Searcharxiv⌕ Search

arXiv subjects

Wei Li

Publications and source records attributed to Wei Li.

At least 19 recordsLinked to original sources
arXiv2026

Demystifying the Impact of Physical Non-Idealities on Kernel Performance

Physical non-idealities in integrated circuit (IC) design, fabrication, and operation are concealed by conventional hardware abstractions, causing unexpected performance variations in kernels. For large-scale workloads, such as AI training and inference, performance unpredictability can lead to substantial losses. We investigate this problem in NVIDIA H100 GPUs with their distributed shared memory (DSMEM) architectural feature that reduces data movement from the L2 cache to Streaming Multiprocessors (SMs) with multicast loads, and uncover that floorsweeping restricts the number of SMs usable for clustered execution, increasing the number of execution waves of kernels. To alleviate this performance penalty, we introduce the wave-aware clustering strategy that accounts for these resource constraints and improves the throughput of the Hopper version of FlashAttention 4 by up to 5.30% over the unclustered baseline. Furthermore, we discover that the power constraint imposed by the thermal design power (TDP) limit causes the DSMEM-based memory bandwidth optimization to improve the throughput even for compute-bound kernels by creating a power headroom for higher SM frequencies through reduced data movement. Extending this analysis beyond GPUs, we find that power constraints also limit throughput scaling on Tenstorrent's Wormhole accelerator, highlighting the broader importance of accounting for physical non-idealities in kernel performance engineering.

cs.AR↗
arXiv2026

UniFlow: Unified Multi-Scale Flow Matching for Time Series Generation

Real-world time series exhibit complex dynamics across multiple temporal resolutions, yet effectively incorporating multi-scale information into generative modeling remains challenging without increasing generative complexity. We propose UniFlow, a unified framework that integrates multi-scale representation learning with efficient non-autoregressive generation through a shared-semantic discrete interface. Specifically, a multi-scale tokenizer employs shared-index voting to consolidate complementary temporal information into a single public-ID sequence, while scale-specific codebooks preserve the information required for reconstruction. Building on this unified representation, we develop geometry-aware categorical flow matching, which initializes transport from learned low-rank anchors and generates public identities non-autoregressively. This design captures multi-scale temporal structure without requiring multiple generative token streams or sequential token decoding. Experiments across four benchmarks and both short- and long-term settings demonstrate improved generative fidelity, strong cross-scale consistency, and substantial inference speedups over representative autoregressive and diffusion-based baselines. Code is available at https://github.com/William-Liwei/uniflow

stat.ML↗
arXiv2026

Hidden Triakontadipole Order in the Triangular-Lattice Magnet CeMgAl$_{11}$O$_{19}$

The absence of magnetic Bragg peaks and a broad excitation continuum are commonly recognized experimental features of a quantum spin liquid. The Ce-based triangular-lattice hexaaluminate CeMgAl$_{11}$O$_{19}$ exhibits both signatures, yet its effective model and ground state remain under debate. Experiments have variously suggested a quantum spin liquid, proximate umbrella point, and disorder-suppressed magnetic order. Here we trace these contradictions to an overlooked multipolar degree of freedom: the transverse pseudospin of the $|\pm5/2\rangle$ Kramers doublet is a rank-5 triakontadipole rather than an ordinary dipole. Analyzing the magnetization, neutron spectra, and thermodynamics on equal footing, we establish a dipole-triakontadipole XXZ model with $J_{\perp} \simeq 0.65$ K and $J_c \simeq -0.13$ K. The resulting planar triakontadipolar superfluid order is hidden from dipolar probes, explaining the absence of a magnetic Bragg peak in neutron scattering. The zero-field excitations separate into a low-energy triakontadipole wave and a broad higher-energy continuum. We further propose nuclear spin-lattice relaxation and spin-current experiments to probe the gapless excitations of this hidden triakontadipole superfluid. Our findings thus open a new avenue for high-rank multipolar physics in rare-earth triangular-lattice magnets.

cond-mat.str-el↗
arXiv2026

Covariance-dependent participation scaling of Comtet disorder on open hyperbolic lattices

Negative curvature makes the electrostatic Green function decay exponentially, producing a finite disorder-variance budget even as radial volume grows exponentially. We investigate whether this near-field concentration also determines wave-function participation on open hyperbolic lattices. A Gaussian control preserves the full one-site distribution of Comtet disorder at every vertex while removing its off-diagonal covariance. Complete-window participation ratios on $\{7,3\}$ graphs with $N=847,2240,5887$ yield an effective-exponent difference of $-0.028$ at $W/t=0.5$ and $+0.116$ at $W/t=3$; both signs persist upon extending to $N=15435$. Size-resolved squared-kernel radii lie near $2.45\ell$ on the larger graphs of this sequence, below the cutoff-matched continuum value $3.12\ell$. Distance-shell weights resolve the finite discrete measure associated with this shift, while higher participation moments and spacing ratios characterize the accompanying wave-function and spectral structure. A marginal-matched exponential-kernel family gives positive exponent contrasts at all five sampled kernel lengths at $W/t=0.5,3$, separating the Comtet response from a description based on the decay length alone. These results distinguish a concentrated variance budget from the covariance-dependent participation response of finite open hyperbolic systems.

cond-mat.dis-nn↗
arXiv2026

Efficient Causal Survival Inference under Limited Surrogate Monitoring

Post-treatment surrogate endpoints can improve the efficiency of causal survival effect estimation, but their monitoring may end before follow-up for the primary outcome. We investigate the impact of limited recurrence monitoring on semiparametric efficiency when mortality follow-up continues. We derive the efficient influence function under limited monitoring and establish an exact expression for the efficiency gain from extending monitoring between any two durations. Our framework encompasses the absence of surrogate information and the usual shared-censoring scheme as boundary cases, and allows comparison with fully observed surrogates. We develop a doubly robust estimator under limited surrogate monitoring and establish asymptotic normality and semiparametric efficiency for its cross-fitted version under suitable conditions. Simulations and an application to breast cancer data illustrate the proposed method.

stat.ME↗
arXiv2026

VEvo-Harness: Causally-Guarded Self-Evolving Harness for Agentic Image Generation

The central bottleneck in visual generation and editing has migrated from achieving pixel-level fidelity to satisfying densely constrained user specifications through increasingly agentic workflows. Prevailing agentic generation pipelines remain shallow generate-verify-rewrite loops treating VLM judgments as coarse visual feedback, either discarding repair experience or admitting ungated updates that poison skill library. With memorize and govern periods left open, they degrade into ungoverned iterators of repeated trial-and-error rather than continual self-evolving harness. Towards this issue, we present \textbf{VEvo-Harness}, the first causally-guarded continual self-evolving harness system for agentic visual generation. VEvo-Harness introduces four coupled innovations: (1) \textit{a visual traceback} that densifies sparse judgments into root-cause attributions with a hierarchical verification tree; (2) \textit{a visual continual governance} that admits only causally valid skill patches through counterfactual dual-arm image generation replay; (3) \textit{a lifecycle-managed harness state} that hierarchically distills root-cause-anchored episodes across working, experience, and pattern tiers into a persistent skill library of routable and causally verified capabilities; (4) \textit{a heterogeneous sandbox policy} that orchestrates a diverse action space of repair primitives, dynamically routing cost-aware interventions based on root causes to select the most effective strategy. VEvo-Harness surpasses state-of-the-art agentic methods and rivals proprietary generators (\textit{e.g.}, GPT-Image, Nano-Pro) while maintaining the continual harness evolution and generalization.

cs.CV↗
arXiv2026

Stochastic dynamics generation of directed percolation using a two-channel convolutional neural network

Neural networks provide a promising approach to learning and generating the stochastic dynamics of nonequilibrium systems. In this study, we propose a stochastic dynamics generative method based on a two-channel convolutional neural network, using the (1+1)-dimensional directed bond percolation model to investigate its ability to learn local stochastic transition rules and reproduce stochastic dynamics across different system sizes. The proposed method takes the current particle configuration and the control parameter as inputs, learns the conditional activation probabilities at the next time step through local convolutions, and generates stochastic temporal evolution through iterative Bernoulli sampling. The results show that the network-generated dynamics agree well with Monte Carlo simulations across different control parameters. Furthermore, applying the trained network to a substantially larger system with $L=10000$ yields an estimated critical probability of $p_c=0.643$ and an effective density decay exponent of $δ\approx0.1592$, close to the established theoretical values. These findings demonstrate that the proposed method can reproduce the stochastic dynamics of directed percolation and capture its macroscopic dynamical and statistical properties across different spatial sizes.

math-ph↗
arXiv2026

Central Limit Theorems and Network-Free Inference for Heterogeneous Spatial Autoregressions on Sparse Graphons

Network dependence is widely regarded as a fundamental obstacle to inference: outcomes linked through a graph are not independent, the graph is rarely observed in full, and models for how dependence propagates are easily misspecified. We show that for a broad class of responses generated through a random network, the obstacle is smaller than it appears. We study a heterogeneous spatial autoregression on a sparse graphon, in which each node's response is a weighted average of its neighbors' responses with a node-specific, nonparametric coefficient plus a node-specific input, and we prove that sample means and smooth functions of vector means are asymptotically Gaussian down to average degree nearly logarithmic in the population size, allowing unbounded graphons with hubs and heavy-tailed inputs. The limiting variance interpolates exactly between two regimes according to the sampling fraction. When the observed nodes are a vanishing fraction of the population, the variance is the marginal variance of the response, the ordinary sample variance is consistent, and the usual $t$-interval is valid with no knowledge of the network. The proof averages over the graph rather than conditioning on it, isolating the behavior of the response through an exact identity for resolvents of Markov kernels and controlling the network noise by an Efron-Stein inequality; neither step requires the walk-by-walk covariance bounds of fixed-network analyses. Consequences for regression with node-level and network-derived covariates are derived.

math.ST↗
arXiv2026

PriorMaskMap: Robust Online Vectorized Map Construction with Biased Priors

Online vectorized map construction provides autonomous vehicles with essential, real-time semantic and geometric scene understanding. While leveraging prior maps presents an opportunity to improve accuracy, they are often biased due to outdated information or temporary road changes. Existing methods, which typically treat priors as fully reliable or only consider limited noise patterns, suffer severe performance degradation when these priors are imperfect. To address this, we propose PriorMaskMap, a robust framework for prior-guided vectorized map construction. Its first innovation is a Prior-Aware Confidence Estimator (PACE) that evaluates the input rasterized map against current visual observations. It assigns confidence scores to regions classified as reliable, erroneous, or blank, creating a foundational mask that guides two subsequent branches. In one branch, the Dual-Prior Query Generator (DPQG) selectively filters and embeds both rasterized and vectorized priors to initialize informative queries from trustworthy map structures. In the other, the Confidence-Guided Prior Fusion (CGPF) module spatially propagates and refines the mask, using the refined version to govern the fusion of BEV features with the prior map, thereby providing rich contextual representations for the map decoder. Evaluation on the nuScenes dataset shows that PriorMaskMap achieves superior mapping precision over existing approaches and exhibits marked robustness when provided with priors containing apparent errors. Our code is released at https://github.com/healenrens/PriorMaskMap.

cs.CV↗
arXiv2026

PulmoWorld: An Asynchronous Multimodal Medical World Model for Longitudinal Pneumonia Dynamics

Pneumonia changes across respiration, inflammation, and organ function, while the measurements that describe it arrive at different times and are often incomplete. We introduce PULMOWORLD, a world model for longitudinal pneumonia dynamics that connects asynchronous state estimation, recursive simulation, and evidence-constrained planning. History-aware filtering separates state propagation from observation-driven correction, so a measurement updates the shared belief only when it is observed. That belief supports recursive forecasts of clinical values and modality representations, and summaries of future prognosis. Cap-MBRL searches imagined continuations inside an authorized action set and replaces a frozen fallback only when a candidate meets the risk budget and a predicted-cost margin. On the pneumonia cohort MASS, recursive forecasts are competitive with task-trained sequence models, and simulated future states retain prognostic information. Trained separately on public MIMIC-IV-CXR chest radiographs and clinical trajectories, this model improves prognostic discrimination over latest-value fusion in development and is competitive on long-horizon error on the locked pneumonia panel. In five known-mechanism MASS-Syn environments, Cap-MBRL lowers selection regret relative to direct fallback execution.

cs.CV↗
arXiv2026

VGIQA: Fine-Grained Image Quality Assessment with Visual Grounding Rewards

Image quality assessment (IQA) aims to evaluate the perceptual visual quality of images. Recently, the rapid development of Multi-modal Large Language Models (MLLMs) has significantly advanced the IQA field. By integrating chain-of-thought (CoT) reasoning and reinforcement learning (RL), these models can generate more accurate and interpretable results while relying on less training data. However, existing MLLM-based methods are predominantly constrained to holistic, image-level evaluations and lack the capability to produce fine-grained, localized assessments, which substantially limits their practical applicability. To address this limitation, we propose VGIQA, a novel IQA model endowed with both local and global quality assessment capabilities. VGIQA is trained via a two-stage paradigm comprising a cold-start phase followed by reinforcement learning. Driven by carefully designed local quality assessment reward functions, our model exhibits exceptional image quality understanding on VGSR-2K, a new dataset we propose specifically for local quality evaluation, significantly outperforming existing open-source MLLMs such as Qwen2.5-VL-72B-Instruct and Qwen3-VL-32B-Instruct. Furthermore, the local quality evaluation task can effectively enhance the model's performance on the degradation perception task.

cs.CV↗
arXiv2026

PatentAgentBench: Benchmarking LLM Agents with Professional Patent Task Suites

Patent practice lies at the intersection of technical, legal, and linguistic work, making it challenging yet crucial to technological innovation. Recently, rapid advances in large language models (LLMs) and agentic systems have demonstrated the potential to complete realistic business workflows and even patent works. This raises a question: can current AI systems perform patent work reliably? Existing benchmarks cover a limited range of patent tasks and simplify evidence acquisition by providing reference materials or candidate documents. In this paper, we introduce PatentAgentBench, a benchmark comprising 510 instances across eight task suites: novelty search, freedom-to-operate (FTO) search, design FTO, invalidity search, claim charting, patent drafting, office-action response, and patent translation. Curated from real patent records and professional materials, the benchmark combines task-specific metrics with expert-reviewed annotations and rubrics. Retrieval tasks such novelty search, FTO, etc., require agents to search and compare patent collections without a supplied candidate pool. The drafting and OAR tasks assess the quality of agents' work given technical disclosure and office action notice. We evaluate seven LLMs across four agent harnesses on PatentAgentBench, finding that current systems still struggle to identify novelty, assess risk, and understand practitioner intent throughout patent workflows.

cs.AI↗
arXiv2026

MedVTok: A General-Purpose Medical Visual Tokenizer

Multimodal medical AI requires a common visual interface for encoding heterogeneous imaging evidence for patient condition modeling. However, existing medical visual tokenizers are often tied to a specific dimension and imaging modality, forcing multimodal systems to rely on fragmented and poorly reusable representations. We introduce MedVTok, a general-purpose visual tokenizer that maps 2D images and 3D volumes from diverse imaging modalities into a unified token space. Building such a tokenizer poses two key challenges: preserving anatomical structure across dimensions and capturing clinical semantics across modalities. To this end, MedVTok introduces (i) slice differential regularization, which explicitly models inter-slice anatomical coherence missing in previous per-pixel/per-voxel supervision, and (ii) multi-expert representation alignment, which integrates knowledge from multiple medical expert encoders while retaining modality-specific complementary knowledge. Trained on 28M images and 70K volumes across 9 imaging modalities, MedVTok supports classification, retrieval, segmentation, synthesis, visual question answering, and medical report generation through a single tokenizer. Across 24 benchmarks, MedVTok achieves state-of-the-art performance, demonstrating a scalable token interface that unifies medical image dimension, modality, and task usage.

eess.IV↗
arXiv2026

Training-Free Inference Acceleration for Video Large Language Models: Methods, Trade-offs, and Evaluation

Video Large Language Models (VLMs) have demonstrated strong capabilities in video understanding and reasoning, but processing long visual sequences incurs substantial computational, latency, and memory overhead. This survey examines training-free inference acceleration, focusing on reducing inference costs while preserving task-relevant spatiotemporal evidence. We organize existing methods by their intervention stages in the inference pipeline, covering input, visual encoding, projector, prefill, and decoding. Post-training quantization (PTQ) is discussed separately as a cross-stage extension with varying calibration requirements. We analyze how intervention position affects evidence preservation and computational savings, and why local efficiency improvements may not translate into end-to-end acceleration. We complement this analysis with quantitative evidence from prior studies and our own experiments under a shared evaluation setting to assess accuracy-latency trade-offs.

cs.CV↗
arXiv2026

The ascending chain condition for radical well-mixed difference ideals

We establish a finite-basis theorem for radical well-mixed difference ideals in a difference polynomial ring with one endomorphism over an arbitrary difference field. The result extends to every finitely difference-generated difference algebra over such a field and is equivalent to the ascending chain condition for radical well-mixed difference ideals, hence resolving a long-standing open problem in difference algebra. The proof proceeds in two stages. Over an inversive difference coefficient field, we establish a uniform inverse-transform bound relating the perfect closure of a finite system to its radical well-mixed closure. This uses Noetherian induction in a fixed truncated polynomial ring, together with generic freeness and going-down for flat ring homomorphisms. We then prove that the radical well-mixed closure of a system over a difference field $K$ is the contraction of its radical well-mixed closure over the inversive closure of $K$. This contraction result follows from successive prime lifting along flat extensions and removes the inversiveness assumption without requiring the coefficient-field extension to be finitely generated.

math.AC↗
arXiv2026

Approximation algorithms for identical parallel machine scheduling with assignment costs

We study non-preemptive scheduling on identical parallel machines in which assigning job $j$ to machine $M_i$ incurs the cost $c_{ij}=a_iw_j+b_j$. The processing times $p_j$ and the cost weights $w_j$ are independent, and the number of machines is part of the input. For every fixed $\varepsilon\in(0,1]$ and every deadline $T$ for which a schedule with makespan at most $T$ exists, we give a polynomial-time algorithm that computes a schedule with makespan at most $(1+\varepsilon)T$ whose total assignment cost does not exceed the minimum under the deadline $T$. Using this algorithm as a subroutine, we obtain a polynomial-time approximation scheme for minimizing the makespan under a total cost budget whose schedules always satisfy the budget; this answers an open question for costs $c_{ij}=l_ic_j$ with $c_j$ independent of $p_j$. For the lexicographic problem, in which the total assignment cost is minimized among all schedules of minimum makespan, we obtain in polynomial time a schedule with makespan at most $1+\varepsilon$ times the minimum and total assignment cost at most that of a lexicographically optimal schedule; this improves the makespan guarantees of earlier algorithms for additive and multiplicative costs. For costs $c_{ij}=s_i+a_iw_j$, where $s_i$ is charged for each job processed on $M_i$, the same guarantees are obtained in quasi-polynomial time.

cs.DS↗
arXiv2026

Weak type $(1,1)$ estimates for noncommutative maximal truncated rough singular integrals

We prove a weak type $(1,1)$ estimate for maximal truncations of noncommutative rough singular integrals whose angular kernels satisfy the classical $L\log L$ integrability and cancellation conditions. The main ingredient of the proof is a new linearization method for handling the maximal truncations acting on the bad functions arising from the noncommutative Calderón--Zygmund decomposition.

math.FA↗
arXiv2026

Perfect Born Sampling of Symmetric Thermal Tensor Network States

Accurate calculations of correlated fermions at low temperatures constitute a major challenge in quantum many-body physics, most notably in the studies of high-$T_c$ superconductivity and ultracold atom optical lattices. Stochastic sampling of tensor-network states offers a promising route to tackle this problem; however, existing schemes have long faced a fundamental dilemma---sampling efficiency and symmetry acceleration \textit{cannot} be achieved simultaneously. Here we propose a perfect Born sampling approach for thermal tensor networks, which performs importance sampling directly from the purified density matrix via the Born rule and incorporates Abelian and non-Abelian symmetries by sampling symmetry quantum numbers. We benchmark the method, realized as both stochastic matrix product states (stoMPS) and stochastic projected entangled pair states (stoPEPS), on large-scale quantum lattice models. Using stoMPS, we accurately simulate the square-lattice Hubbard model on cylinders up to width $W=10$, and study the triangular-lattice Hubbard model down to $T/t = 1/64$. We reveal scalar chiral order at half filling and kinetic ferromagnetism upon electron doping, with the latter directly compared to ultracold atom experiments. We further extend the stoMPS method to compute finite-temperature quantum dynamics, as demonstrated by the optical conductivity of the Hubbard model. We showcase the generalization of stoPEPS on up to $20\times20$ square-lattice quantum Ising model, where accurate results are obtained near its quantum critical point. Our method combines high sampling efficiency with full symmetry acceleration. The resulting state-of-the-art framework captures both equilibrium and dynamical properties down to ultralow temperatures, thereby providing a powerful theoretical tool for correlated quantum systems, particularly fermion Hubbard quantum gas microscopy.

cond-mat.str-el↗