arXiv subjects

Qiang Wang

Publications and source records attributed to Qiang Wang.

At least 19 recordsLinked to original sources

arXiv2026

Laplacian Frequency Hierarchies for Efficient 3D Gaussian Splatting Training

A key bottleneck in 3D Gaussian Splatting training is the continual growth of Gaussian primitives, which increases optimization cost and slows convergence, especially at high resolutions. We propose Laplacian Frequency Hierarchies, a simple yet efficient 3DGS scheme that combines Laplacian image decomposition with coarse-to-fine, frequency-staged training. After fitting lower-frequency structure, we archive the corresponding Gaussian field so that subsequent fields can optimize higher-frequency residuals without carrying the full primitive burden, and we compose the rendered components in the image domain via a Laplacian-style reconstruction at inference time. This design reduces the number of active Gaussians during training, thereby lowering optimization overhead and accelerating training. The proposed scheme is plug-and-play and orthogonal to prior 3DGS accelerations: it can be directly combined with strong backbones such as Taming-3DGS and FastGS to improve training speed with competitive reconstruction quality. It achieves average speedups of 1.73x and 1.21x at 1K setting, and 1.74x and 1.33x at 4K setting on Taming-3DGS and FastGS, with larger gains on more challenging scenes and increasingly pronounced benefits at higher resolutions.

cs.CV↗

arXiv2026

Constructions of complete permutations over $\mathbb{F}_q^n$

Complete permutation polynomials play an important role in cryptography, particularly in the design of cryptographic primitives such as the Lai--Massey scheme and S-boxes. We generalize a result of Sun, Li, Guo, and Qu (2021) by characterizing the complete permutation behavior of the mapping $\Psi(X)=M(X+\psi(AX))$ over $\mathbb{F}_q^n$, where $\mathbb{F}_q$ is a finite field of $q$ elements with $q$ being a prime power, $M\in GL(n, \mathbb{F}_q)$, $GL(n, \mathbb{F}_q)$ is the general linear group of order $n$ over $\mathbb{F}_q$, $A_{m \times n}$ is a full-rank matrix over $\mathbb{F}_q$, and $\psi=(\psi_1,\psi_2,\ldots,\psi_n)$ with each component function $\psi_i:\mathbb{F}_q^m\to\mathbb{F}_q$. Furthermore, we establish criteria for the permutation and complete permutation properties of the mapping $F(X)=T(X+B^tf(AX))$ over $\mathbb{F}_{q}^n$, $f: \mathbb{F}_{q}^{m} \rightarrow \mathbb{F}_{q}^{n-m}$, $T \in GL(n, \mathbb{F}_q)$, $A_{m \times n}$ and $B_{(n-m)\times n}$ are full-rank matrices over $\mathbb{F}_q$, $B^t$ represents the transpose of the matrix $B$, and $0<m<n$ are integers. These results also generalize an earlier result of Gravel and Panario (2023), who showed that any arbitrary function $f$ from $\mathbb{F}_q^m$ to $\mathbb{F}_q^{\,n-m}$ can be extended to a bijection over $\mathbb{F}_{q}^n$ through the mapping $F(X)=T(X+B^tf(AX))$, under the condition $AB^t=0$. Here we do not impose the restriction that $AB^t=0$.

cs.CR↗

arXiv2026

BPG: Balancing Plasticity and Generalization for Domain Incremental Learning

Deep neural networks excel in various tasks but struggle to generalize across evolving data distributions, leading to significant performance degradation under domain shifts. Domain incremental learning (DIL) addresses this challenge by enabling models to continuously adapt while retaining prior knowledge. Among existing DIL approaches, the parameter-isolation paradigm achieves state-of-the-art performance. However, these methods often adopt a one-size-fits-all approach to adapt to new domains, resulting in either insufficient learning capacity or redundant parameters. In this work, we propose BPG, a unified framework that addresses both challenges through two complementary components: BPG-Adapter, which dynamically determines each domain's adapter hidden dimension based on domain-specific feature separability, and BPG-Inference, a soft domain mixture strategy that integrates multiple domain-specific models at test time, mitigating domain ID misselection. Experimental results on DomainNet, CDDB, and CORe50 demonstrate that BPG consistently outperforms uniform adapter-based approaches and hard domain selection strategies, achieving state-of-the-art average accuracy while reducing forgetting to as low as 0.22% on DomainNet.

cs.CV↗

arXiv2026

NodeJEPA: Structure-Conditioned Latent Prediction for Node-Level Graph Self-Supervised Learning

Self-supervised learning on graphs is largely shaped by contrastive methods that depend on carefully designed augmentations, and by generative methods that reconstruct node attributes in the input space. Both paradigms can entangle representations with low-level input statistics rather than with relational structure. Joint-embedding predictive architectures (JEPA) instead learn by predicting latent targets rather than reconstructing inputs. Recent work has explored this idea for graph-level representation learning, but how to design JEPA-style objectives for node-level tasks, and which structural signals the predictor should condition on, remains less clear. We present NodeJEPA, a joint-embedding predictive architecture for node-level graph self-supervised learning. NodeJEPA masks structure-aware k-hop ego-subgraphs and trains a context encoder to predict the latent representations of the masked nodes. These targets come from an EMA-updated target encoder with stop-gradient. A structure-conditioned predictor integrates spectral and centrality descriptors through cross-attention. Variance, covariance, and Laplacian spectral regularizers help stabilize the embedding geometry, and an optional curriculum gradually increases masking difficulty during training. Because prediction occurs in latent space, NodeJEPA does not rely on input reconstruction or hand-crafted graph augmentations. We evaluate NodeJEPA on standard node classification benchmarks under linear probing and fine-tuning protocols, and conduct ablations on masking, prediction, and regularization design choices. Our study offers a practical recipe for node-level JEPA-style latent prediction on graphs, and clarifies when structural conditioning helps representation learning. Code, configurations, and evaluation scripts are publicly available at https://github.com/OliverZ-dot/Node-Jepa.

cs.LG↗

arXiv2026

Real-time decoding of quantum error correction codes using high-performance computing

Quantum error correction (QEC) is indispensable for building scalable fault-tolerant quantum computers. Effective QEC demands stringent real-time decoding: the decoder must process syndrome measurements and determine corrections within a time scale--typically on the order of microseconds, to avoid data backlog. Scaling to large number of logical qubits further necessitates significant computational resources. In this work, we propose an architecture, called \emph{THQLink}, for real-time decoding of quantum error correction codes using high-performance computing (HPC) resources. The network connecting the HPC and the control system of quantum processing unit (QPU) is built on TH-Express and can be adapted to different quantum technologies and their associated control stacks. We report a round-trip latency of 2.944 $\mu$s on average, with an incremental overhead of 130 ns per additional hop. Using a parallel window strategy, we demonstrate real-time decoding (1 $\mu$s per QEC round) of the surface code up to distance 19 using a matching-based decoder on CPUs. Our work presents a scalable framework for real-time decoding in fault-tolerant quantum computing. It can be readily applied to quantum-centric supercomputers that feature tight integration between QPU and HPC resources, thereby enabling efficient support for hybrid quantum-classical algorithms and computation-intensive workloads offloaded from the QPU.

quant-ph↗

arXiv2026

Learning a System-Level Surrogate for Hydraulic Excavators: A Simulation-to-Real LSTM Approach

Developing autonomous hydraulic excavators is constrained by limited access to physical machines and the high cost of real-world experimentation. This paper proposes a simulation-to-real framework for learning a system-level digital surrogate using Long Short-Term Memory (LSTM) networks. Instead of modeling internal dynamics, the excavator is treated as an input-output operator, and the surrogate is trained to reproduce its closed-loop behavior under identical control inputs. The approach is first validated in a MuJoCo simulation environment and then transferred to a real excavator. To address measurement inconsistencies in real-world data, a consistency-aware state estimation method based on adaptive Kalman filtering is introduced. Experimental results demonstrate that the learned surrogate achieves high fidelity in both angular velocity and long-horizon trajectory reproduction under closed-loop autoregressive evaluation. These results confirm that the proposed model can serve as a drop-in surrogate for both simulation and physical systems, enabling scalable and efficient development of excavation automation algorithms.

cs.RO↗

arXiv2026

Dynamic Constraint Reconstruction Based Control Barrier Functions for Safety-Critical Control of High-Dimensional Manipulators

Control barrier functions (CBFs) provide formal safety guarantees for constrained nonlinear systems, but their effectiveness relies on accurate system dynamics. In high-dimensional manipulators subject to unknown disturbances and model uncertainties, fixed safety constraints constructed from nominal dynamics may become inconsistent with the actual system behavior, leading to safety degradation or excessive conservatism. This paper proposes a dynamic constraint reconstruction based control barrier function (DCR-CBF) framework for safety-critical control of disturbed robotic manipulators. An extended state observer is employed to estimate lumped disturbances online, and the estimated disturbance is incorporated into high-order control barrier functions to reconstruct safety constraints according to the estimated true dynamics. To address estimation inaccuracies, a safety margin is introduced, and a sufficient condition is derived to guarantee forward invariance under bounded estimation errors. Simulation studies on a 4-DOF excavation manipulator demonstrate that the proposed DCR-CBF method achieves zero safety violation under strong unknown disturbances while significantly improving trajectory-tracking performance compared with standard and robust CBF methods.

eess.SY↗

arXiv2026

MCPEvol-Bench: Benchmarking LLM Agent Performance Across Dynamic Evolutions of MCP Servers

As Model Context Protocol (MCP) servers emerge as the core infrastructure for connecting LLMs with external tools, existing benchmarks leverage real-world MCP servers to evaluate LLM agents' tool-using capabilities. However, these benchmarks overlook the continuous evolution of tool interfaces and functionalities within MCP servers, resulting in flawed assessments that fail to capture the agent's adaptability in changing tool landscapes. To bridge this gap, we introduce \textbf{MCPEvol-Bench}, a novel benchmark for evaluating the task-solving capabilities of LLM agents under dynamic toolset evolution. Inspired by large-scale empirical study, we propose 11 mutation operators to simulate realistic tool evolution within 123 MCP servers. We benchmark 12 state-of-the-art LLMs on multiple versions of MCP servers, revealing that even frontier models struggle to adapt to evolving tools. For instance, GPT-5.4 and Claude-Sonnet-4-6 exhibit performance declines of 13.7\% and 14.4\% in evolved MCP servers, respectively, accompanied by substantial increases in planning and reasoning errors. These findings highlight the vulnerability of LLM-driven workflows, establishing MCPEvol-Bench as a standard for evaluating agent adaptability in dynamic tool environments.

cs.AI↗

arXiv2026

LPM: Industrial-Scale Generative Video Restoration

We present the Large Processing Model (LPM), a diffusion-based generative framework for photorealistic video restoration under complex, in-the-wild degradations. To our knowledge, LPM is the first generative video restoration model deployed at industrial scale. LPM addresses the diverse degradations in user-generated content (UGC) through a unified system encompassing large-scale data engineering, foundation-model training, and efficient inference. Its enhanced architecture, progressive training strategy, and temporal-pyramid inference mechanism jointly enable high-fidelity, temporally consistent restoration of arbitrarily long videos across the broad content distribution encountered on UGC platforms. LPM has been deployed in production at Kuaishou, where videos processed by the model account for approximately 45% of total viewing time, delivering consistent improvements across key quality-of-experience metrics. Beyond perceptual enhancement, LPM delivers substantial system-level benefits: at comparable perceptual quality, it reduces bitrate by 20% relative to Kuaishou's in-house codec, yielding annual bandwidth cost savings on the order of hundreds of millions. Its low serving cost also enables integration into products such as Kling, demonstrating that generative restoration can be practical, scalable, and cost-effective for large-scale video processing.

cs.CV↗

arXiv2026

Bidirectional Resource Scheduling for Disaggregated and Asynchronous RL Post-Training

It is well established that the reasoning capabilities of large language models (LLMs) can be improved by applying reinforcement learning (RL) in a post-training stage. In a standard RL iteration, the current model (the policy) generates experience through rollouts, and the resulting data is then used to update the policy during training. High-performance RL frameworks such as StreamRL and AReaL employ a disaggregated architecture and asynchronous rollouts to better exploit both rollout and training resources, thereby increasing overall system throughput. Nonetheless, across varying RL setups (e.g., hardware configurations, model scales, staleness levels, and hyperparameters) and under changing workloads, it remains common for both rollout and training resources to experience idle periods. In this paper, we present BiDiRL, a hybrid time-space multiplexing architecture for asynchronous, disaggregated RL designed to reduce resource idleness. First, we develop a hot-switch runtime that enables rapid switching between rollout and training resources with negligible overhead. Second, we propose a static, scheduling-aware planner based on time-performance modeling that chooses a hot-switch-friendly resource partition, so that rollout and training durations are roughly balanced at a coarse level. Third, at execution time, we introduce a bidirectional scheduler that further exploits runtime bubbles through fine-grained resource switching, allowing the bottleneck stage to temporarily borrow idle resources from the other pool. Across a wide range of workloads, datasets, and models on two 32-GPU testbeds, BiDiRL increases RL training throughput by up to 1.94x compared with RL systems including veRL, AReaL, and ROLL, without affecting convergence behavior.

cs.DC↗

arXiv2026

Attractor Flow Versus Hesse Flow in Wall-Crossing Structures

We recast the physics discussions in the paper of Dieter Van den Bleeken \cite{svan2012bps} within the context of wall-crossing structure \`a la Kontsevich and Soibelman \cite{kontsevich2014wall}. In particular, we compare the Hesse flow given in \cite{svan2012bps} and the attractor flow on the base of the complex integrable system, and show that both can be used in the formalism of wall-crossing structure. We also propose the notions of dual Hesse flow and dual attractor flow, and show that under the rotation of the $\mathbb{Z}$-affine structure, the Hesse flow can be transformed into the dual attractor flow, while the attractor flow into the dual Hesse flow. This suggests its possible use in Mirror Symmetry.

math-ph↗

arXiv2026

Secure-CHG: A Comprehensive Framework for Robust and Fair Federated Learning via Hybrid Defense and Contribution-Aware Trust

Federated Learning (FL) is highly susceptible to stealthy backdoor attacks, which aim to force a model into predicting an attacker-chosen target class for inputs containing a specific trigger. However, existing statistical defenses primarily focus on the early stages of model convergence. In this paper, we identify a fundamental vulnerability termed ``Late-stage Failure.'' We demonstrate that as the global model converges, decaying gradient norms render malicious and benign updates morphologically indistinguishable. This vanishing statistical variance effectively blinds traditional defenses, enabling adaptive adversaries to remain dormant and subsequently hijack the training process. To overcome these constraints, we propose Secure-CHG, a hybrid framework that pivots the defense paradigm from superficial morphological detection toward intrinsic semantic contribution verification. Secure-CHG employs an adaptive defense pipeline: a cascaded statistical filter stabilizes optimization during the early oscillatory phase, while a novel CHG-Shapley mechanism takes over during late-stage convergence. By leveraging sample hardness (i.e., local training loss) to project updates into a composite Hardness-Gradient space, it effectively amplifies adversarial semantic traces, enabling the isolation of stealthy attackers even as gradient norms vanish. Furthermore, we derive a closed-form solution for CHG-Shapley, facilitating low-complexity, retraining-free node valuation and trust-modulated aggregation. Extensive evaluations on CIFAR-10, MedMNIST, and NEU-SDDB demonstrate that Secure-CHG effectively mitigates Late-stage Failure. Specifically, it significantly suppresses advanced backdoor attacks, reducing their attack success rate by 2.3$\times$ and 2.0$\times$ relative to the mainstream Krum and Trimmed Mean baselines, respectively.

cs.CR↗

arXiv2026

Heterogeneous Tactile Transformer

Tactile sensors are inherently heterogeneous: a model trained on one sensor cannot be directly used on another, which limits learning contact-rich manipulation policies from diverse tactile data at scale. To bridge this gap, we propose the Heterogeneous Tactile Transformer (HTT), a framework that learns shared tactile representations across heterogeneous sensors. HTT consists of sensor-specific encoders and a shared transformer trunk, and is pretrained with per-modality masked reconstruction together with cross-modal alignment between paired sensors. Pretraining uses our novel Heterogeneous Paired Tactile (HPT) dataset, containing 1.6M synchronized paired frames across four vision- and array-based tactile sensors. Across distinct tactile perception and real-world manipulation tasks, HTT is shown to learn transferable representations that adapt to new tasks and previously unseen sensors. Dataset, code, and model checkpoints will be released upon publication at https://jxbi1010.github.io/htt-gh-page/.

cs.RO↗

arXiv2026

Geometry of BPS Attractor, Hessian, and Spectral Flows

We provide a systematic and rigorous geometric framework that relates three structures naturally associated to BPS central charges in $\mathcal{N}=2$ supersymmetric gauge theories: the split attractor flow (SAF) of $|Z|$, the Hessian flow (HF) of $\operatorname{Im}(e^{-i\vartheta}Z)$, and the spectral network (SN) on the base curve of the Hitchin fibration. Our main contributions are: (i) a concise proof of orthogonality between SAF and gradient Hessian flow using only the K\"ahler structure; (ii) a precise lift--projection duality showing that the spectral network projects to the \emph{characteristic Hessian flow} (the Hamiltonian flow of $\operatorname{Im}(e^{-i\vartheta}Z)$) on the Hitchin base, clarifying a crucial distinction; (iii) a complete proof of the Kontsevich--Soibelman (KS) equivariance by induction on the SAF tree depth, with the geometric ordering provided by the characteristic Hessian flow. We illustrate the framework with detailed and nontrivial examples: $SU(2)$ pure and $N_f=4$ (including BPS indices for higher flavour charges), $SU(3)$ pure (full BPS spectrum reconstruction), $SU(4)$, the Kronecker $3$-quiver, and we apply the induction to derive a closed-form BPS spectrum for the Argyres--Douglas $H_1$ theory, $\Omega(n\alpha_1+m\alpha_2)=\frac{1}{n+m}\binom{n+m}{n}\binom{n+m}{n+1}$, which is known from Cecotti--Vafa and serves as a strong consistency check of our geometric recursion. In the tropical limit we obtain an explicit generating function for disk counts in $SU(N)$ gauge theories, $Z_{\mathrm{disk}}^{SU(N)}(y) = \exp\!\,\Bigl( \sum_{\alpha\in\Phi_+} \sum_{k=1}^{\infty} \frac{1}{k}\binom{k+\mathrm{ht}(\alpha)-1}{\mathrm{ht}(\alpha)-1} e^{-k\langle\alpha,y\rangle} \Bigr) $, which reproduces the standard scattering diagram result and confirms the geometric framework.

math-ph↗

arXiv2026

HyperDFlash: Hyper-Connection-Aligned Block Speculative Decoding with Gated Residual Reduction

We present HyperDFlash, a block-parallel speculative decoding framework tailored to DeepSeek-V4's Hyper-Connections (HC). Despite the strong performance of DeepSeek-V4's native Multi-Token Prediction (MTP) module on initial token drafting, its draft accuracy degrades sharply at later positions, as error accumulation from unverified intermediate tokens harms draft acceptance rates. Although the original DFlash method supports efficient one-pass block drafting, it cannot be seamlessly adapted to the HC paradigm, since DeepSeek-V4's multi-path residual stream induces inherent feature misalignment with conventional drafting designs. To resolve this architectural mismatch, we propose two dedicated, model-aligned optimizations for HC residual streams. First, we adopt pre-collapse residual states as the exclusive conditioning signal, preserving complete multi-path structural information and better aligning the drafter with the target's native prediction pathway. Second, we replace the heavy generic linear compressor with a lightweight gated residual reducer, whose parameters are directly inherited from the target model's built-in hc_head module. This design yields input-aware path aggregation with three orders of magnitude fewer parameters while maintaining precise architectural alignment. We further enhance model training via a targeted KL distillation loss applied to the LM-head, regularizing predictions against the target distribution to improve early draft quality. Extensive experiments across math reasoning, code synthesis, and conversational benchmarks demonstrate that HyperDFlash consistently outperforms both the native MTP baseline and vanilla DFlash adaptation, achieving substantial gains in average accepted draft length and decoding speedup. These results validate HC alignment, gated reduction, and targeted distillation for high-performance speculative decoding.

cs.LG↗

arXiv2026

What Transfers from Text to Vision? Capability Scaling Laws and Transfer Dynamics for VLMs

Choosing the right large language model (LLM) backbone is the most consequential decision when building a vision-language model (VLM), yet it remains fundamentally unprincipled: compute-based scaling laws fail to generalize across model families, and no framework exists for directly predicting VLM performance before training begins. We propose the Capability-Driven Multimodal Scaling Law, the first cross-family framework that predicts VLM benchmark accuracy from directly observable textual capability. Given a low-dimensional capability score $S$ extracted from LLM textual benchmarks via PCA, we model VLM performance as a function of $S$, with a per-backbone transfer rate and an absorption rate that quantifies data-scaling efficiency. To fit and validate the framework, we train over 150 VLMs on 34 LLMs spanning 7 model families under a strictly controlled recipe. Evaluations on more than 200 textual and 50 multimodal benchmarks show that the law accurately extrapolates transfer rate from models up to 8B parameters to 72B-scale backbones, predicts full VLM training trajectories with high fidelity, and generalizes to entirely held-out model families. Beyond the scaling law, our analysis surfaces actionable insights: certain textual benchmarks negatively correlate with multimodal performance, exposing latent benchmark-gaming behavior; base LLMs outperform instruction-tuned counterparts as VLM backbones due to higher absorption rates and lower data-scaling decay; and different model families occupy distinct positions in the transfer--absorption space. The framework turns backbone selection from costly empirical sweeps into a principled, quantitative decision. Code and data are available at https://github.com/wangq-dev/CDMScaling.

cs.CL↗

arXiv2026

Accelerating Disaggregated RL for Visual Generative LLMs with Diffusion-Based Parallelism and Trainer-Assisted Generation

Reinforcement learning (RL) has become a dominant post-training paradigm, driving the emergence of high-performance RL systems such as veRL for autoregressive large language models (LLMs). In parallel, diffusion-oriented RL algorithms, e.g., DanceGRPO and FlowGRPO, have rapidly expanded the scope of RL from language reasoning to diffusion-based visual and flow-based generation. However, efficient RL systems for diffusion generative LLMs remain underexplored. Existing implementations, e.g., veRL-Omni, still rely on colocated execution, which simplifies synchronization but couples rollout and training resources, limits heterogeneous deployment, and constrains independent scaling. To this end, we introduce DigenRL, a disaggregated RL framework for diffusion-based generative LLMs that supports flexible resource allocation, accommodates heterogeneous GPUs, and facilitates efficient task scheduling. To maximally reduce the execution bubbles in the disaggregated architecture, we propose: 1) a generation-axis pipeline (GAP) and time-step parallelism (TSP) in the diffusion architecture to enable finer-grained pipelining between rollout and training; 2) an elastic trainer-assisted generation (TAG) approach to enable the trainer GPU resources to dynamically assist in executing rollout generations; and 3) a tightly one-step constrained asynchronous strategy to further utilize the tail bubble in the pipeline. Extensive experiments are conducted on three hardware testbeds with 16-32 GPUs using HunyuanVideo-13B, Wan2.1-14B, FLUX.1-12B, and QwenImage-20B generative models. Experimental results show that DigenRL achieves 1.56-2.10x throughput improvements over state-of-the-art diffusion RL systems, veRL-Omni and GenRL.

cs.AI↗

arXiv2026

A novel approach for studying two-particle momentum correlation function in relativistic nuclear collisions

Two particle momentum correlation functions provide a nontrivial tool for probing the strong interaction and/or extracting particle emission source information in relativistic nuclear collisions. Although transport models can describe the microscopic phase-space evolution of the collision system, calculating correlation functions within the framework of transport models remains challenging. In this paper, we employ the mixed-event technique to calculate two particle momentum correlation function as $C(k^*)= \mathcal{N} \xi(k^*)\frac{N_{\mathrm{same}}(k^*)}{N_{\mathrm{mixed}}(k^*)}$ based on the parton and hadron cascade model PACIAE simulated final hadronic state (FHS) with introducing a modification factor $\xi(k^*)$ to improve the treatment of final-state interactions and quantum statistics effects in the PACIAE model. The simulated results show good agreement with the ALICE data for $Kp$, $pp$, $p\Lambda$, and $\Lambda\Lambda$ momentum correlation functions in $pp$ collisions at $\sqrt{s}=7$ TeV. On the other hand, the particle emission source radius of the correlated pairs are also evaluated based on the simulated FHS self-consistently. Since the PACIAE model employs hadron-hadron cross sections derived from the additive quark model, the calculation of two-particle momentum correlation functions does not require prior assumptions about the interaction between the two correlated particles. This successful ``PACIAE + modification factor" approach may shed light on the future study of momentum correlation functions for dimesons, dibaryons, and even diexotic hadrons.

hep-ph↗