arXiv subjects

Zeyu Xia

Publications and source records attributed to Zeyu Xia.

At least 19 recordsLinked to original sources

arXiv2026

ScatterPrism: convergence for generative simulation and inverse problems in particle and nuclear physics

High-fidelity simulations and complex inverse problems, such as detector modeling and unfolding, are computationally intensive bottlenecks across subatomic physics, yet essential for accurate physical interpretation. While Conditional Flow Matching (CFM) offers a robust acceleration approach, we demonstrate its standard training loss is fundamentally misleading. Specifically, utilizing a Jefferson Lab Nuclear Physics (NP) kinematic dataset ($γp \to ρ^0 p \to π^+π^- p$), we expose that CFM loss plateaus prematurely, obscuring ongoing physical refinement. To verify this disconnect is a dataset-agnostic pathology, we introduce ScatterPrism, an efficient generative surrogate evaluated against both the NP data and synthetic stress tests modeling challenging 1D distribution topologies. Coupling these benchmarks, we establish that physics-informed metrics continue improving long after standard loss converges. Consequently, we propose a multi-metric diagnostic protocol to ensure true kinematic fidelity without data memorization. Driven by NP challenges relevant to the forthcoming Electron-Ion Collider (EIC), this unified machinery has strong potential to extend to High-Energy Physics (HEP) applications, such as jet modeling. Furthermore, the framework holds promise for broader domains requiring rigorous generative reliability, including medical imaging, astrophysics, and quantitative finance.

cs.LG↗

arXiv2026

Numerical solutions of an accurate diffuse interface model of the incompressible resistive MHD free surface flow

In this paper, we derive a new model to simulate the incompressible resistive magnetohydrodynamic (MHD) free surface flow. A thermodynamically consistent diffuse interface method is adopted to characterize the moving interface in the modeling process. The formal convergence of the proposed MHD free surface flow model to the sharp interface model is established via a matched asymptotic argument, and the model can be solved without the need for sophisticated free surface capturing schemes. We design a fully decoupled linear finite element scheme that preserves the divergence-free constraint of the magnetic field at a discrete level. The reliability and robustness of the proposed model and algorithm are validated through numerical investigations of the magnetic damping effect on bubble dynamics. In particular, we provide a quantitative numerical comparison of the present results with those obtained from an inductionless MHD model and a sharp interface arbitrary Lagrangian--Eulerian model.

math.NA↗

arXiv2026

VASP Agent: An Agentic Framework for Autonomous First-principles Calculations

Large Language Models (LLMs) are increasingly embedded in agentic frameworks for scientific discovery. First-principles materials computation imposes a demanding standard for autonomy: successful execution depends on internally consistent inputs, supervision of long-running calculations, and verified outputs. Here we present VASP Agent, a coding-agent-centered system that combines reusable domain skills, deterministic tools, workspace-state inspection, runtime evidence, and scientific guardrails to execute multi-step VASP calculations. The system is evaluated across multiple tasks including structural relaxation, bandgap calculation, equilibrium lattice constant determination, and CO/Pt(111) adsorption. VASP Agent completes all evaluated cases, and its computed numerical results are compared with those obtained using pymatgen and other agentic tools. When large deviations occur, the calculation parameters produced by VASP Agent are more appropriate than those produced by LLM-based workflows. Failure analysis shows that errors that terminate fixed pipelines can be diagnosed and recovered under agentic control.

cs.AI↗

arXiv2026

PyTorchFire: A GPU-Accelerated Wildfire Simulator with Differentiable Cellular Automata

Accurate and rapid prediction of wildfire trends is crucial for effective management and mitigation. However, the stochastic nature of fire propagation poses significant challenges in developing reliable simulators. In this paper, we introduce PyTorchFire, an open-access, PyTorch-based software that leverages GPU acceleration. With our redesigned differentiable wildfire Cellular Automata (CA) model, we achieve millisecond-level computational efficiency, significantly outperforming traditional CPU-based wildfire simulators on real-world-scale fires at high resolution. Real-time parameter calibration is made possible through gradient descent on our model, aligning simulations closely with observed wildfire behavior both temporally and spatially, thereby enhancing the realism of the simulations. Our PyTorchFire simulator, combined with real-world environmental data, demonstrates superior generalizability compared to supervised learning surrogate models. Its ability to predict and calibrate wildfire behavior in real-time ensures accuracy, stability, and efficiency. PyTorchFire has the potential to revolutionize wildfire simulation, serving as a powerful tool for wildfire prediction and management.

cs.CE↗

arXiv2026

Accurate identification and measurement of the precipitate area by two-stage deep neural networks in novel chromium-based alloys

The performance of advanced materials for extreme environments is underpinned by their microstructure, including the size and distribution of reinforcing phases. Chromium-based superalloys are a recently proposed alternative to conventional face-centred-cubic superalloys for high-temperature applications, such as Concentrated Solar Power, and their development requires efficient measurement of precipitate volume fraction and size distribution from electron microscopy images. Traditional fixed-threshold image processing is sensitive to background noise, generalises poorly across materials, and requires substantial manual measurement effort. To address these bottlenecks, this study proposes DT-SegNet, an end-to-end two-stage deep learning scheme based on YOLOv5 and SegFormer for object detection and segmentation in electron microscopy images. The approach combines the training efficiency of convolutional neural networks at the detection stage with the segmentation accuracy of a Vision Transformer. Numerical experiments show that DT-SegNet substantially outperforms state-of-the-art segmentation tools offered by Weka and ilastik across metrics including accuracy, precision, recall, and F1-score. The model provides a useful tool for alloy-development microstructure examinations and helps address the large datasets associated with high-throughput alloy development.

cs.CV↗

arXiv2026

FireDataForge: A Unified Framework for Multi-Source Wildfire Data Retrieval and Integration

Wildfire research, modeling, and education require geospatial data from multiple sources that vary in formats, coordinate systems, spatial resolutions, and temporal cadences. This preprocessing burden limits reproducible reuse. We present FireDataForge, an open-source Python framework that automates retrieval and harmonization of 11 wildfire-related sources spanning fire behavior, weather, land cover, vegetation, elevation, built environment, wildland-urban interface, fire history, and satellite imagery. Given an MTBS Event ID, FireDataForge retrieves relevant datasets, aligns them to a common grid, and outputs analysis-ready NumPy arrays with embedded metadata. Batch processing of historical fires demonstrates support for fire behavior simulation, educational visualization, machine learning, and AI-assisted wildfire analysis.

cs.CE↗

arXiv2026

Cosmo3DFlow: Wavelet Flow Matching for Spatial-to-Spectral Compression in Reconstructing the Early Universe

Reconstructing the early universe from the evolved present-day universe is a challenging and computationally demanding problem in modern astrophysics. We devise a novel generative framework, Cosmo3DFlow, designed to address dimensionality and sparsity, the critical bottlenecks inherent in current state-of-the-art methods for cosmological inference. By integrating 3D Discrete Wavelet Transform (DWT) with flow matching, we effectively represent high-dimensional cosmological structures. The Wavelet Transform addresses the ``void problem'' by translating spatial emptiness into spectral sparsity. It decouples high-frequency details from low-frequency structures, and wavelet-space velocity fields facilitate stable ordinary differential equation (ODE) solvers with large step sizes. Using large-scale cosmological $N$-body simulations at $128^3$ resolution, we achieve up to $46\times$ faster sampling than diffusion models. Our results enable initial conditions to be sampled in seconds, compared to minutes for previous methods.

astro-ph.IM↗

arXiv2026

FunctionEvolve: Structure-Guided Symbolic Regression with LLMs

Symbolic regression aims to uncover explicit scientific laws from data. Recent methods use LLMs to guide mutation from background text, which is more directed than random genetic programming. However, exact symbolic recovery requires both semantic guidance and explicit structure, so that domain-informed search are carried out through valid symbolic representation. Current LLM-driven systems remain structure-blind: they select among opaque candidates, lack explicit mechanisms for local mutation, and rely on brittle coefficient fitting that can undervalue correct skeletons. We propose FunctionEvolve, an evolutionary framework using expression trees to organize the whole search: structural summaries promote diverse parent selection, local tree edits preserve useful subexpressions, and structure-aware fitting decomposes, constrains, and simplifies coefficients for more reliable scoring. It uses only elementary function families, without additional domain-specific rules limiting generalization. On the 129-task synthetic subset of LLM-SRBench, FunctionEvolve with \emph{Claude Opus 4.6} recovers 107 exact forms, reaching 82.9% SA@50, 4.5x above same-backbone baselines, and 55.8% SA@1, 3.6x above the strongest previously published top-1 result. Ablations show that structure-visible search is central to reliable recovery, with LLM-guided refinements and structure-aware coefficient optimization serving as essential proposal and scoring mechanisms. We also audit the benchmark and show that collinearity in its materials-science subset creates identifiability issues.

cs.LG↗

arXiv2026

DMAConv: Dual Mask-Adaptive Convolution for Remote Sensing Pansharpening

Pansharpening aims to fuse a high-resolution panchromatic image with a low-resolution multispectral image. Existing deep learning methods, including recent adaptive convolutions, struggle with regional heterogeneity in remote sensing images and often incur prohibitive computational costs. To address these challenges, we propose Dual Mask-Adaptive Convolution (DMAConv), a novel operator that dynamically allocates computational resources based on feature characteristics. DMAConv first employs a lightweight module to generate soft and hard masks. The hard mask separates features into a compact branch for processing redundant information globally and a focused branch that models complex, heterogeneous regions with greater computational investment. The soft mask then preliminarily modulates the input features for both branches. This dual-branch, mask-adaptive design significantly enhances feature representation while minimizing computational overhead. Extensive experiments demonstrate that our method achieves SOTA on a broad array of quantitative benchmarks, with substantially lower parameter counts and the minimal computational cost among adaptive convolution models.

cs.CV↗

arXiv2026

Helix: Evolutionary Reinforcement Learning for Open-Ended Scientific Problem Solving

Large language models (LLMs) with reasoning abilities have demonstrated growing promise for tackling complex scientific problems. Yet such tasks are inherently domain-specific, unbounded and open-ended, demanding exploration across vast and flexible solution spaces. Existing approaches, whether purely learning-based or reliant on carefully designed workflows, often suffer from limited exploration efficiency and poor generalization. To overcome these challenges, we present HELIX -- a Hierarchical Evolutionary reinforcement Learning framework with In-context eXperiences. HELIX introduces two key novelties: (i) a diverse yet high-quality pool of candidate solutions that broadens exploration through in-context learning, and (ii) reinforcement learning for iterative policy refinement that progressively elevates solution quality. This synergy enables the discovery of more advanced solutions. On the circle packing task, HELIX achieves state-of-the-art result with a sum of radii of 2.63598308 using only a 14B model. Across standard machine learning benchmarks, HELIX further surpasses GPT-4o with a carefully engineered pipeline, delivering an average F1 improvement of 5.95 points on the Adult and Bank Marketing datasets.

cs.LG↗

arXiv2025

Training and Inference within 1 Second -- Tackle Cross-Sensor Degradation of Real-World Pansharpening with Efficient Residual Feature Tailoring

Deep learning methods for pansharpening have advanced rapidly, yet models pretrained on data from a specific sensor often generalize poorly to data from other sensors. Existing methods to tackle such cross-sensor degradation include retraining model or zero-shot methods, but they are highly time-consuming or even need extra training data. To address these challenges, our method first performs modular decomposition on deep learning-based pansharpening models, revealing a general yet critical interface where high-dimensional fused features begin mapping to the channel space of the final image. % may need revisement A Feature Tailor is then integrated at this interface to address cross-sensor degradation at the feature level, and is trained efficiently with physics-aware unsupervised losses. Moreover, our method operates in a patch-wise manner, training on partial patches and performing parallel inference on all patches to boost efficiency. Our method offers two key advantages: (1) $\textit{Improved Generalization Ability}$: it significantly enhance performance in cross-sensor cases. (2) $\textit{Low Generalization Cost}$: it achieves sub-second training and inference, requiring only partial test inputs and no external data, whereas prior methods often take minutes or even hours. Experiments on the real-world data from multiple datasets demonstrate that our method achieves state-of-the-art quality and efficiency in tackling cross-sensor degradation. For example, training and inference of $512\times512\times8$ image within $\textit{0.2 seconds}$ and $4000\times4000\times8$ image within $\textit{3 seconds}$ at the fastest setting on a commonly used RTX 3090 GPU, which is over 100 times faster than zero-shot methods.

cs.CV↗

arXiv2025

Modeling and simulation of inductionless magnetohydrodynamic free surface problems with unmatched densities

We propose a new diffuse interface model for simulating an inductionless magnetohydrodynamic (MHD) free surface problem. By using the Onsager's variational principle and the laws of thermodynamics, we derive a thermodynamically consistent system that couples the Cahn--Hilliard equation modeling phase separation, the Navier--Stokes equations governing fluid motion, and a generalized Darcy's law accounting for electromagnetic effects. In contrast to existing diffuse interface MHD models, the proposed model can handle general material properties in practical engineering applications. Furthermore, through asymptotic arguments, we investigate the sharp interface limit, and then demonstrate that the classical sharp interface model can be recovered as the interface thickness approaches zero, theoretically validating the proposed diffuse interface model as an approximate approach. An efficient decoupled, linear, and charge-conservative finite element scheme is designed, and it significantly facilitates the large-scale and accurate numerical simulations involving large parameter ratios. Finally, we present several three-dimensional numerical experiments of magnetic damping effects on bubble dynamics for the demonstration of the capability of the proposed model and method in capturing complex MHD phenomena.

math.NA↗

arXiv2025

Convergence analysis of a second-order SAV-ZEC scheme for the Cahn-Hilliard-Navier-Stokes system

Incorporating the scalar auxiliary variable (SAV) method and the zero energy contribution (ZEC) technique, we analyze a linear and fully decoupled numerical scheme for the Cahn-Hilliard-Naiver-Stokes (CHNS) system. More precisely, the fully discrete scheme combines the marker-and-cell (MAC) finite difference spatial approximation and BDF2 temporal discretization, as well as the Adams-Bashforth extrapolation for the nonlinear terms, based on the SAV-ZEC reformulation. A pressure correction approach is applied to decouple the Stokes equation. Only constant-coefficient Poisson-like solvers are needed in the implementation for the resulting numerical system. The numerical scheme is unconditionally stable with respect to a rewritten total energy functional, represented in terms of one auxiliary variable in the double-well potential, another auxiliary variable to balance all the nonlinear and coupled terms, the surface energy in the original phase variable, combined with the kinematic energy part. Specifically, the error estimate for the phase variable in the $\ell^{\infty}(0,T;H_h^1)\cap\ell^2(0,T;H_h^3)$ norm, the velocity variable in the $\ell^{\infty}(0,T;\ell^2)\cap\ell^2(0,T;H_h^1)$ norm, is derived with optimal convergence rates.

math.NA↗

arXiv2025

SWIFT: A General Sensitive Weight Identification Framework for Fast Sensor-Transfer Pansharpening

Pansharpening aims to fuse high-resolution panchromatic (PAN) images with low-resolution multispectral (LRMS) images to generate high-resolution multispectral (HRMS) images. Although deep learning-based methods have achieved promising performance, they generally suffer from severe performance degradation when applied to data from unseen sensors. Adapting these models through full-scale retraining or designing more complex architectures is often prohibitively expensive and impractical for real-world deployment. To address this critical challenge, we propose a fast and general-purpose framework for cross-sensor adaptation, SWIFT (Sensitive Weight Identification for Fast Transfer). Specifically, SWIFT employs an unsupervised sampling strategy based on data manifold structures to balance sample selection while mitigating the bias of traditional Farthest Point Sampling, efficiently selecting only 3\% of the most informative samples from the target domain. This subset is then used to probe a source-domain pre-trained model by analyzing the gradient behavior of its parameters, allowing for the quick identification and subsequent update of only the weight subset most sensitive to the domain shift. As a plug-and-play framework, SWIFT can be applied to various existing pansharpening models. Extensive experiments demonstrate that SWIFT reduces the adaptation time from hours to approximately one minute on a single NVIDIA RTX 4090 GPU. The adapted models not only substantially outperform direct-transfer baselines but also achieve performance competitive with, and in some cases superior to, full retraining, establishing a new state-of-the-art on cross-sensor pansharpening tasks for the WorldView-2 and QuickBird datasets.

cs.CV↗

arXiv2025

STAIR: Improving Safety Alignment with Introspective Reasoning

Ensuring the safety and harmlessness of Large Language Models (LLMs) has become equally critical as their performance in applications. However, existing safety alignment methods typically suffer from safety-performance trade-offs and the susceptibility to jailbreak attacks, primarily due to their reliance on direct refusals for malicious queries. In this paper, we propose STAIR, a novel framework that integrates SafeTy Alignment with Itrospective Reasoning. We enable LLMs to identify safety risks through step-by-step analysis by self-improving chain-of-thought (CoT) reasoning with safety awareness. STAIR first equips the model with a structured reasoning capability and then advances safety alignment via iterative preference optimization on step-level reasoning data generated using our newly proposed Safety-Informed Monte Carlo Tree Search (SI-MCTS). We further train a process reward model on this data to guide test-time searches for improved responses. Extensive experiments show that STAIR effectively mitigates harmful outputs while better preserving helpfulness, compared to instinctive alignment strategies. With test-time scaling, STAIR achieves a safety performance comparable to Claude-3.5 against popular jailbreak attacks. Relevant resources in this work are available at https://github.com/thu-ml/STAIR.

cs.CL↗

arXiv2025

Optimal ${L^2}$ error estimates for 2D/3D incompressible Cahn--Hilliard--magnetohydrodynamic equations

This paper focuses on an optimal error analysis of a fully discrete finite element scheme for the Cahn--Hilliard--magnetohydrodynamic (CH-MHD) system. The method use the standard inf-sup stable Taylor--Hood/MINI elements to solve the Navier--Stokes equations, Lagrange elements to solve the phase field, and particularly, the Nédélec elements for solving the magnetic induction field. Suffering from the strong coupling and high nonlinearity, the previous works just provide suboptimal error estimates for phase field and velocity field in $L^{2}/Ł^2$-norm under the same order elements, and the suboptimal error estimates for magnetic induction field in $\H(\rm curl)$-norm. To this end, we utilize the Ritz, Stokes, and Maxwell quasi-projections to eliminate the low-order pollution of the phase field and magnetic induction field. In addition to the optimal $Ł^2$-norm error estimates, we present the optimal convergence rates for magnetic induction field in $\H(\rm curl)$-norm and for velocity field in $\H^1$-norm. Moreover, the unconditional energy stability and mass conservation of the proposed scheme are preserved. Numerical examples are illustrated to validate the theoretical analysis and show the performance of the proposed scheme.

math.NA↗

arXiv2025

CAT: A Conditional Adaptation Tailor for Efficient and Effective Instance-Specific Pansharpening on Real-World Data

Pansharpening is a crucial remote sensing technique that fuses low-resolution multispectral (LRMS) images with high-resolution panchromatic (PAN) images to generate high-resolution multispectral (HRMS) imagery. Although deep learning techniques have significantly advanced pansharpening, many existing methods suffer from limited cross-sensor generalization and high computational overhead, restricting their real-time applications. To address these challenges, we propose an efficient framework that quickly adapts to a specific input instance, completing both training and inference in a short time. Our framework splits the input image into multiple patches, selects a subset for unsupervised CAT training, and then performs inference on all patches, stitching them into the final output. The CAT module, integrated between the feature extraction and channel transformation stages of a pre-trained network, tailors the fused features and fixes the parameters for efficient inference, generating improved results. Our approach offers two key advantages: (1) $\textit{Improved Generalization Ability}$: by mitigating cross-sensor degradation, our model--although pre-trained on a specific dataset--achieves superior performance on datasets captured by other sensors; (2) $\textit{Enhanced Computational Efficiency}$: the CAT-enhanced network can swiftly adapt to the test sample using the single LRMS-PAN pair input, without requiring extensive large-scale data retraining. Experiments on the real-world data from WorldView-3 and WorldView-2 datasets demonstrate that our method achieves state-of-the-art performance on cross-sensor real-world data, while achieving both training and inference of $512\times512$ image within $\textit{0.4 seconds}$ and $4000\times4000$ image within $\textit{3 seconds}$ at the fastest setting on a commonly used RTX 3090 GPU.

cs.CV↗

arXiv2025

Optimal error estimates of a second-order temporally finite element method for electrohydrodynamic equations

In this work, we mainly present the optimal convergence rates of the temporally second-order finite element scheme for solving the electrohydrodynamic equation. Suffering from the highly coupled nonlinearity, the convergence analysis of the numerical schemes for such a system is rather rare, not to mention the optimal error estimates for the high-order temporally scheme. To this end, we abandon the traditional error analysis method following the process of energy estimate, which may lead to the loss of accuracy. Instead, we note that the charge density also possesses the "energy" decaying property directly derived by its governing equation, although it does not appear in the energy stability analysis. This fact allows us to control the error terms of the charge density more conveniently, which finally leads to the optimal convergence rates. Several numerical examples are provided to demonstrate the theoretical results, including the energy stability, mass conservation, and convergence rates.

math.NA↗