arXiv subjects

Min Li

Publications and source records attributed to Min Li.

At least 19 recordsLinked to original sources

arXiv2026

Behind Video Generation: Observations and Insights on Multimodal Understanding from Kling

Video generation has progressed from isolated clips to coherent short-form audiovisual sequences, and the bottleneck is shifting from rendering to semantic control. This report summarizes the observations and insights on multimodal understanding gained from successive generations of Kling. We argue that generation quality is the projection of multimodal understanding through the entire data--conditioning--alignment--evaluation chain, and we examine this chain module by module. Kling-MLLM, a shared multimodal foundation model that unifies image, video, audio, synchronized audio--video, and interleaved inputs, is specialized for each downstream module. Because a DiT is trained on far fewer training samples than a language model, we treat captioning as a semantic contract that maximizes description density, and extend it to reference-grounded captioning that makes reference bindings explicit. Prompt enhancement translates user intent into this caption distribution: basic PE is derived from captions of compelling videos and refined with text-level and generator-in-the-loop reinforcement learning, while agentic PE refines the story and retrieves compelling examples. We show that a reliable text encoder comes from perceptual training rather than from stronger language ability, and that candidates can be screened from text alone before end-to-end training. Reward modeling pairs a general reward model for explainable, localized critiques with a decision reward model for fast multidimensional scoring, treating perceptual bandwidth as a first-class design parameter. Finally, we review task-specific automatic evaluation, focusing on protocols that expose scoring shortcuts and help trace failures to their upstream source. We hope these lessons offer practical guidance for building the understanding stack behind controllable video generation.

cs.CV↗

arXiv2026

BoundedGS: Memory-Efficient High-Resolution Training for 3D Gaussian Splatting

High-resolution 3D Gaussian Splatting training increases pixel and Gaussian-tile intersection counts, expanding rasterization caches that coexist with projection and gradient state at peak memory. A complete-view update aggregates gradients across the view; each region's forward and backward caches can be released after its contribution is collected. BoundedGS separates rasterization and projection backward passes and bounds the active local rasterization working set according to its measured load. Candidate-indexed projection and compact gradient storage address the view-level peaks exposed by this change, while CPU-assisted structure rebuilding and bounded image supply manage resources outside rendering. In a resolution sweep at a fixed count of about 6M Gaussians, the measured rendering working set stays within 1.073-1.077 GiB from 1K to 16K; at 11K, a warm pass takes 19.79 seconds versus 21.76 for full-frame execution. Across five native 5K outdoor scenes, full training reduces the mean rendering working set from 9.97 to 1.76 GiB (82.3%), retains similar reconstruction quality, and shortens training time from 36.7 to 33.9 minutes. Project page: https://xiaobin2001.github.io/BoundedGS-Web/

cs.CV↗

arXiv2026

ParetoGS: High-Quality Continuous-Budget 3D Gaussian Splatting with Sparse Adapted Anchors

Scalable 3D Gaussian Splatting allows one scene representation to serve different Gaussian-count budgets, yet its quality-budget curve may lag behind models optimized separately for each budget. Reducing the number of Gaussians changes both the selected members and the reconstruction tasks carried by the survivors; parameters inherited from a high-budget model may fit a low-budget representation poorly. We propose ParetoGS, which describes member changes and parameter evolution with a small set of adapted budget anchors and decodes intermediate budgets without separate optimization at every requested count. ParetoGS alternates pruning and fixed-member adaptation to build nested anchors with persistent identities, and encodes their parameter changes relative to the full model in a compact residual atlas. At decoding, it interpolates the states of shared members and gradually introduces the rendering contributions of new members, connecting sparse anchors into a continuous-budget representation. Across 13 scenes, ParetoGS attains better average PSNR, SSIM, and LPIPS with fewer Gaussians than the compared prefix methods at 10% of each method's own full model. From a shared full model, it also outperforms FlexGS-imp at its trained 1%, 5%, and 10% budgets and NanoGS-imp at the 1% and 0.1% budgets. On Garden, its non-anchor quality approaches an empirical reference from independently optimized GNS models at higher retention ratios and exceeds it at some lower-budget points. Across 13 scenes, encoding reduces mean additional state from 390.0 to 85.7 MB with an average PSNR change of 0.01 dB. Project page: https://xiaobin2001.github.io/ParetoGS-Web/

cs.CV↗

arXiv2026

LIDL: LLM Integration Defect Localization via Knowledge Graph-Enhanced Multi-Agent Analysis

LLM-integrated software, which embeds or interacts with large language models (LLMs) as functional components, exhibits probabilistic and context-dependent behaviors that fundamentally differ from those of traditional software. This shift introduces a new category of integration defects that arise not only from code errors but also from misaligned interactions among LLM-specific artifacts, including prompts, API calls, configurations, and model outputs. However, existing defect localization techniques are ineffective at identifying these LLM-specific integration defects because they fail to capture cross-layer dependencies across heterogeneous artifacts, cannot exploit incomplete or misleading error traces, and lack semantic reasoning capabilities for identifying root causes. To address these challenges, we propose LIDL, a multi-agent framework for defect localization in LLM-integrated software. LIDL (1) constructs a code knowledge graph enriched with LLM-related annotations that represent interaction boundaries across source code, prompts, and configuration files, (2) fuses complementary sources of error evidence to surface candidate defect locations, and (3) applies context-aware validation that uses counterfactual reasoning to distinguish true root causes from propagated symptoms. We evaluate LIDL on 146 real-world defect instances collected from 105 GitHub repositories and 16 agent-based systems. The results show that LIDL significantly outperforms seven state-of-the-art baselines in localization accuracy, improving Top-3 accuracy by 19.15%, 21.69%, and 21.65% at the file, function, and line levels. Notably, LIDL achieves these gains at a lower cost than most baselines. Our localization results also improve downstream repair success by 30.77% over the best-performing baselines.

cs.SE↗

arXiv2026

k-Submodular Cover Problem with a Matroid Constraint

We investigate k-submodular cover problem with a matroid constraint on an n-element ground set V , which seeks k disjoint subsets of V with minimal cost, where their union is an independent set for some matroid defined on V and the k-submodular utility function surpasses a specified threshold. This problem generalizes the k-submodular cover problem and has a wide range of applications in artificial intelligence and combinatorial optimization. In this paper, we propose a matroid extended greedy algorithm for this problem, analyze its bi-criteria approximation ratio, and further derive an improved approximation ratio under the special case of unit costs.

cs.GT↗

arXiv2026

Transdimensional Chern insulators in rhombohedral multilayer graphene superlattices

Recent experiments on lightly electron-doped rhombohedral multilayer graphene (RMG) in the "transdimensional" regime revealed anomalous Hall hysteresis under both perpendicular and in-plane magnetic fields, pointing to the emergence of transdimensional orbital magnetism in an interaction-driven "Fermi-lune" metal. Here we demonstrate that, by coupling to the moiré potential, this transdimensional orbital ferromagnetic metal can evolve into quantized Chern insulator phases in twisted RMG/hBN moiré superlattices. Using unrestricted Hartree-Fock calculations, we find that interaction-driven reconstruction isolates a spin-valley-polarized $|C| = 1$ Chern band, leading to a "transdimensional Chern insulator" (TDCI) state exhibiting both in-plane and out-of-plane orbital magnetizations. A minimal two-band model is introduced to describe the unique properties of TDCI, which unveils the intimate connection between non-reciprocal velocity and in-plane orbital magnetization. We further analyze the collective excitations of TDCIs, and identify soft charge and spin modes that may lead to density wave instabilities in certain cases. An in-plane magnetic field can harden the soft density-wave mode and may drive a transition from density-wave state to TDCI. Finally, when the $|C| = 1$ band is fractionally filled, the many-body ground state may be a fractional Chern insulator with broken rotation and mirror symmetries.

cond-mat.mes-hall↗

arXiv2026

Beam Spill Structure Measurements at CEE Using the T0 Detector

The Cooling Storage Ring External-target Experiment (CEE) at the Heavy Ion Research Facility in Lanzhou uses a slow-extracted heavy-ion beam whose temporal profile directly affects detector occupancy. The existing CEE T0 detector, originally designed to provide the event start time, is extended in this work to measure the spill structure immediately upstream of the fixed target. The detector consists of a 0.8-mm-thick EJ-200 plastic scintillator read out by 16 silicon photomultipliers at each end. The SiPM signals are conditioned and discriminated by the NINO front-end electronics, and the resulting timing signals are recorded using an FPGA-based time-to-digital converter. In laboratory tests, with the laser trigger set to its maximum available repetition frequency of 100~MHz, the recorded T0 hit-rate distribution was concentrated around 100~MHz. During the June 2024 beam test with 350-MeV/u Kr ions, the accelerator team varied several operating settings, including RF-KO, sextupole, and feedback settings, while T0 independently recorded the corresponding spill profiles. An empirical representation was used to extract the rise, plateau, and fall durations from the stored profiles. The plateau duration was used as the primary comparison metric while accelerator stability and RF-system load were also considered. The measurements clearly distinguished the spill-profile responses associated with different accelerator settings. The spill obtained with the selected settings exhibited a more nearly rectangular envelope than the initial profile, although residual fluctuations and a terminal spike remained. These results demonstrate that the existing CEE T0 detector can provide independent beam-structure information during accelerator tuning.

physics.ins-det↗

arXiv2026

Fermi lune and non-reciprocal transport in rhombohedral multilayer graphene

We develop a microscopic theory of Fermi-lune metals in rhombohedral multilayer graphene and investigate their non-reciprocal longitudinal transport. We show that tuning the layer number and displacement field drives Lifshitz transitions from circular to annular non-interacting Fermi surfaces accompanied by redistributions of electronic weight among layers. Combining renormalization-group corrections with unrestricted Hartree-Fock calculations, we further demonstrate how Coulomb interactions reconstruct the non-interacting annular Fermi surface into crescent-shaped Fermi lune through spontaneous breaking of time-reversal, mirror and rotation symmetries. We determine competing mean-field phases for four to nine layers and distinguish two types of transdimensional orbital magnetic orders. Beyond the Fermi-lune state that exhibits transdimensional anomalous Hall effect reported recently, we also identify a mirror-preserving Fermi-lune phase with spontaneous in-plane orbital magnetization but a symmetry-forbidden anomalous Hall response. The maximum calculated in-plane orbital moment increases approximately linearly with layer number. The asymmetric Fermi-velocity distributions of these states also lead to intrinsic non-reciprocal longitudinal transport. Measurements presented here in a nine-layer device show a non-reciprocal resistance ratio reaching 23% at a current magnitude of 17 nA. These results elucidate the formation mechanism, competing symmetries, and thickness dependence of Fermi-lune states, and connect their electronic structure to a longitudinal-transport signature beyond the previously reported anomalous Hall phenomenology.

cond-mat.mes-hall↗

arXiv2026

S2M-Trek: From Single to Multi-Sphere Transport via Per-Frame Deep Sets on a Wheel-Legged Robot

We study the problem of scaling dynamic loco-manipulation from a single free-rolling sphere to multiple spheres transported simultaneously on the back of a wheel-legged quadruped, without fences, grippers, or mechanical stops. Multiple identical free-rolling spheres form an unordered set with no persistent identity: their ordering may change independently at each history frame, creating a \emph{per-frame permutation symmetry} that standard history-concatenation set encoders do not explicitly enforce -- these encoders impose only a shared, diagonal permutation symmetry over the full history. We show that this symmetry mismatch leads to a concrete failure mode in curriculum-based reinforcement learning. Within the same PPO training budget, flat MLPs and branch-wise encoders plateau at or below the two-sphere stage, while a history-concatenation Deep Sets baseline (\HCDS) fails to progress past the two-sphere stage in our runs unless ball-to-slot assignments are randomised during training, suggesting that it exploits slot indices as a curriculum shortcut rather than learning identity-free multi-sphere dynamics. We propose \textbf{Per-Frame Deep Sets (\PFDS)}, which performs permutation-invariant pooling within each history frame before temporal readout; we prove that \PFDS is $\Gframe$-invariant and universally approximates continuous $\Gframe$-invariant policies. A $2{\times}2$ ablation over encoder architecture and slot randomisation separates the architectural and data-augmentation pathways, and \PFDS reaches the five-sphere stage with 100\% no-drop transport in simulation across all five random seeds. We further distill the \PFDS teacher into \TactSet via DAgger, replacing privileged sphere-state observations with a $16{\times}16$ Boolean union contact map, yielding a compact and naturally $\Gframe$-invariant tactile representation.

cs.RO↗

arXiv2026

From Charts to Code: A Hierarchical Benchmark for Multimodal Models

We introduce Chart2Code, a new benchmark for evaluating the chart understanding and code generation capabilities of large multimodal models (LMMs). Chart2Code is explicitly designed from a user-driven perspective, capturing diverse real-world scenarios and progressively increasing task difficulty. It consists of three levels: Level 1 (Chart Reproduction) reproduces charts from a reference figure and user query; Level 2 (Chart Editing) involves complex modifications such as changing chart types or adding elements; and Level 3 (Long-Table to Chart Generation) requires models to transform long, information-dense tables into faithful charts following user instructions. To our knowledge, this is the first hierarchical benchmark that reflects practical chart2code usage while systematically scaling task complexity. In total, Chart2Code contains 2,023 tasks across 22 chart types, paired with multi-level evaluation metrics that assess both code correctness and the visual fidelity of rendered charts. We benchmark 25 state-of-the-art (SoTA) LMMs, including both proprietary and the latest open-source models such as GPT-5, Qwen2.5-VL, InternVL3/3.5, MiMo-VL, and Seed-1.6-VL. Experimental results demonstrate that even the SoTA model GPT-5 averages only 0.57 on code-based evaluation and 0.22 on chart-quality assessment across the editing tasks, underscoring the difficulty of Chart2Code. We anticipate this benchmark will drive advances in multimodal reasoning and foster the development of more robust and general-purpose LMMs. Our code and data are available on Chart2Code.

cs.SE↗

arXiv2026

CableVLA: Simulation-Privileged Global-Local Representation Learning for Cable Routing

Cable routing requires coordinated control of global cable topology and changing local contacts. We present CableVLA, an end-to-end multimodal vision-language-action framework that converts simulation-privileged supervision into deployable cable-topology and tactile representations. TopoHead distills node-level physics and current and future cable-topology information into causal visual context for the action expert. TacSense uses complementary frame and taxel branches to learn contact dynamics from resistive arrays, with simulator-derived kinematics and contact events providing supervision beyond the measured force map. A contact gate activates force-tactile residuals that refine the next 8 arm-and-gripper actions of a frozen topology-conditioned policy. Across 345 MuJoCo evaluations, CableVLA improves success from 62.6% for the $π_{0.5}$-V visual baseline to 84.9%. TacSense achieves pronounced gains in slip-transition recognition over a CNN-LSTM baseline with a similar parameter count, and this advantage persists under frozen-encoder probes. Topology prediction and 57-task tactile evaluations assess representation quality, while policy adaptation studies evaluate downstream control performance. Cross-simulator and real-robot comparisons further examine zero-shot policy transfer under changes in dynamics and sensing.

cs.RO↗

arXiv2026

On the stability of the solutions to compressible Navier-Stokes equations

This paper investigates the continuous dependence (stability) of solutions to the barotropic compressible Navier--Stokes equations in critical Besov spaces. Using the Lagrangian approach developed in \cite{Danchin2014}, it was shown that, for $1<p<2d$, the flow map $(a_0,u_0)\mapsto (\bar a,\bar u)=(a\circ X,u\circ X)$ is Lipschitz continuous from $\dot B_{p,1}^{d/p}\times \dot B_{p,1}^{d/p-1}$ into $\mathcal{C}([0,T];\dot B_{p,1}^{d/p})\times E_p(T)$. However, this result does not directly imply the corresponding continuous dependence in the original Eulerian coordinates because of the low regularity of the critical initial data. Previously, only for $p<d$, continuous dependence was known only in certain lower-regularity spaces with a loss of one derivative relative to the natural solution spaces arising essentially as a by-product of the uniqueness argument. We close this gap and prove that for $1<p<2d$ the flow map $(a_0,u_0)\mapsto(a,u)$ is continuous (not Lipschitz continuous) from $\dot B_{p,1}^{d/p}\times\dot B_{p,1}^{d/p-1}$ to $\mathcal{C}([0,T];\dot B_{p,1}^{d/p})\times E_p(T)$ in Eulerian coordinates without any loss of regularity, which together with the known existence and uniqueness theory \cite{Danchin2014} completes Hadamard well-posedness in the critical spaces.

math.AP↗

arXiv2026

SafeStyle: Calibrated Style Residual Injection for Controllable Style-Leakage Trade-off in Diffusion Stylization

Reference-guided diffusion stylization aims to transfer visual style from a reference image while preserving the semantics specified by a text prompt. However, image conditioning often entangles transferable style cues with reference-specific content, leading to an inherent trade-off: stronger conditioning improves style fidelity but increases content leakage, whereas aggressive suppression reduces leakage at the cost of style expression. This challenge is further complicated by the distinct spatial organization of texture- and geometry-dominant styles. To address these issues, we propose SafeStyle, a training-free framework for calibrated style residual injection in frozen diffusion models. SafeStyle first estimates style-supported and content-associated subspaces from compact calibration sets, preserving their informative overlap while suppressing useless content variations. It then transports the purified style evidence over adaptive spatial granularity and constrains its effective influence through an explicit residual-norm budget. Experiments across texture- and geometry-dominant styles show that SafeStyle achieves a DINO style similarity of 0.432 while maintaining competitive text alignment. On a semantically disjoint leakage-stress benchmark, it further achieves a DINO style similarity of 0.474 with only 0.8\% semantic leakage, demonstrating an effective balance between style fidelity and reference-content suppression.

cs.CV↗

arXiv2026

PACC: Propagation-Aware Channel Charting with Physics-Guided Metric Learning

Channel charting has emerged as a promising paradigm that maps high-dimensional channel state information into a low-dimensional latent space, facilitating tasks such as radio environment sensing and beam management. However, existing methods often rely on precise user locations or timestamp-based pseudo-labels, which are difficult to obtain in privacy-sensitive scenarios and rapidly varying wireless environments. To address these limitations, we propose propagation-aware channel charting with physics-guided metric learning (PACC), a location-free framework that constructs channel-domain supervision from propagation characteristics without requiring explicit geographic information. Specifically, PACC designs a propagation-aware dissimilarity metric that adapts to line-of-sight and non-line-of-sight propagation conditions, thereby preserving both local neighborhood relationships and the intrinsic geometry of the radio environment. Simulation results demonstrate that PACC consistently outperforms both classical dimensionality-reduction methods and state-of-the-art learning-based channel-charting approaches under diverse propagation conditions.

eess.SP↗

arXiv2026

Depth-Dominant Skeleton Detection for Natural Scenes

To date, all natural scene skeleton detection follows the paradigm of taking RGB images as the sole input; despite notable progress, methods under this paradigm suffer significant performance degradation on complex-content images. We observe that depth images are inherently insensitive to color and texture, and can provide clear regional contours and inter-region spatial relationships, which naturally alleviates the difficulty of skeleton detection in complex scenarios. Motivated by this observation, this paper proposes for the first time a novel skeleton detection paradigm where depth images serve as the dominant modality and RGB images act as the auxiliary, and accordingly presents a model DDSkel (short for Depth-Dominant Skeleton Detection) under this paradigm. DDSkel employs an asymmetric encoder design to fuse RGB information into depth features, with the RGB modality branch having only 12% the parameters of the depth modality branch. DDSkel has a simple structure without intricate designs. Nevertheless, with only 36% of the trainable parameters of the current best method, DDSkel outperforms all state-of-the-art approaches on SymPASCAL, the most challenging dataset with a large volume of complex images.

cs.CV↗

arXiv2026

Spec-Driven Hardware Evolution via Executable Contract Refinement and Proof-Guided RTL Update

Hardware development is inherently evolutionary: major revisions typically begin by changing intended behavior and then updating a previously validated implementation, rather than regenerating RTL from scratch. Yet most recent LLM-based hardware research still frames the task primarily as prompt-to-RTL generation, offering limited support for semantic version evolution of trusted legacy designs. We present spec-driven hardware evolution, a contract-centered formulation for RTL version iteration. Instead of treating a new feature request as a direct prompt for RTL generation, we refine it into a reviewed executable contract for the next version. This contract specifies what must hold at the externally visible transactional level through a behavior-level reference together with explicit observation and checking semantics, while leaving how the change is realized in RTL to the evolution process. Based on this formulation, we organize hardware evolution into four stages: Specify, Plan, Implement, and Validate. After contract approval, the remaining stages proceed automatically: Plan derives cross-version semantic deltas and localizes affected RTL regions, aided by mutation-based semantic probing; Implement and Validate then perform legacy-aware RTL update under proof-guided checking and iterative repair. We evaluate the framework on a controlled version-evolution case study of a representative TPU datapath block under data-format changes. The results support the feasibility of contract-driven hardware evolution and demonstrate that the proposed backend workflow can effectively drive validated legacy RTL toward next-version functional convergence under a reviewed executable contract. An anonymous artifact for reproducibility is available at https://anonymous.4open.science/r/SDHE-3A6C.

cs.AR↗

arXiv2026

Seed2GS: Camera-Free, Training-Free Object Extraction from 3D Gaussian Scenes via a Single Reference-View Grounding

Extracting a target object from a pre-built 3D Gaussian Splatting (3DGS) scene enables interactive 3D editing. Existing methods either train for tens of minutes per scene, sacrifice accuracy, or require original reconstruction cameras that pre-built assets may not include. We present Seed2GS, which achieves the highest reported LERF-MASK accuracy without original reconstruction cameras or scene-specific representation training. Its key insight is to separate target identity from 3D coverage. QD-SAM3 selects one reliable reference mask from several open-vocabulary candidates, fixing identity once. Seed lift and visibility-adaptive virtual orbits then expose the object from new viewpoints, while tracking propagates the seed without repeated detection. Because the scene remains frozen, these masks supervise only one temporary foreground logit per Gaussian. On LERF-MASK, Seed2GS reaches 92.1% mean intersection over union (mIoU) with a measured compute-only latency of 9.3 seconds, 3.7 points above the strongest scene-trained baseline and 7.6 points above the closest camera-free baseline. With one fixed test reference per scene, the complete pipeline retains 91.1% mIoU; replacing its predicted seed with a ground-truth mask improves mIoU by only 0.72 points. On 3D-OVS, Seed2GS reaches 95.7% mIoU.

cs.CV↗

arXiv2026

Correlated states in charge-transfer heterostructures based on rhombohedral multilayer graphene

Charge transfer is a common phenomenon in van der Waals heterostructures with proper work function mismatch, which enables electrostatic gating to control band alignment and interlayer charge distributions. This provides a tunable platform for studying coupled bilayer correlated electronic systems. Here, we theoretically investigate heterostructures of rhombohedral multilayer graphene (RMG) and an insulating substrate with gate-tunable band alignment. We first develop a self-consistent electrostatic theory for layer charge densities incorporating charge transfer, which reproduces the experimentally observed broadened and bent charge neutrality region. When the substrate's band edge has a much larger effective mass than RMG, its carriers may form a Wigner crystal at low densities. This generates a quantum superlattice that induces topological flat bands in the RMG layer through interlayer Coulomb interactions, which may further lead to Chern insulators driven by Coulomb interactions within the RMG layers. Conversely, with comparable effective masses, we find an interlayer excitonic insulator state at charge neutrality stabilized by interlayer Coulomb coupling. Our work establishes these charge-transfer heterostructures as a rich platform for topological and excitonic correlated states.

cond-mat.str-el↗