arXiv subjects

Meng Zhao

Publications and source records attributed to Meng Zhao.

At least 19 recordsLinked to original sources

arXiv2026

GeoSim: Benchmarking Multi-Field Neural Operators for Geodynamics

Learning reliable surrogates for geodynamic models requires benchmarks that resolve coupled fields and temporal stability. We present GeoSim, a benchmark of high-resolution two-dimensional mantle-plume simulations generated with I2VIS, comprising 237 trajectories across two parameter regimes, with six coupled fields, physical timestamps, and trajectory-level splits. GeoSim defines short-window prediction and autonomous rollout, and evaluates representative neural operators field by field. Across models, we observe a pronounced scalar--velocity gap: scalar fields are reconstructed accurately while velocity errors remain large, and short-window accuracy does not imply stable rollout. We also evaluate MFGNO (Multi-Field Graph Neural Operator), a joint multi-field graph baseline designed for this benchmark; on GeoSim-Plume1, under a fixed per-field objective, it reduces horizontal and vertical velocity errors by 58.4% and 38.5% relative to an independent-field graph baseline, while an explicit field-difference input shows no measurable benefit at this scale. GeoSim provides a reproducible benchmark for field-wise diagnosis and long-horizon evaluation of geodynamic surrogates.

physics.geo-ph↗

arXiv2026

AdMuse: Skill-Guided Creative Advertising Image Generation

An outstanding creative advertising image should both attract attention and communicate a product's positioning. How to choose a persuasive conceptual bridge before rendering begins is challenging for advertising generation. We present AdMuse, a skill-guided multi-agent framework that distills high-quality advertisements into reusable creative skills for cross-domain advertising strategy. Rather than retrieving isolated exemplars, AdMuse abstracts recurring bridge patterns into a skill library that can guide new product briefs. At generation time, AdMuse coordinates skill routing, multi-candidate ideation and selection, and VLM-guided visual refinement: diverse skills propose alternative concepts, relative judgment selects the most promising direction, and vision-language feedback helps realize the selected intent in the final image. This design treats advertising creativity as a structured decision process rather than a single prompt-writing step. Experiments across diverse product and brand positionings show that AdMuse produces visually compelling advertisements that faithfully communicate product positioning.

cs.CV↗

arXiv2026

Cross-Dimension and Cross-Modality Relational Tensor Representation Learning for Multimodal Health Data Modeling

Multimodal health data integrate information from heterogeneous clinical sources, creating substantial opportunities for data-driven decision support. Realizing these opportunities, however, requires effectively modeling complex dependencies that arise both within individual modalities and across modalities. This paper proposes Relational Tensor Representation (RTR), a unified framework that jointly learns dimension-specific tensor representations and cross-dimensional and cross-modality relationships. RTR introduces learnable relational bridge matrices that maintain relational information while preserving the structural identities of tensor dimensions and modalities. A structured fusion mechanism further integrates factor-level and relationship-level information into compact representations for downstream prediction. Experiments on simulated data and two multimodal neurophysiological datasets, covering emotion recognition and brain activity classification, demonstrate that RTR consistently outperforms representative tensor learning, multimodal fusion, and Transformer-based baselines. These results highlight the value of explicitly modeling cross-dimensional and cross-modality relationships for learning informative representations from heterogeneous multimodal tensor data. The source code is planned for public release to support reproducibility and future research.

cs.LG↗

arXiv2026

Polynomial mixing for infinite-energy solutions of the white-forced Navier--Stokes equation in an infinite pipe

We study the long-time behavior of the infinite-energy solutions to the white-forced Navier--Stokes equation in an infinite pipe $\mathbb{R}\times (-1,1)$ with no-slip boundary conditions. The equation is globally well-posed in uniformly local spaces satisfying the divergence-free condition. Under suitable degeneracy and spatial decay assumptions on the noise, we prove the uniqueness of the invariant measure and establish polynomial mixing in the dual-Lipschitz metric. This provides an affirmative answer to Open Problem 1 posed by Kuksin and Shirikyan \cite{KS17} in the case of zero cross-sectional flux.

math.AP↗

arXiv2026

NarrativeSteward: Coordinating Delegation, Guidance, and Verification in Agent-Assisted Interactive Narrative Authoring

Autonomous AI agents can turn authors' goals into interactive narratives by independently organizing and carrying out generation and revision. As agents generate and revise extensive content, authors struggle to grasp its overall structure, local details, and relationships, complicating continued guidance. We present NarrativeSteward, an authoring environment that organizes outlines, worldbuilding, and narrative graphs as linked artifacts for agent implementation and author guidance. Agent dialogue and project-wide structural review help authors understand the evolving work and guide local and cross-layer revisions, while change records and execution verification help authors assess the resulting work. Technical tests validated the system's change records, recovery mechanisms, and execution diagnostics. In a 12-participant within-subject study, NarrativeSteward supported easier formulation of revision requests and inspection of changes, and greater perceived understanding of changes and story structure, than general-purpose agents. Qualitative findings show how reviewing the work and feedback helps authors develop requirements and guide subsequent delegation. We open-source NarrativeSteward at https://github.com/Tencent/NarrativeSteward.

cs.HC↗

arXiv2026

GradAgent: A Knowledge-Guided Multi-Agent System for Structure-Preserving Gradient-Flow Computation with an Application to Multicomponent Vesicle Dynamics

High-order differential operators and nonlinear coupling make it challenging to construct conservative and energy-stable schemes for coupled gradient-flow systems. We present GradAgent, a knowledge-guided multi-agent system that coordinates three agents across model analysis, algorithm design and proofs, and numerical implementation and validation. Independent audits strengthen reliability by uncovering mathematical errors and proof gaps, guiding revisions, and maintaining consistency across stages. In Reconstruction Mode, GradAgent reconstructs 20 published studies and organizes audited knowledge in an extensible knowledge graph (KG), GradAgent-KG, linking model structures, discretization strategies and proofs, implementations, and numerical evidence. In Design Mode, the agents assess the applicability of retrieved knowledge and develop new schemes informed by relevant discretization strategies. Applied to the fully coupled multicomponent vesicle phase-field-fluid model, GradAgent yields three first-order and three second-order schemes across three algorithmic families, including four linear, decoupled schemes. Under stated assumptions, all six schemes conserve membrane component mass and vesicle volume and dissipate their respective temporally discrete energies unconditionally. Comparisons with and without GradAgent-KG show that it promotes diversity in structure-preserving scheme design for this target model. Numerical tests confirm second-order spatial accuracy, the expected temporal orders, conservation, and temporally discrete energy dissipation, while three-dimensional shear-flow simulations agree qualitatively with experiments. These results demonstrate GradAgent's ability to combine reusable knowledge, coordinated reasoning, and independent auditing to develop and validate structure-preserving algorithms for complex coupled systems.

math.NA↗

arXiv2026

Domain-Division based Progressive Learning for Source-Free Domain Adaptation

With growing privacy and portability concerns, source-free domain adaptation requires only a source pre-trained model and an unlabeled target domain, allowing for effective adaptation to the target data. Most existing self-training methods focus on selecting and exploiting samples with reliable predictions, often neglecting others. Inspired by the finding that deep models learn clean samples faster than noisy ones, we propose a domain-division based progressive learning method named DPL. Specifically, our approach consists of two alternating stages, each beginning with the division of the target domain into easy-to-adapt and hard-to-adapt subdomains based on adaptation difficulty, followed by neighborhood-based pseudo label assignment. In stage one, we enhance classification accuracy through uncertainty-aware self-training and alignment of corresponding classes between subdomains. Stage two then applies tailored learning strategies to each subdomain, starting with consistency learning on the easy-to-adapt samples and progressing to utilizing local structural information for the more challenging ones, thereby mining the intrinsic properties of the target data. Extensive experiments on several widely used benchmarks validate the effectiveness of our approach, demonstrating superior performance compared to state-of-the-art methods. Our code is available at https://github.com/iamjingli/DPL.

cs.CV↗

arXiv2026

LFM: Leveraging Foundation Models for Source-Free Universal Domain Adaptation

Source-free universal domain adaptation (SF-UniDA) adapts a pre-trained source model to an unlabeled target domain under both covariate and label shifts, without access to source data. However, existing SF-UniDA methods rely on inefficient techniques such as threshold tuning and clustering. Foundation models (FMs), known for their generalization and zero-shot capabilities, remain underexplored in SF-UniDA. In this paper, we propose a framework that leverages foundation models (LFM) for SF-UniDA. We use a vision-language model (VLM) to compute similarities between target samples and text labels, including those for unknown classes generated by prompting a large language model. The label shift type is determined by analyzing the coefficient of variation of a similarity-based sample-level score. Unknown samples are identified using a binary Gaussian mixture model fitted to another similarity-based metric. Under a consensus strategy, the pseudo-labels generated by the VLM are refined by the target model initialized with the pre-trained source model, integrating knowledge from both the source domain and foundation models. Finally, these refined pseudo-labels are used to train the target model. Extensive experiments across all possible label shifts and multiple benchmarks demonstrate the effectiveness and superiority of our proposed LFM framework. Our code is available at https://github.com/iamjingli/LFM.

cs.CV↗

arXiv2026

DaV-Gen: End-to-End Generative Retrieval via Draft-and-Verify

Mainstream industrial information retrieval systems (e.g., search and recommendation) are usually built upon Multi-Stage Cascade Architectures (MCAs), which balance effectiveness and efficiency through a coarse-to-fine ``retrieval-ranking'' pipeline. However, the optimization objectives across different stages are substantially inconsistent, propagating or even amplifying the early-stage errors that ultimately degrade the quality of final results. While emerging end-to-end generative models offer a potential solution by unifying the pipeline, their online serving performance is severely hindered by the auto-regressive process inherited from the standard decoder-only structure. To bridge this gap, we introduce \textbf{DaV-Gen}, a novel unified solution designed to fundamentally refactor the paradigm for both search and recommendation via a ``Draft-and-Verify'' mechanism. Inspired by the process used by speculative decoding, our framework redesigns the generation task into two synergistic operations within a single model. During training, the model is concurrently optimized for both candidate drafting and fine-grained verification. This is achieved by a composite loss function that jointly trains the model on two distinct but related objectives: 1) a contrastive loss that structures the embedding space for efficient drafting, and 2) a fusion loss that combines generative likelihood with vector similarity to produce a superior verification score. This integrated training strategy equips the model with dual capabilities. At inference time, it first performs highly efficient vector-based drafting to generate a candidate set, and then verifies these candidates using the more powerful fused scoring function, thereby achieving both the speed of sparse drafting and the precision of advanced generative models within a unified, end-to-end architecture.

cs.IR↗

arXiv2026

Delayed blow-up by transport noise for the 3D Navier-Stokes equation with Navier-slip boundary conditions

We study the vorticity formulation of the 3D Navier-Stokes equation driven by transport noise in a periodic channel with Navier-slip boundary conditions. We consider both non-degenerate transport noise and degenerate tangential transport noise. For any prescribed $T>0$ and $ε>0$, we prove that, by choosing the noise intensity sufficiently large and concentrating the noise on sufficiently high modes, the solution exists up to $T$ with probability at least $1-ε$. A main contribution of this work is to identify and analyze the interaction between enhanced dissipation induced by transport noise and physical boundary effects. The no-flux condition breaks the isotropy of the noise and changes the scaling limit of the Itô-Stratonovich corrector. In the non-degenerate case, a boundary feedback term appears in the limiting effective operator; in the degenerate case, the limiting operator is a nonlocal anisotropic tangential dissipation. The proof is based on a combination of a boundary correction operator, a Meyers-type estimate, a scaling-limit analysis of the Itô-Stratonovich corrector, and resolvent estimates for the deterministic limiting equations.

math.AP↗

arXiv2026

Distribution and Evolution of the Debris Cloud from the Fragmentation of Intelsat 33E

The breakup of Intelsat 33E on 19 October 2024 posed a potential risk to satellites in the Geostationary Earth Orbit (GEO). This study analyzes the evolution and distribution of these fragments using a probabilistic approach. The initial distribution of the fragments, derived from the NASA Standard Breakup Model, indicates the generation of 4,393 fragments larger than 1 centimeter. The spatial propagation of these fragments is modeled analytically in the Earth-Centered Earth-Fixed reference frame, showing the formation of high-density ring structures in the equatorial plane from 24 hours to 28 days after the breakup. The orbits of 36 cataloged fragments are retrieved and compared with the probability density. Furthermore, Monte Carlo simulations validate the probabilistic model and highlight its efficiency in capturing low-probability events. Collision risks to other GEO satellites are assessed, showing that the top 10\% of satellites encounter a collision probability of up to $10^{-8}$ after 28 days. Satellites near the equatorial plane are at higher risk, whereas those with higher inclinations are less affected. These findings underscore the need for enhanced monitoring and mitigation strategies for GEO breakup events, given the challenges in detecting small fragments.

astro-ph.EP↗

arXiv2026

STRIDE-ED: A Strategy-Grounded Stepwise Reasoning Framework for Empathetic Dialogue Systems

Empathetic dialogue requires not only recognizing a user's emotional state but also making strategy-aware, context-sensitive decisions throughout response generation. However, the lack of a comprehensive empathy strategy framework, explicit task-aligned multi-stage reasoning, and high-quality strategy-aware data fundamentally limits existing approaches, preventing them from effectively modeling empathetic dialogue as a complex, multi-stage cognitive and decision-making process. To address these challenges, we propose STRIDE-ED, a STRategy-grounded, Interpretable, and DEep reasoning framework that models Empathetic Dialogue through structured, strategy-conditioned reasoning. To support effective learning, we develop a strategy-aware data refinement pipeline integrating LLM-based annotation, multi-model consistency-weighted evaluation, and dynamic sampling to construct high-quality training data aligned with empathetic strategies. Furthermore, we adopt a two-stage training paradigm that combines supervised fine-tuning with multi-objective reinforcement learning to better align model behaviors with target emotions, empathetic strategies, and response formats. Extensive experiments demonstrate that STRIDE-ED generalizes across diverse open-source LLMs and consistently outperforms existing methods on both automatic metrics and human evaluations. Our data and code are publicly available at https://github.com/jicoder-nwpu/STRIDE-ED.

cs.CL↗

arXiv2026

O-ConNet: Geometry-Aware End-to-End Inference of Over-Constrained Spatial Mechanisms

Deep learning has shown strong potential for scientific discovery, but its ability to model macroscopic rigid-body kinematic constraints remains underexplored. We study this problem on spatial over-constrained mechanisms and propose O-ConNet, an end-to-end framework that infers mechanism structural parameters from only three sparse reachable points while reconstructing the full motion trajectory, without explicitly solving constraint equations during inference. On a self-constructed Bennett 4R dataset of 42,860 valid samples, O-ConNet achieves Param-MAE 0.276 +/- 0.077 and Traj-MAE 0.145 +/- 0.018 (mean +/- std over 10 runs), outperforming the strongest sequence baseline (LSTM-Seq2Seq) by 65.1 percent and 88.2 percent, respectively. These results suggest that end-to-end learning can capture closed-loop geometric structure and provide a practical route for inverse design of spatial over-constrained mechanisms under extremely sparse observations.

cs.RO↗

arXiv2026

Platform and Framework for Time-Resolved Nanoscale Thermal Transport Measurements in STEM

Understanding heat transport at the nanometer scale is critical for semiconductor devices, quantum materials, and thermal management of nanostructures, yet direct local measurements of thermal conductivity and heat capacity remain scarce. We developed a laser-excitation system integrated into a scanning transmission electron microscope (STEM) for nanoscale thermal transport measurements using ultra-high-resolution electron energy-loss spectroscopy (EELS). A fiber-coupled laser is introduced via a modified aperture mechanism, enabling flexible holder geometries and large tilt angles without optical elements in the polepiece gap. Synchronization of pulsed laser excitation with an externally gated direct electron detector provides temporal resolution about 50 ns at <10 meV energy resolution. Local temperatures are determined via the principle of detailed balance, and thermal transport parameters are extracted by fitting a forward-time central-space heat diffusion model including radiative losses. For amorphous carbon films, we obtain a thermal conductivity of 1.24 $\frac{W}{m\cdot K}$ and a heat capacity of 821 $\frac{J}{kg\cdot K}$, consistent with literature. This framework enables time-resolved nanoscale measurements of thermal transport in materials and devices.

cond-mat.mtrl-sci↗

arXiv2026

Large-scale Integration of Experimental and Computational Data for 2D Materials

The past decade has seen rapid growth in the number of experimentally realized two-dimensional (2D) materials with diverse chemical and physical properties. However, information on their crystal structure, synthesis routes, and measured or predicted properties, remains scattered across thousands of publications. Here we consolidate this fragmented knowledge by establishing X2DB - an open infrastructure that integrates experimental and computational data on 2D materials. Using extensive literature mining and direct community uploads, we identify 370 unique 2D materials that have been realized in monolayer or few-layer form, and link them to their digital counterparts in computational databases, enabling consistent ab initio characterization of their properties across monolayer, bilayer and bulk forms. We describe the structure and content of the database highlighting its support for community uploads, illustrate how it can be used to generate new scientific insight and introduce a hierarchical classification of the known set of 2D materials. Our work provides a foundation for the integration and cross-fertilization of experimental and theoretical knowledge, opening new avenues for data-driven, predictive synthesis of novel 2D materials.

cond-mat.mtrl-sci↗

arXiv2025

MedualTime: A Dual-Adapter Language Model for Medical Time Series-Text Multimodal Learning

The recent rapid advancements in language models (LMs) have garnered attention in medical time series-text multimodal learning. However, existing contrastive learning-based and prompt-based LM approaches tend to be biased, often assigning a primary role to time series modality while treating text modality as secondary. We classify these approaches under a temporal-primary paradigm, which may overlook the unique and critical task-relevant information embedded in text modality like clinical reports, thus failing to fully leverage mutual benefits and complementarity of different modalities. To fill this gap, we propose a novel textual-temporal multimodal learning paradigm that enables either modality to serve as the primary while being enhanced by the other, thereby effectively capturing modality-specific information and fostering cross-modal interaction. In specific, we design MedualTime, a language model composed of dual adapters to implement temporal-primary and textual-primary modeling simultaneously. Within each adapter, lightweight adaptation tokens are injected into the top layers of LM to encourage high-level modality fusion. The shared LM pipeline by dual adapters not only achieves adapter alignment but also enables efficient fine-tuning, reducing computational resources. Empirically, MedualTime demonstrates superior performance on medical data, achieving notable improvements of 8% accuracy and 12% F1 in supervised settings. Furthermore, MedualTime's transferability is validated by few-shot label transfer experiments from coarse-grained to fine-grained medical data. https://github.com/start2020/MedualTime

cs.LG↗

arXiv2025

Toward Temporal Causal Representation Learning with Tensor Decomposition

Temporal causal representation learning is a powerful tool for uncovering complex patterns in observational studies, which are often represented as low-dimensional time series. However, in many real-world applications, data are high-dimensional with varying input lengths and naturally take the form of irregular tensors. To analyze such data, irregular tensor decomposition is critical for extracting meaningful clusters that capture essential information. In this paper, we focus on modeling causal representation learning based on the transformed information. First, we present a novel causal formulation for a set of latent clusters. We then propose CaRTeD, a joint learning framework that integrates temporal causal representation learning with irregular tensor decomposition. Notably, our framework provides a blueprint for downstream tasks using the learned tensor factors, such as modeling latent structures and extracting causal information, and offers a more flexible regularization design to enhance tensor decomposition. Theoretically, we show that our algorithm converges to a stationary point. More importantly, our results fill the gap in theoretical guarantees for the convergence of state-of-the-art irregular tensor decomposition. Experimental results on synthetic and real-world electronic health record (EHR) datasets (MIMIC-III), with extensive benchmarks from both phenotyping and network recovery perspectives, demonstrate that our proposed method outperforms state-of-the-art techniques and enhances the explainability of causal representations.

cs.LG↗

arXiv2025

Level-3 large deviations for the white-forced 2D Navier-Stokes system in a bounded domain

We study the large deviations principle (LDP) of Donsker-Varadhan type for the white-forced Navier-Stokes system in a bounded domain. Under the assumption that the noise is non-degenerate, we establish level-2 and level-3 LDPs with rate functions given by the Donsker-Varadhan formulas. The proof relies on an improved version of Kifer's criterion, a lift argument inspired from [DV83], an improved abstract result on the large-time asymptotics of generalized Markov semigroups, and a delicate approximation scheme utilizing the resolvent operators of the Markov semigroup.

math.AP↗