependencies in the Fourier domain. Furthermore, we explore a range of applications of FNO in physical-layer signal processing for next-generation MIMO systems. Representative case studies on channel modeling and estimation for novel MIMO architectures demonstrate the superior performance of FNO compared to state-of-the-art methods. Finally, we discuss open challenges and outline future research directions, positioning FNO as a promising technology for enabling the enormous potential of next-generation MIMO systems.

cs.IT↗

arXiv2025

Huisken's Distance Comparison Principle in Higher Codimension

We establish a variant of Huisken's distance comparison principle for reflection symmetric immersed Curve Shortening flow in $\mathbb R^n,n\geq2$. As an application, we show that certain symmetric Curve Shortening flow with a one-to-one convex projection develops a Type~I singularity and becomes asymptotically circular.

math.DG↗

arXiv2025

Towards Robust Agentic CUDA Kernel Benchmarking, Verification, and Optimization

Recent advances in large language models (LLMs) demonstrate their effectiveness in scaling test-time compute for software engineering tasks. However, these approaches often focus on high-level solutions, with limited attention to optimizing low-level CUDA kernel implementations. Additionally, existing kernel generation benchmarks suffer from exploitable loopholes and insufficient diversity in testing conditions, hindering true generalization assessment. To address these limitations, we introduce robust-kbench, a new benchmark for rigorous evaluation of kernel performance and correctness across varied scenarios. Furthermore, we present a comprehensive agentic framework that automates CUDA kernel discovery, verification, and optimization. This pipeline enables frontier LLMs to translate torch code to CUDA kernels and iteratively improve their runtime within our robust evaluation setting. Our sequential workflow first translates PyTorch code into equivalent CUDA kernels. It then optimizes their runtime using a novel evolutionary meta-generation procedure tailored to the CUDA ecosystem, guided by LLM-based verifiers for correctness and efficient filtering. Evaluated on robust-kbench, our approach produces CUDA kernels outperforming torch implementations for practical applications, including forward and backward passes. It can fuse operations and deploy various runtime optimization strategies. The verifier workflow accurately classifies incorrect kernels, enhancing hardware verification efficiency.

cs.SE↗

arXiv2025

Learning Singularity-Encoded Green's Functions with Application to Iterative Methods

Green's function provides an inherent connection between theoretical analysis and numerical methods for elliptic partial differential equations, and general absence of its closed-form expression necessitates surrogate modeling to guide the design of effective solvers. Unfortunately, numerical computation of Green's function remains challenging due to its doubled dimensionality and intrinsic singularity. In this paper, we present a novel singularity-encoded learning approach to resolve these problems in an unsupervised fashion. Our method embeds the Green's function within a one-order higher-dimensional space by encoding its prior estimate as an augmented variable, followed by a neural network parametrization to manage the increased dimensionality. By projecting the trained neural network solution back onto the original domain, our deep surrogate model exploits its spectral bias to accelerate conventional iterative schemes, serving either as a preconditioner or as part of a hybrid solver. The effectiveness of our proposed method is empirically verified through numerical experiments with two and four dimensional Green's functions, achieving satisfactory resolution of singularities and acceleration of iterative solvers.

math.NA↗