arXiv subjects

Jie Xu

Publications and source records attributed to Jie Xu.

At least 19 recordsLinked to original sources

arXiv2026

Agentic Wireless Optimization: Workflow Automation, Algorithm Design, and Multi-Timescale Network Operations

Optimization is fundamental to wireless communications, from characterizing channel capacity and designing transmission schemes to allocating network resources. However, conventional workflows require substantial expertise and manual effort, particularly for high-dimensional or nonconvex problems, black-box objectives, and interdependent decisions across time scales. Recent advances in agentic artificial intelligence (AI), particularly agents based on large language models (LLMs), offer new opportunities to address these challenges by integrating reasoning, tool use, and numerical feedback. This article presents an overview of agentic AI for wireless optimization, focusing on workflow automation, algorithm design, and multi-timescale network operations. First, LLM-based agents streamline optimization workflows by automating problem formulation, code generation, numerical solution, and result verification in a closed loop. Second, they facilitate algorithm design by combining and configuring existing numerical methods and refining learning-based optimizers, using numerical feedback to improve solution quality and computational efficiency for nonconvex and black-box problems. Third, they support multi-timescale network operations by decomposing coupled problems into interdependent subtasks and coordinating their solutions to satisfy network-level objectives and service requirements. A multicast beamforming case study demonstrates performance gains from agent-developed solvers, particularly in execution time, and a network energy-saving example illustrates how agents coordinate dynamic base station activation, user association, and beamforming across time scales. Finally, we discuss open challenges and research opportunities in agentic wireless optimization.

eess.SP↗

arXiv2026

ST-CGMDiff: Sparse-Conditioned Cross-Frequency Spatio-Temporal Latent Diffusion for Dynamic 3D Channel Gain Map Reconstruction

Accurate and efficient reconstruction of channel gain maps (CGMs) is essential for environment-aware wireless communications. Unlike prior works that focus on static CGMs at a single frequency band, this paper studies the construction of dynamic three-dimensional (3D) CGMs across multiple frequency bands over consecutive time slots, using limited channel measurements. Specifically, for a given transmitter (Tx) location, we represent the CGM at any given receiver (Rx) height as a two-dimensional (2D) grayscale image, where each pixel indicates the average channel gain over the corresponding Rx spatial grid cell. Accordingly, we propose ST-CGMDiff, a spatio-temporal latent diffusion framework that integrates sparse-observation conditioning and cross-frequency modeling to reconstruct dynamic 3D CGMs across multiple carrier frequencies and varying Rx heights within a unified model. In this framework, we jointly encode the input parameters, including the current sparse observations, observation masks, physical Tx and Rx z-coordinates, and carrier frequencies as spatial conditions. A frequency-conditioned convolutional gated recurrent unit (ConvGRU) then extracts causal information from previously constructed CGMs, while a spatially adaptive fusion mechanism balances historical information and current observations. Furthermore, bidirectional convolutional long short-term memory (BiConvLSTM) modules are introduced to capture the ordered dependencies among the selected Rx z-layers. Finally, we perform few-step latent diffusion followed by high-resolution residual refinement to generate the final CGM with improved spatial fidelity.

eess.SP↗

arXiv2026

Scale Anchoring via Implicit Semantics and Sequence-Consistent Loop Closure for VGGT-Based SLAM

Feed-forward geometric foundation models enable dense monocular reconstruction without requiring camera calibration, but independently reconstructed submaps can accumulate scale drift, and even a valid revisit can yield an inconsistent cross-submap measurement. We extend VGGT-SLAM 2.0 with fixed-root scale anchoring and short-sequence-assisted direct loop connections. SALAD selects an image pair between the initial submap and each eligible new submap; joint VGGT inference then uses learned implicit semantic-geometric cues to supply a scale-only anchor while preserving the original overlap chain. For loop closure, neighboring image pairs support a consistency check and translation-scale alignment. Each accepted candidate contributes one robust Sim(3) factor between existing center-frame nodes, without an auxiliary loop submap. A 17-sequence comparison lowers mean Sim(3) ATE on nine TUM sequences from 0.1018 to 0.0451 m relative to the VGGT-SLAM 2.0 baseline. Anchoring improves all seven no-loop module ablations. On three Clio scenes, a fixed-loop comparison favors the direct connection pipeline over the tested auxiliary-submap pipeline. A separate candidate audit exposes additional rejections of visually overlapping pairs by the short-sequence check. These comparisons distinguish the gains from long-range scale anchoring and loop connection choices from the limitations of candidate verification.

cs.RO↗

arXiv2026

Scalar and Mean Curvature Comparison on Compact Cylinder

Let $ X $ be a closed, oriented Riemannian manifold. Denote by $ (M = X \times I = X \times [0, 1], \partial M = X \times \lbrace 0 \rbrace \cup X \times \lbrace 1 \rbrace, g) $ a compact cylinder with smooth boundary, $ \dim M \geqslant 3 $. In this article, we address the following question: If $ g $ is a Riemannian metric having (i) positive scalar curvature (PSC metric) on $ M $ and nonnegative mean curvature on $ \partial M $; and (ii) $ χ(X) = 0 $ and $ \dim X \geq 5 $, then there exists a PSC metric on $ X $. This generalizes a result of Gromov and Lawson for $ X = \mathbb{T}^{n} $. Our result also implies an infinite band width consequence complementary to Gromov's band width conjecture, which roughly says no additional obstruction or flexibility for the existence of PSC on $ X $ is gained by passing to $ X \times [0, 1] $. Our result relies on geometric PDE analysis with respect to a new conformal invariance---the $ g $-angle condition between normal vector field along $ \partial M $ and the global vector field tangential to the $ I $-direction.

math.DG↗

arXiv2026

A Design-Intent-Preserving Agent Harness for Parametric CAD Modeling

Parametric CAD modeling requires not only matching the target shape, but also preserving design intent under parameter edits. However, most existing methods focus on improving static shape generation accuracy, while the dependencies that govern later edits are often overlooked. To tackle this issue, we present DIP-CAD, a design-intent-preserving agent harness for generating parametric CAD models from images. DIP-CAD combines vision-language model reasoning with CAD kernel operations to establish dependencies that encode the design intent. In parametric CAD, these dependencies are mainly established through geometry references in modeling commands and constraints in parametric sketches. At the command level, ID-based explicit geometry reference binds modeling commands to selected faces, edges, or vertices. At the sketch level, a local-coordinate bridge ensures accurate sketch placement, providing the basis for correctly establishing sketch constraints. Constraints among sketch primitives and constraints to external projected geometry further preserve both internal sketch relationships and dependencies on existing geometry. To provide a more comprehensive data foundation for parametric CAD modeling tasks, we introduce the Geometry Explicit Reference Dataset (GERD), whose structured data format links stepwise modeling commands to intermediate CAD models through explicit geometry references. On GERD-EVL and a DeepCAD test subset, DIP-CAD improves mIoU for static model generation by 11.1% and 5.6%, respectively. In parameter-editing evaluations, the corresponding mIoU improvements are 7.0% and 4.2%. Case studies further demonstrate how parameter edits drive shape updates while preserving design intent.

cs.CV↗

arXiv2026

Physics-informed neural network (PINN) approximations of capillary menisci in polygonal channels

In a zero-gravity cylindrical container, a meniscus can span a square or equilateral-triangular cross-section only at contact angles at or above the respective Concus-Finn critical angles of 45 degrees and 60 degrees. We study strong-form physics-informed neural networks (PINNs) across these thresholds under fixed collocation and optimization budgets. Classical capillary flux balance yields a lower bound on the combined interior and wall residual loss for arbitrary positive sampling densities. In the tested nonexistence regime, several degrees below the critical angles, independently evaluated network losses are 1.6-1.9 times this bound. A polynomial least-squares comparison supports the same geometric interpretation. The PINNs still return smooth graphs with finite axial extent, even though the prescribed equilibrium problem has no spanning graph solution. Their small interior residuals coexist with wall-flux deficits near corners, where adjoining contact-angle conditions demand more flux than a finite-slope graph can supply. Residuals vary on both sides of the critical angle, while the bound tends uniformly to zero at threshold for fixed wall weight. These results give a geometric scale for interpreting the computed loss and locate the incompatibility that a small interior residual can conceal.

physics.flu-dyn↗

arXiv2026

Understanding and Mitigating Under-Confidence in GNNs from the Final Layer

Graph Neural Networks (GNNs) have demonstrated remarkable effectiveness on graph-based tasks. However, their predictive confidence is often miscalibrated, typically exhibiting under-confidence, which harms the reliability of their decisions. Existing calibration methods for GNNs normally introduce additional calibration components, which fail to capture the intrinsic relationship between the model and the prediction confidence, resulting in limited theoretical guarantees and increased computational overhead. To address this issue, we propose a simple yet efficient graph calibration method. We establish a unified theoretical framework revealing that model confidence is jointly governed by class-centroid-level and node-level calibration at the final layer. Based on this insight, we theoretically show that reducing the weight decay of the final-layer parameters alleviates GNN under-confidence by acting on the class-centroid level, while node-level calibration acts as a finer-grained complement to class-centroid-level calibration, which encourages each test node to be closer to its predicted class prototype in the final-layer representations. Extensive experiments validate the superiority of our method. The code is released at https://github.com/huangJC0429/SCAR.

cs.LG↗

arXiv2026

Collaborative In-Context Learning for Indoor Radio Map Estimation

Indoor radio map estimation (RME) is essential for wireless applications such as coverage planning, localization, and network deployment. However, most existing methods follow a supervised learning paradigm that requires costly labeled data and task-specific model training. Moreover, indoor environments involve privacy-sensitive information, such as building layouts and structures, that often cannot be centralized. To address these challenges, we make the first attempt to formulate indoor RME as a collaborative in-context learning (ICL) problem. We propose Poly-Ripple, a framework in which multiple clients perform ICL using large autoregressive vision models (LAVMs) without model updates. Each client is prompted by its private demonstrations and predicts the radio map for the same scene query. To capture the causal structure of wireless propagation outward from the transmitter (Tx), we adopt ripple-ordered autoregressive modeling. We further introduce importance-aware aggregation to exploit heterogeneous contextual knowledge distributed across clients under the context length constraint. We provide theoretical analysis and conduct extensive experiments on a public indoor RME benchmark. The experimental results show that Poly-Ripple consistently outperforms representative ICL baselines, while aligning with our theoretical analysis.

cs.LG↗

arXiv2026

REST: Reliability-Guided Spatial Transcriptomics Supervision for Gene Expression Prediction from Histopathological HE-Stained Images

Histopathology is central to cancer diagnosis, but morphology in Hematoxylin and Eosin (HE)-stained images does not directly reveal the molecular states that characterize disease heterogeneity and inform tumor characterization, prognostic assessment, and biomarker discovery. Spatial Transcriptomics (ST) measures spatially resolved gene expression, providing molecular information within tissue architecture, yet differences in assay coverage and spot informativeness make paired HE--ST observations unequally reliable for representation learning. We propose Reliability-Guided Spatial Transcriptomics (REST) supervision with Biological Program Bank (BPB), a program-structured and reliability-guided ST-supervised training framework for predicting molecular information from H\&E images. REST constructs BPB from curated signature, pathway, and cell-type gene sets to provide a shared structure for ST encoding, reliability estimation, and molecular reconstruction. Guided by BPB, Spot-Reliability Contrastive Alignment (SRCA) measures the reliability of each spot from assay-level program coverage and spot-level activity dispersion, and reweights HE--ST alignment accordingly. Biology-Informed Program Reconstruction (BIPR) further uses learnable program queries to reconstruct member-gene expression from histological morphology. REST's training corpus comprised 209 samples (407,475 spots across 16 organs); evaluation used 70 disjoint samples (230,056 spots across eight organs) for gene-expression, pathway-activity, and cell-signature prediction. REST achieved the highest Macro correlation among state-of-the-art ST-supervised training methods and outperformed existing pathology foundation models. Code is available at https://github.com/lanfz2000/REST.

cs.CV↗

arXiv2026

Three-body corrections to Brownian hard-sphere viscosity

For Brownian hard spheres without hydrodynamic interactions, we obtain a conditional enclosure between 1.8266948 and 1.8267157 for the third-order zero-shear Brownian viscosity coefficient. It applies when the physical three-body correction coincides with the finite-energy weak solution and the remote remainder in the pair adjoint identity vanishes. Kirkwood superposition gives zero because three nonzero pair contributions cancel exactly. An elementary radial identity exposes this cancellation, and a common contact-stress convention allows direct comparison of closure predictions. Although superposition solves the bulk equation, it leaves a shared-contact flux error. Correcting this error defines a unique weak response with positive Dirichlet energy. Explicit symmetric trials yield a variational lower bound and a residual upper bound using the rotational structure of shear and three-sphere geometry. Exact integration and outward interval evaluation enclose the fixed-trial functionals. Under the stated physical conditions, the resulting interval excludes all closure coefficients compared here and supplies a quantitative benchmark for dilute three-body Brownian stress.

cond-mat.soft↗

arXiv2026

Resonant-pair energy return beyond the early Green-Kubo plateau in a quartic chain

Distinguishing rapid local energy exchange from sustained transport matters when thermal conductivity is inferred from finite simulations. In a weakly coupled chain of separable quartic oscillators, we identify a quantitative mechanism that changes an early Green--Kubo plateau: resonant neighbors exchange energy and coherently return it. The pair heat-current integral maps to the classical pendulum trapping response. Quartic homogeneity gives an energy-independent reference period $T_{\rm lib}=7.77ε^{-1/2}$, where $ε$ is the coupling. Without fitted parameters, the leading response predicts a negative first minimum at $0.952T_{\rm lib}$ with depth $-0.74$ times the independent-cell exchange coefficient. Isolated-pair simulations test this reference and resolve finite-coupling corrections. The chain reaches its first minimum on the same time scale, with a shallower minimum and a small later residual. Comparisons with chaotic cells and energy-conserving noise show regimes in which coherent return is absent or interrupted over the measured windows. These finite-time results provide a quantitative reference for separating the early exchange plateau from subsequent correlated dynamics.

cond-mat.stat-mech↗

arXiv2026

Calibrated shape-only recovery of membrane viscosities

Nearby relaxation rates make two membrane viscosities difficult to distinguish from a shape transient. We show how calibrated preparation removes this loss of local sensitivity in a linear Kelvin-Voigt sphere with finite area elasticity and intrinsic membrane dissipation. Two independent releases, with a known signed amplitude ratio and calibrated relative gain and timing, yield normal-displacement records whose fixed linear combination separates the shear and dilatational rates. Their common amplitude remains unknown. For fixed sampling, positive-definite weights, bounded positive rates and nonzero excitation, the amplitude-projected sensitivity has a positive lower bound even at rate coincidence. Without relative calibration, the paired sensitivity decreases linearly with the rate gap; the single tangential-release record decreases quadratically. The calibrated bound is compatible with fixed finite initial energy. A positive local bound also persists with independent ratio calibration of fixed positive precision under a Gaussian error model. Supplied elastic coefficients convert the recovered rates into the two surface viscosities. An unbounded spherical Stokes extension quantifies how bulk dissipation changes the modal patterns and provides a sufficient full-Jacobian criterion for retaining local material sensitivity.

cond-mat.soft↗

arXiv2026

The Segal-Bargmann Transform in Neural Partial Differential Equations

We introduce a novel neural operator method based on Segal-Bargmann transform and Bargmann-Fock space for learning the solution operators of partial differential equations (PDEs). The method brings data encoding, latent-space representation, operator learning, and decoding into an unifed framework. From only finitely many observations, we establish a sup-norm reconstruction error of order $ L^{-\frac{1}{2}} $, where L is the number of complex sampling points that are decoupled with the original samplings and observations. The reconstruction error is independent of the spatial dimension and the number of original sampling points. We show that this new mesh-free Segal-Bargmann Neural Operator approach has superior performance compared to existing NO approaches like FNO, BelNet, GNOT, etc.,\ particularly in terms of pointwise control $ \sup_{x}\lvert\hat{u}_{NO} - u_{truth}\rvert $ for solutions of PDEs $ \mathcal{L}u = f $ which provides pointwise accuracy and interpretability as classical numerical solvers. We also demonstrates the reconstructions of the solution operators for higher dimensional nonlinear PDEs, which suggest that our proposed method has great potential to be applied to more challenging PDEs and to even higher spatial dimensions.

math.NA↗

arXiv2026

DrugHarness: An Evolvable Agent Harness for Computational Drug Discovery

Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with conditional skill induction. Its environment integrates over 80 scientific operations, biomedical resources and task-specific code, preserving associations among candidates, results and artifacts. From execution trajectories and outcome feedback, DrugHarness induces action hypotheses with observable applicability conditions, recommended actions and supporting evidence. An independent review agent evaluates candidate skills derived from these hypotheses through source verification and, when needed, targeted execution trials. Admitted skills guide subsequent tasks and are revised through feedback. Evaluation on MolBench and SMDD-Bench demonstrates capabilities in molecular screening, optimization and constrained design. Without evolution, DrugHarness exceeds reported MolClaw-CC results by 10.8 percentage points in affinity-comparison accuracy and 55% in mean top-three active hits on MolBench. Retrospective evaluation on skill-development tasks shows that evolution increases Qwen's affinity-comparison accuracy from 62.2% to 81.1% and optimization success from 79.5% to 87.2%. In a prospective HPK1 study, DrugHarness achieves a more favorable median docking score than Biomni among system-specific candidates. A synthesized compound achieves a cellular IC50 of 14.19 nM, a 2.64-fold potency improvement over the best input reference, with activity competitive with two selected compounds designed by experienced medicinal chemists. These results support conditional skill evolution for scientific agents and demonstrate a workflow connecting computational drug design with experimental validation.

cs.AI↗

arXiv2026

An AI-Assisted Formalization of the Poincaré Conjecture

We present an AI-assisted Lean 4 formalization of the Poincaré conjecture. The project began with limited reusable formal infrastructure for the geometric analysis behind the proof. To organize this work, we combined a proof blueprint prepared by mathematicians with explicit milestone statements. These milestones enabled parallel agent work and gave mathematicians clear points to locate blockers and provide effective mathematical guidance. Our analysis identifies the human interventions and organizational choices behind this workflow. The project provides a starting point toward reusable infrastructure for future formalization projects; such infrastructure, once developed, could eventually reduce the cost of verifying mathematical results in geometric analysis.

cs.AI↗

arXiv2026

FlashNeRD: Performance-First Contact-Rich Neural Robot Dynamics

Compared with analytical physics, learned dynamics models promise robot simulation that is faster, inherently differentiable, and easily adaptable to real data. Neural Robot Dynamics (NeRD) pursues this by keeping collision detection analytical and replacing a simulator's numerical dynamics for the robot with a learned model. Three limitations remain. NeRD offers little speedup over the simulator it learned from, accepts contact only at predefined points, and has no two-way coupling with objects it manipulates. FlashNeRD removes all three with a parallel streaming architecture that makes each prediction faster and more accurate, an encoder that accepts contacts wherever they occur, and two-way coupling with objects simulated by analytical solvers. Experiments across five robots show faster and more accurate dynamics, faster policy learning, and faster inference-time planning. Across three robots, FlashNeRD's dynamics model is up to $55\times$ faster than an optimized NeRD and more accurate over long rollouts. This speedup extends to policy learning, where PPO trains an ANYmal locomotion policy in 34 seconds, $3.8\times$ faster than the analytical simulator and $2.7\times$ faster than an optimized NeRD. With DIAL-MPC, a sampling-based MPC method, the robot climbs all six test platforms where fixed-contact NeRD manages one, at approximately half the analytical simulator's planning time. Cube-reorientation policies trained with FlashNeRD complete within 2% of the simulator-trained policy's target count.

cs.RO↗

arXiv2026

Beyond the Embedding Bottleneck: Adaptive Retrieval-Augmented 3D CT Report Generation

Automated radiology report generation from 3D CT volumes often suffers from incomplete pathology coverage. We provide empirical evidence that this limitation stems from a representational bottleneck: contrastive 3D CT embeddings encode discriminative pathology signals, yet exhibit severe dimensional concentration, with as few as 2 effective dimensions out of 512. Corroborating this, scaling the language model yields no measurable improvement, suggesting that the bottleneck lies in the visual representation rather than the generator. This bottleneck limits both generation and retrieval; naive static retrieval fails to improve clinical efficacy and can even degrade performance. We propose \textbf{AdaRAG-CT}, an adaptive augmentation framework that compensates for this visual bottleneck by introducing supplementary textual information through controlled retrieval and selectively integrating it during generation. On the CT-RATE benchmark, AdaRAG-CT achieves state-of-the-art clinical efficacy, improving Clinical F1 from 0.420 (CT-Agent) to 0.480 (+6 points); ablation studies confirm that both the retrieval and generation components contribute to the improvement. Code is available at https://github.com/renjie-liang/Adaptive-RAG-for-3DCT-Report-Generation.

cs.CV↗

arXiv2026

When to Compile a Computer-Use Agent? Measuring Payback and Making Compilation Decisions for Token Efficiency

Compiling GUI procedures that agents execute repeatedly into programs can reduce their token costs. However, measuring payback and deciding when to compile have two challenges. First, compilation costs are uncertain because attempts can require repair and still fail to produce a usable program. Second, future reuse is unknown because tasks may stop arriving or GUI drift may stop the program from working. To address these challenges, we propose PACE (Payback-Aware Compilation from Experience), a system with a measurement protocol and an online compilation algorithm. The measurement protocol records successful and failed compilation costs, and compares agent and program execution costs on matched task inputs to estimate per-use savings and payback counts. Using these measurements, the online algorithm compares estimated future savings with compilation costs, including failed attempts, based on past task arrivals and compilation outcomes. It checks execution and compilation charges against a cumulative budget determined by observed task arrivals before allowing either action. Under stated action-cost assumptions, total cost after each arrival is at most $1+ε$ times the cost of running every task with the agent. For successful compilation attempts, estimated payback counts excluding source agent runs are 2-16 uses. In simulations using recorded task arrivals, PACE reduces token costs by 17.3% compared with ReAct, 24.9% with the AutoRPA adaptation, and 17.3% with the ToolPro adaptation on average ($ε=0.25$).

cs.AI↗