arXiv subjects

Lu Chen

Publications and source records attributed to Lu Chen.

At least 19 recordsLinked to original sources

arXiv2026

Agentic Data Sampling Framework for Multilingual ASR via Linguistic-Aware Error-Driven Optimization

Multilingual automatic speech recognition (ASR) has achieved rapid progress with large-scale pretrained models and multilingual speech corpora. Previous studies have proved the pivotal role of data sampling in multilingual ASR performance. However, existing sampling strategies largely rely on time-consuming manual, heuristic-based designs. To address this, we propose a linguistic-aware agentic data sampling framework that introduces a data sampling LLM agent into the multilingual ASR training loop. The agent is driven by a linguistically grounded rubric that analyzes ASR errors from phonological, morphological, and syntactic perspectives, and uses the resulting feedback together with training statistics, budget constraints, and algorithm profiles to select domain-level, sample-level, and token-level sampling strategies. This design turns evaluation results into adaptive data-selection decisions rather than treating WER/CER as final endpoints only. We conduct experiments on twelve Common Voice languages using Whisper-medium as the ASR backbone. The results show that our method achieves performance better than full-data training while using less training data and lower computational resources, and substantially accelerates training convergence compared to conventional sampling strategies.

cs.CL↗

arXiv2026

RxnExBench: Benchmarking Vision-Language Models for Structured Chemical Reaction Extraction from Scientific Literature

Screen Tables and Substrate Scopes are information-dense figures widely used in synthetic chemistry literature, yet automated extraction of structured reaction data from them remains challenging. We introduce RxnExBench, a multimodal benchmark for fine-grained reaction-level evaluation. Built from high-impact chemistry journals, it contains 3,565 expert-annotated reactions and 2,361 molecular structures. The dataset adopts a unified schema with 15 reaction roles and 26 core fields, supporting downstream AI-for-Chemistry tasks such as retrosynthetic planning, yield prediction, and reaction condition recommendation. RxnExBench further provides a reasoning-driven difficulty taxonomy, a modular annotation pipeline, and the Oracle-MolRec protocol, which disentangles molecular recognition from higher-level reasoning. Evaluation of 9 state-of-the-art VLMs shows that performance degrades substantially as schema complexity and reasoning demands increase. Under the Oracle-MolRec setting, results further reveal that accurate molecular recognition alone is insufficient, with multimodal integration and chemistry-aware reasoning remaining the primary bottlenecks. We release all data and code to facilitate future research in chemistry-aware multimodal understanding and AI-for-Chemistry dataset construction. The dataset and code will be released at https://github.com/opendatalab/RxnExBench.

cs.AI↗

arXiv2026

Sharp Moser-Onofri inequality on the CR sphere under zero-moment constraints and its application to the stability of the CR Moser-Onofri inequality

In this paper, we establish the sharp Moser--Onofri inequality in CR sphere under zero-moment constraints. By combining a decomposition into low and high spherical harmonics with the known CR localization estimates, we first prove the existence of extremals of the subcritical Moser--Onofri inequality under zero-moment constraints. We then derive their Euler--Lagrange equation, establish their smoothness, and eliminate the moment multiplier by means of the infinitesimal CR conformal action. Using the summation identity (4.9) on the Hardy space $H^2(B^n)$, we show that the extremals of subcritical Moser-Onofri inequality in CR sphere under zero-moment constraints are constants. By a limiting argument, we deduce the sharp Moser--Onofri inequality in CR sphere under zero-moment constraints. As an application, we establish the stability of Moser--Onofri inequality in CR sphere with the best stability constant.

math.AP↗

arXiv2026

Cognitively Decoupled Meta-Induction for Out-of-Distribution Generalization

In-context learning (ICL) tackles downstream tasks conditioned on demonstrations. This paradigm relies heavily on neighbor-based matching between queries and demonstrations, limiting out-of-distribution (OOD) performance. For better generalization, rule-based inductive reasoning derives task-level regularities, which we term task-level inductive representations (TIR), before directly answering. However, existing methods lack generalizable inductive ability across diverse tasks. Inspired by meta-learning, we propose Cognitive-Decoupling meta-Induction (CoDI) for TIR-based inductive reasoning on general tasks. CoDI comprises a Cognitive-Decoupling Framework for query-blind TIR induction, and a Dual-Loop Meta-Inductive Learning method for cross-task inducer optimization. Empirically, CoDI achieves substantial performance gains on five ID and OOD benchmarks, and preserves OOD gains without explicit induction at inference. Code is available at https://github.com/Yuqian-Wang2277/codi-meta-induction.

cs.LG↗

arXiv2026

Hydrogen Ion-Cut Single-Crystal Diamond Membranes for Wide and Ultrawide Bandgap Semiconductor Heterointegration

Single-crystal diamond membranes (SCDms) were created using a hydrogen ion-cut process combined with a two-step post-implantation annealing, electrochemical etching, and micro-transfer printing. Monte Carlo simulations comparing H, He, C, and O implantation predicted a narrower vacancy-distribution profile for hydrogen under the implantation conditions examined. Raman spectroscopy was used to evaluate implantation-induced lattice damage and its recovery during annealing and membrane release, while optical bandgap analysis provided further evidence of restoration of the diamond lattice structure. After delamination and removal of residual graphitic material, freestanding SCDms with improved crystalline quality were obtained. The SCDms were subsequently transferred to GaN, SiC, AlGaN, and AlN substrates without an adhesive layer. The successful formation of diamond/WBG and diamond/UWBG heterostructures demonstrates the feasibility of integrating the SCDms with a range of semiconductor platforms. Overall, these results establish hydrogen implantation as an effective route for producing high-quality SCDms for future diamond-based heterogeneous integration applications.

cond-mat.mtrl-sci↗

arXiv2026

The cost of connectivity: Network tradeoffs in subway expansion and historic street preservation

As cities become increasingly interconnected, expanding regional mobility networks can come at the cost of local social networks. This study examines such tradeoffs in the case of Miaoqianzhijie, a historic street in Guangzhou, China, which was planned to be demolished for building a new interchange subway at Line 10. We introduce a network tradeoff framework that analyzes subway lines as spatial networks and his-toric streets as social networks within a single analytic frame. Empirically, we combine a participatory GIS-based social network survey with origin-destination (OD) flow analysis of Guangzhou's subway system. Our findings show that Miaoqianzhijie functions as a dense corridor of social infrastructure that support everyday routines, social ties, mutual aid, and symbolic place attachment for residents, workers, and frequent visitors. Our OD analysis found that Line 10's mobility benefits are unevenly distributed and the station closest to the demolition area sees only modest improve-ments. These findings reveal an asymmetric network tradeoff that highlights concentrated social connectivity losses versus diffuse and limited local mobility gains. This study contributes a case study and network-based methodology for representing such tradeoffs and highlights how these representations can inform more equitable planning debates around subway expansion and historic street preservation.

physics.soc-ph↗

arXiv2026

ARGUS: An Agenda-Driven Framework for Reliable User Simulation

High-fidelity user simulators are essential for evaluating and training agents in interactive environments. However, current LLM-based simulators struggle with reliability issues, such as goal omission and policy non-compliance. We propose ARGUS (Agenda-dRiven Goal-oriented User Simulator), a neuro-symbolic framework that maintains the user's goal state using a dynamic agenda stack manipulated via symbolic operations. By rigorously decoupling state tracking from natural language generation, ARGUS facilitates strict goal progression and constraint-aligned interactions. For cost-efficient deployment, we distill ARGUS into compact models using a unified pipeline: evaluator-guided SFT followed by online RL with dense step-level rewards. To rigorously evaluate behavioral fidelity, we introduce a rubric-based metric Composite Behavioral Score (CBS). Evaluations on $τ^3$-Bench and MultiWOZ demonstrate the significant superiority of the ARGUS architecture and the efficacy of our training pipeline. Code and data will be released soon.

cs.AI↗

arXiv2026

Optimizers for the finite-rank fractional Lieb--Thirring inequality

The finite-rank fractional Lieb--Thirring inequality provides an estimate on the Riesz sum of the first $N$ negative eigenvalues of the fractional Schrödinger operator $(-Δ)^s-V$ in terms of the $L^{γ+\frac{d}{2s}}(\mathbb R^d)$ norm of the potential $V$, where the Riesz exponent $γ>\max\left\{0,1-d/(2s)\right\}$. We prove the existence of optimizers for every finite rank by establishing a result on the spectrum in the case of dichotomy. We further characterize the spectral gap and qualitative properties of optimizers, including the Euler--Lagrange equation, regularity, and asymptotic behavior at infinity. In particular, under the condition $γ+{d}/{(2s)}\geq 3+{4s}/{d}$, we show that all normalized maximizing sequences are precompact modulo translations. Moreover, we derive a strict binding inequality for the optimal constants in the finite-rank setting, which implies that the optimal fractional Lieb--Thirring constant, if it exists, cannot be attained by a potential with finitely many negative eigenvalues.

math.AP↗

arXiv2026

Quantization analysis of Moser-Trudinger equations in the Poincaré disk and applications

In this paper, we first establish the quantitative properties for positive solutions to the Moser-Trudinger equation in the two-dimensional Poincaré disk $\mathbb{B}^2$: \begin{equation*}\label{mt1} \left\{ \begin{aligned} &-Δ_{\mathbb{B}^2}u=λue^{u^2},\ x\in\mathbb{B}^2, &u\to0,\ \text{when}\ ρ(x)\to\infty, &\|\nabla_{\mathbb{B}^2} u\|_{L^2(\mathbb{B}^2)}^2\leq M_0, \end{aligned} \right. \end{equation*} where $0<λ<\frac{1}{4}=\inf\limits_{u\in W^{1,2}(\mathbb{B}^2)\backslash\{0\}}\frac{\|\nabla_{\mathbb{B}^2}u\|_{L^2(\mathbb{B}^2)}^2}{\|u\|_{L^2(\mathbb{B}^2)}^2}$, $ρ(x)$ denotes the geodesic distance between $x$ and the origin and $M_0$ is a fixed large positive constant (see Theorem \ref{thm1}). Furthermore, by carefully analyzing the asymptotic expansion of the Dirichlet energy $\|\nabla_{\mathbb{B}^2} u\|_{L^2(\mathbb{B}^2)}^2$ as $λ\to 0$, and establishing that \[ \lim_{\substack{λ\to 1/4}} \int_{\mathbb{B}^2} |\nabla_{\mathbb{B}^2} u_λ|^2 \, dV_{\mathbb{B}^2} = +\infty \] for~the ~nonzero~least energy solution, we are able to prove the existence of a critical point for the Moser-Trudinger functional $F(u) = \int_{\mathbb{B}^2} (e^{u^2} - 1) \, dV_{\mathbb{B}^2}$ under the constraint $\int_{\mathbb{B}^2} |\nabla_{\mathbb{B}^2} u|^2 \, dV_{\mathbb{B}^2} = Λ$ for $Λ\in (4π, +\infty)$, up to some Möbius transformation. This result is substantially different from the non-existence of critical points for the Moser-Trudinger functional in the Euclidean disk. Finally, by applying a local Pohozaev identity around the origin and exploiting the uniqueness of solutions to the Cauchy initial value problem for ODEs, we prove that the Moser-Trudinger equation has a unique positive solution up to some Möbius transformation when $λ$ is close to $0$.

math.AP↗

arXiv2026

FlowCred: Hierarchical Credit Assignment and Functional Coordination for Joint Text-to-Video Post-Training

Prompt optimization and generator post-training are two complementary ways to improve text-to-video (T2V) generation, and recent systems increasingly train both modules jointly. However, optimizing a prompt policy and a video generator from the same terminal reward creates two ambiguities: prompt quality is confounded with stochastic generation, and the two modules may realize overlapping local improvements. We introduce FlowCred, a two-level framework for joint T2V post-training. First, hierarchical credit assignment uses matched prompt-seed rollout grids to separate prompt-level credit from prompt-conditioned generator residuals under shared randomness. Second, functional coordination probes alternative current-policy prompts at states from actual generation trajectories, providing generator-aware evidence for prompt optimization while selectively attenuating generator updates already supported by conditioning changes. Across multiple T2V benchmarks, FlowCred consistently improves matched joint-training baselines. On LTX-2.5, FlowCred improves a strictly matched joint-training control from 61.47 to 64.46 on T2V-CompBench and from 55.58 to 58.08 on StoryEval, without increasing decoded-video or reward-model-call budgets. Factorial and realized-update analyses show complementary contributions from prompt shaping and generator routing, while human evaluation and transfer across backbones and updater families provide further supporting evidence.

cs.CV↗

arXiv2026

VARI: Robust Text-to-Video Generation via Group-Structured Intent Alignment

Modern text-to-video (T2V) models generate visually compelling videos, yet meaning-preserving changes in prompt wording can substantially alter semantic realization and overall quality. Existing prompt rewriting modifies the inference input, whereas conventional post-training constructs supervision from one prompt formulation at a time; neither explicitly optimizes reliability across equivalent forms. We introduce VARI (Verified Alignment over Representations of Intent), a group-structured framework that treats automatically verified formulations of the same visual intent as a shared supervision unit. VARI builds matched form-seed rollout pools, cross-scores each video against all forms, and applies a wording-independent structured intent audit to identify valid realizations and hard semantic failures. Group-Consistent Distillation (GCD) aligns every verified form with interchangeable valid realizations, while Form-Contrastive Flow-DPO (FC-DPO) learns group-level preferences from audited positive-negative pairs, prioritizing same-seed, different-form contrasts. On HunyuanVideo-1.5, VARI attains the highest T2V-CompBench, VBench, and StoryEval scores among the evaluated methods. On prompt-form extensions of T2V-CompBench and VBench, it improves weakest-form alignment over the strongest prompt-level baselines by 2.45 and 3.55 points and reduces cross-form drop by 30.5% and 32.2%, respectively, while preserving visual quality. Human evaluation, natural-caption controls, prompt-enhancement experiments, diversity analyses, and transfer to LTX-2.3 further support the robustness and transferability of these gains, while VARI requires no additional inference-time module.

cs.CV↗

arXiv2026

Attainment of the sharp stability constant of Beckner conformal log-Sobolev inequality

The existence of minimizer for stability of the fractional Sobolev inequality and the stability of Beckner conformal Log-Sobolev inequality (the endpoint case of fractional Sobolev inequality) have been established by König and Chen-Lu-Tang. The existence of minimizer for stability of the Log-Sobolev inequality is unsolved. In this paper, we solve this problem. For every $n\geq2$, we prove that the sharp $L^2$-stability constant of Beckner conformal log-Sobolev inequality on $\mathbb S^n$ is attained and is strictly less than $\frac{2\log2}{n}$. Moreover, the normalized stability constant of the fractional Sobolev inequality converges to the stability constant of the log-Sobolev inequality as $s\to0^+$. After Hersch normalization, every sequence of $L^{p_{s_k}}$-normalized minimizers for stability of the fractional Sobolev inequality, with $s_k\to0^+$, has a subsequence converging strongly in $L^2(\mathbb S^n)$ to an $L^2$-normalized minimizer of stability of the log-Sobolev inequality.

math.AP↗

arXiv2026

Asymptotic Neutrality for the Ionization Conjecture in the Two-Center Thomas-Fermi-Dirac-von Weizsäcker Theory

The ionization conjecture for the Thomas-Fermi-Dirac-von Weizsäcker model asks how many electrons a given nuclear configuration can bind. So far, only the single-center case has been settled by Frank, Nam, and Van Den Bosch, while the original molecular case has still remained open. We advance this conjecture in the two-center case. We prove that for two distinct nuclei of total charge $Z$, the number $N$ of bound electrons satisfies $N\le Z+C Z^{153/154}(1+\log Z)^{3/8}$, with constants independent of the nuclear positions and the nuclear charges. The proof relies on a comparison with the neutral Thomas-Fermi density and on localization and exterior $L^1$-estimates coupled through the screened potential. Our results could be extended to molecules with any fixed number $K$ of nuclei.

math-ph↗

arXiv2026

Gradient rearrangement estimates for singular and degenerate anisotropic $p$-Laplace equations

In this paper, we prove global pointwise estimates on the decreasing rearrangement $u^{\ast}$ and $[H(\nabla u)]^{\ast}$ of SOLAs $u$ of an elliptic Poisson equation driven by the anisotropic $p$-Laplace operator $Δ_p^H$ for given $L^1$-data and equipped with either homogeneous Dirichlet or Neumann boundary conditions. Here, we assume that the dimension $n\ge 2$, $1<p<n$ and $H$ is norm on $\R^n$. In the degenerate range $2 \leq p <n$, this gives an anisotropic extension and a full-interval completion of the rearrangement theory of Cianchi-Maz'ya [JEMS, 2014]. In particular, we obtain a new rearrangement estimate for gradients by decomposing the data for singular range $1<p<2$, where the nonlinear interpolation method aren't available. Moreover, the endpoint weak-type gradient estimate is obtained for all $1<p<n$. In the range $p_*(n)<p<n$ with $p_*(n):=\frac{3n-2}{2n-1}$, we prove a stability estimate for differences of gradients. These results extends the known ones in the literature to the singular range $p_*(n)<p<2$.

math.AP↗

arXiv2026

Determining Schrodinger Potentials on Riemannian Manifolds

We prove that the Dirichlet-to-Neumann map uniquely determines a smooth real Schrödinger potential on any prescribed compact connected smooth Riemannian manifold of dimension at least three with nonempty smooth boundary. The proof combines local continuation of Green kernels with uniform estimates for the associated integral representations on shrinking spheres. These estimates yield local recovery of the potential, and a continuation argument extends this recovery throughout the manifold.

math.AP↗

arXiv2026

Observable sets for the Schrödinger equation with potential $\lvert x\rvert$

In this paper, we study observable sets for the one-dimensional Schrödinger equation associated with the operator $H=-d^2/dx^2+|x|$ on $L^2(\mathbb R)$. We prove that a measurable set is observable at every positive time if and only if it is thick. For each fixed observation time, the observability constant is uniform over all measurable sets with prescribed thickness parameters. The theorem answers the question concerning the relationship between thick and observable set raised by Gengsheng Wang [12] for this anharmonic oscillator.

math.AP↗

arXiv2026

PACE: Stage-Consistent Long-Horizon Robot Manipulation via Progress-Aligned Context for Execution

Demonstration-conditioned policies provide a natural interface for specifying robot behavior, yet long-horizon manipulation remains difficult when visually similar states recur across different stages or when demonstrations and executions proceed at different speeds. We identify the resulting failure mode as stage confusion and introduce Progress-Aligned Context for Execution (PACE), a stateful method that continually reinterprets a complete demonstration according to realized execution progress. PACE compresses the demonstration into ordered multimodal prompt tokens and uses training-only dual-edge attention supervision to expose its latent stage structure. During execution, an episode-local fast-weight memory causally encodes realized action-observation transitions and modulates prompt cross-attention, producing a progress-aligned context for a unified diffusion action expert without test-time stage labels or stage-specific policies. PACE improves success from 88.9% to 94.0% on LIBERO-Gen Goal Chain, from 79.1% to 83.3% on Spatial Combination, and from 33.3% to 73.3% on the two-step Block Routing tasks. Failure analysis further indicates that structured demonstration alignment and causal execution memory jointly mitigate stage confusion.

cs.RO↗

arXiv2026

Disambiguate Gripper State in Grasp-Based Tasks: Pseudo-Tactile as Feedback Enables Pure Simulation Learning

Grasp-based manipulation tasks are fundamental to robots interacting with their environments, yet gripper state ambiguity significantly reduces the robustness of imitation learning policies for these tasks. Data-driven solutions face the challenge of high real-world data costs, while simulation data, despite its low costs, is limited by the sim-to-real gap. We identify the root cause of gripper state ambiguity as the lack of tactile feedback. To address this, we propose a novel approach employing pseudo-tactile as feedback, inspired by the idea of using a force-controlled gripper as a tactile sensor. This method enhances policy robustness without additional data collection and hardware involvement, while providing a noise-free binary gripper state observation for the policy and thus facilitating pure simulation learning to unleash the power of simulation. Experimental results across three real-world grasp-based tasks demonstrate the necessity, effectiveness, and efficiency of our approach.

cs.RO↗