Searcharxiv⌕ Search

arXiv subjects

Xue Hu

Publications and source records attributed to Xue Hu.

14 recordsLinked to original sources
arXiv2026

SA-Bench: Evaluating Semantic Alignment in LLM-Based Paper Reproduction

LLM agents can generate paper reproduction code, yet often produce scientifically unfaithful implementations. We define this failure mode as semantic drift, where generated code silently diverges from the paper's specifications. We introduce SemanticAlign-Bench(SA-Bench), a diagnostic benchmark covering 30 papers from ICLR, ICML and NeurIPS 2025. For each paper, we decompose its specifications into atomic and verifiable implementation claims, which we call Semantic Alignment Units (SAUs) and evaluate repositories along four diagnostic dimensions spanning numerical, methodological, protocol and ordering drift. In total, we construct 1,491 SAUs across five ML domains and evaluate 12 generator configurations (4 models $\times$ 3 scaffolds). Even the strongest configuration (Claude+PaperCoder) achieves a mean SAU score of only 0.301 out of 1.0, with an overall mean of 0.221 across 360 evaluations. A failure taxonomy reveals that agents attempt most requirements but implement them incorrectly, with implementation mismatch and stubs accounting for the majority of zero-scored claims. Our analysis further indicates that scaffolds optimized for executability provide limited leverage for scientific reproduction; narrowing the gap requires scaffolds that prioritize semantic specification verification. The benchmark, annotations and evaluation pipeline are publicly available.

cs.AI↗
arXiv2026

ReproAgent: Contract-Guided Paper-to-Code Reproduction

Paper-to-code reproduction asks scientific AI agents to turn research papers into executable repositories that preserve the paper's method, protocol and artifacts. This is difficult because the specification is split: explicit paper content such as algorithms, metrics and artifacts is often lost across long agent trajectories, while implicit details such as framework defaults and conventions inherited from related work are absent from the paper. We introduce ReproAgent, a four-stage Prepare--Plan--Generate--Repair pipeline built around a persistent implementation contract with two channels: an implementation-requirement channel that turns paper snippets into code obligations, and a reference-evidence channel that retrieves content and structure evidence from related repositories. Both are bound to work packages, projected into file-level contracts, and consumed across generation and repair. On PaperBench Code-Dev, ReproAgent reaches the highest mean score among same-backbone scaffolds under both Claude-Sonnet-4.5 and Gemini-3-Flash. End-to-end channel ablations and per-paper cases support the contribution of both channels. Code and experimental artifacts are publicly available.

cs.AI↗
arXiv2026

Design of Grid Forming Multi Timescale Coordinated Control Strategies for Dynamic Virtual Power Plants

As the penetration level of distributed energy resources (DERs) continues to rise, traditional frequency and voltage support from synchronous machines declines. This weakens grid stability and increases the need for fast and adaptive control in a dynamic manner, especially in weak grids. However, most virtual power plants (VPPs) rely on static aggregation and plan based resource allocation strategies. These methods overlook differences in device response times and limit flexibility for ancillary services. To address this issue, we propose a dynamic virtual power plant (DVPP) that coordinates heterogeneous resources across multiple time scales using grid forming control. We first contrast grid following and grid forming converters: grid following designs rely on a phase locked loop which can undermine stability in weak grids, whereas our DVPP applies virtual synchronous generator control at the aggregate level to provide effective inertia and damping. Then, we introduce a dynamic participation factor framework that measures each device s contribution through the frequency active power and voltage reactive power loops. Exploiting device heterogeneity, we adopt a banded allocation strategy: slow resources manage steady state and low frequency regulation; intermediate resources smooth transitions; and fast resources deliver rapid response and high frequency damping. Comparative simulations demonstrate that this coordinated, timescale aware approach enhances stability and ancillary service performance compared to conventional VPPs.

eess.SY↗
arXiv2025

Aurora: Architecting Argonne's First Exascale Supercomputer for Accelerated Scientific Discovery

Aurora is Argonne National Laboratory's pioneering Exascale supercomputer, designed to accelerate scientific discovery with cutting-edge architectural innovations. Key new technologies include the Intel(TM) Xeon(TM) Data Center GPU Max Series (code-named Sapphire Rapids) with support for High Bandwidth Memory (HBM), alongside the Intel(TM) Data Center GPU Max Series (code-named Ponte Vecchio) on each compute node. Aurora also integrates the Distributed Asynchronous Object Storage (DAOS), a novel exascale storage solution, and leverages Intel's oneAPI programming environment. This paper presents an in-depth exploration of Aurora's node architecture, the HPE Slingshot interconnect, the supporting software ecosystem, and DAOS. We provide insights into standard benchmark performance and applications readiness efforts via Aurora's Early Science Program and the Exascale Computing Project.

cs.DC↗
arXiv2025

A Joint Planning Model for Fixed and Mobile Electric Vehicle Charging Stations Considering Flexible Capacity Strategy

The widespread adoption of electric vehicles (EVs) has significantly increased demand on both transportation and power systems, posing challenges to their stable operation. To support the growing need for EV charging, both fixed charging stations (FCSs) and mobile charging stations (MCSs) have been introduced, serving as key interfaces between the power grid and traffic network. Recognizing the importance of collaborative planning across these sectors, this paper presents a two-stage joint planning model for FCSs and MCSs, utilizing an improved alternating direction method of multipliers (ADMM) algorithm. The primary goal of the proposed model is to transform the potential negative impacts of large-scale EV integration into positive outcomes, thereby enhancing social welfare through collaboration among multiple stakeholders. In the first stage, we develop a framework for evaluating FCS locations, incorporating assessments of EV hosting capacity and voltage stability. The second stage introduces a joint planning model for FCSs and MCSs, aiming to minimize the overall social costs of the EV charging system while maintaining a reliable power supply. To solve the planning problem, we employ a combination of mixed-integer linear programming, queueing theory, and sequential quadratic programming. The improved ADMM algorithm couples the siting and sizing decisions consistently by introducing coupling constraints, and supports a distributed optimization framework that coordinates the interests of EV users, MCS operators, and distribution system operators. Additionally, a flexible capacity planning strategy that accounts for the multi-period development potential of EVCS is proposed to reduce both the complexity and the investment required for FCS construction. Finally, a case study with comparative experiments demonstrates the effectiveness of the proposed models and solution methods.

eess.SY↗
arXiv2023

Learning Exhaustive Correlation for Spectral Super-Resolution: Where Spatial-Spectral Attention Meets Linear Dependence

Spectral super-resolution that aims to recover hyperspectral image (HSI) from easily obtainable RGB image has drawn increasing interest in the field of computational photography. The crucial aspect of spectral super-resolution lies in exploiting the correlation within HSIs. However, two types of bottlenecks in existing Transformers limit performance improvement and practical applications. First, existing Transformers often separately emphasize either spatial-wise or spectral-wise correlation, disrupting the 3D features of HSI and hindering the exploitation of unified spatial-spectral correlation. Second, existing self-attention mechanism always establishes full-rank correlation matrix by learning the correlation between pairs of tokens, leading to its inability to describe linear dependence widely existing in HSI among multiple tokens. To address these issues, we propose a novel Exhaustive Correlation Transformer (ECT) for spectral super-resolution. First, we propose a Spectral-wise Discontinuous 3D (SD3D) splitting strategy, which models unified spatial-spectral correlation by integrating spatial-wise continuous splitting strategy and spectral-wise discontinuous splitting strategy. Second, we propose a Dynamic Low-Rank Mapping (DLRM) model, which captures linear dependence among multiple tokens through a dynamically calculated low-rank dependence map. By integrating unified spatial-spectral attention and linear dependence, our ECT can model exhaustive correlation within HSI. The experimental results on both simulated and real data indicate that our method achieves state-of-the-art performance. Codes and pretrained models will be available later.

eess.IV↗
arXiv2023

Computational Spectral Imaging with Unified Encoding Model: A Comparative Study and Beyond

Computational spectral imaging is drawing increasing attention owing to the snapshot advantage, and amplitude, phase, and wavelength encoding systems are three types of representative implementations. Fairly comparing and understanding the performance of these systems is essential, but challenging due to the heterogeneity in encoding design. To overcome this limitation, we propose the unified encoding model (UEM) that covers all physical systems using the three encoding types. Specifically, the UEM comprises physical amplitude, physical phase, and physical wavelength encoding models that can be combined with a digital decoding model in a joint encoder-decoder optimization framework to compare the three systems under a unified experimental setup fairly. Furthermore, we extend the UEMs to ideal versions, namely, ideal amplitude, ideal phase, and ideal wavelength encoding models, which are free from physical constraints, to explore the full potential of the three types of computational spectral imaging systems. Finally, we conduct a holistic comparison of the three types of computational spectral imaging systems and provide valuable insights for designing and exploiting these systems in the future.

eess.IV↗
arXiv2022

Disentangling 3D Attributes from a Single 2D Image: Human Pose, Shape and Garment

For visual manipulation tasks, we aim to represent image content with semantically meaningful features. However, learning implicit representations from images often lacks interpretability, especially when attributes are intertwined. We focus on the challenging task of extracting disentangled 3D attributes only from 2D image data. Specifically, we focus on human appearance and learn implicit pose, shape and garment representations of dressed humans from RGB images. Our method learns an embedding with disentangled latent representations of these three image properties and enables meaningful re-assembling of features and property control through a 2D-to-3D encoder-decoder structure. The 3D model is inferred solely from the feature map in the learned embedding space. To the best of our knowledge, our method is the first to achieve cross-domain disentanglement for this highly under-constrained problem. We qualitatively and quantitatively demonstrate our framework's ability to transfer pose, shape, and garments in 3D reconstruction on virtual data and show how an implicit shape loss can benefit the model's ability to recover fine-grained reconstruction details.

cs.CV↗
arXiv2021

Rotation-constrained optical see-through headset calibration withbare-hand alignment

The inaccessibility of user-perceived reality remains an open issue in pursuing the accurate calibration of optical see-through (OST) head-mounted displays (HMDs). Manual user alignment is usually required to collect a set of virtual-to-real correspondences, so that a default or an offline display calibration can be updated to account for the user's eye position(s). Current alignment-based calibration procedures usually require point-wise alignments between rendered image point(s) and associated physical landmark(s) of a target calibration tool. As each alignment can only provide one or a few correspondences, repeated alignments are required to ensure calibration quality. This work presents an accurate and tool-less online OST calibration method to update an offline-calibrated eye-display model. The user's bare hand is markerlessly tracked by a commercial RGBD camera anchored to the OST headset to generate a user-specific cursor for correspondence collection. The required alignment is object-wise, and can provide thousands of unordered corresponding points in tracked space. The collected correspondences are registered by a proposed rotation-constrained iterative closest point (rcICP) method to optimise the viewpoint-related calibration parameters. We implemented such a method for the Microsoft HoloLens 1. The resiliency of the proposed procedure to noisy data was evaluated through simulated tests and real experiments performed with an eye-replacement camera. According to the simulation test, the rcICP registration is robust against possible user-induced rotational misalignment. With a single alignment, our method achieves 8.81 arcmin (1.37 mm) positional error and 1.76 degree rotational error by camera-based tests in the arm-reach distance, and 10.79 arcmin (7.71 pixels) reprojection error by user tests.

cs.GT↗
arXiv2021

Occlusion-robust Visual Markerless Bone Tracking for Computer-Assisted Orthopaedic Surgery

Conventional computer-assisted orthopaedic navigation systems rely on the tracking of dedicated optical markers for patient poses, which makes the surgical workflow more invasive, tedious, and expensive. Visual tracking has recently been proposed to measure the target anatomy in a markerless and effortless way, but the existing methods fail under real-world occlusion caused by intraoperative interventions. Furthermore, such methods are hardware-specific and not accurate enough for surgical applications. In this paper, we propose a RGB-D sensing-based markerless tracking method that is robust against occlusion. We design a new segmentation network that features dynamic region-of-interest prediction and robust 3D point cloud segmentation. As it is expensive to collect large-scale training data with occlusion instances, we also propose a new method to create synthetic RGB-D images for network training. Experimental results show that our proposed markerless tracking method outperforms recent state-of-the-art approaches by a large margin, especially when an occlusion exists. Furthermore, our method generalises well to new cameras and new target models, including a cadaver, without the need for network retraining. In practice, by using a high-quality commercial RGB-D camera, our proposed visual tracking method achieves an accuracy of 1-2 degress and 2-4 mm on a model knee, which meets the standard for clinical applications.

cs.CV↗
arXiv2020

NNSC-Cobordism of Bartnik Data in High Dimensions

In this short note, we formulate three problems relating to nonnegative scalar curvature (NNSC) fill-ins. Loosely speaking, the first two problems focus on: When are $(n-1)$-dimensional Bartnik data $\big(Σ_i ^{n-1}, γ_i, H_i\big)$, $i=1,2$, NNSC-cobordant? (i.e., there is an $n$-dimensional compact Riemannian manifold $\big(Ω^n, g\big)$ with scalar curvature $R(g)\geq 0$ and the boundary $\partial Ω=Σ_{1} \cup Σ_{2}$ such that $γ_i$ is the metric on $Σ_i ^{n-1}$ induced by $g$, and $H_i$ is the mean curvature of $Σ_i$ in $\big(Ω^n, g\big)$). If $\big(\mathbb{S}^{n-1},γ_{\rm std},0\big)$ is positive scalar curvature (PSC) cobordant to $\big(Σ_1 ^{n-1}, γ_1, H_1\big)$, where $\big(\mathbb{S}^{n-1}, γ_{\rm std}\big)$ denotes the standard round unit sphere then $\big(Σ_1 ^{n-1}, γ_1, H_1\big)$ admits an NNSC fill-in. Just as Gromov's conjecture is connected with positive mass theorem, our problems are connected with Penrose inequality, at least in the case of $n=3$. Our third problem is on $Λ\big(Σ^{n-1}, γ\big)$ defined below.

math.DG↗
arXiv2011

Static flow on complete noncompact manifolds I: short-time existence and asymptotic expansions at conformal infinity

In this paper, we study short-time existence of static flow on complete noncompact asymptotically static manifolds from the point of view that the stationary points of the evolution equations can be interpreted as static solutions of the Einstein vacuum equations with negative cosmological constant. For a static vacuum $(M^n,g,V),$ we also compute the asymptotic expansions of $g$ and $V$ at conformal infinity.

math.DG↗
arXiv2009

Regularity and rigidity of asymptotically hyperbolic manifolds

In this paper, we study some intrinsic characterization of conformally compact manifolds. We show that, if a complete Riemannian manifold admits an essential set and its curvature tends to -1 at infinity in certain rate, then it is conformally compactifiable and the compactified metrics can enjoy some regularity at infinity. As consequences we prove some rigidity theorems for complete manifolds whose curvature tends to the hyperbolic one in a rate greater than 2.

math.DG↗