arXiv subjects

Yue Hu

Publications and source records attributed to Yue Hu.

At least 109 records · Page 6Linked to original sources

arXiv2025

NL2GenSym: Natural Language to Generative Symbolic Rules for SOAR Cognitive Architecture via Large Language Models

SOAR, a classic symbol-based cognitive architecture, has been fostering the development of general, human-like intelligent agents. Nevertheless, its practical adoption is hindered by the laborious manual rule coding. Emerging Large Language Models (LLMs) present the immense potential for efficient rules generation. However, there is a critical gap that current research predominantly focuses on conceptual frameworks and lacks robust experimental validation. To bridge this gap, we propose \textit{N}atural \textit{L}anguage to \textit{Gen}erative \textit{Sym}bolic Rules (NL2GenSym), a novel framework that integrates LLMs with SOAR to autonomously produce generative symbolic rules from natural language. Specifically, our framework introduces a novel Execution-Grounded Generator-Critic mechanism. The LLM-based Generator, guided by a Retrieval-Augmented Generation-accessed self-evolving domain knowledge base, proposes rules from natural language. Subsequently, these rules are immediately executed within the SOAR environment to rigorously validate their correctness. Based on this execution-grounded feedback, a reflective LLM-based Critic drives the iterative refinement of these rules. Experiments on our specialized Water Jug Problem (WJP) dataset, utilizing both Gemini and Qwen series models, validate the efficacy of our framework. It achieves a success rate over 86\% in generating rules from natural language. Crucially, the framework also generates novel heuristic rules, reducing average decision cycles for solving the WJP to 1.98 times the optimal solution and 1/1000 of baseline methods. Additionally, our initial experiments show that NL2GenSym enables smaller-parameter models to achieve better performance than larger counterparts.

cs.CL↗

arXiv2025

UniOTalign: A Global Matching Framework for Protein Alignment via Optimal Transport

Protein sequence alignment is a cornerstone of bioinformatics, traditionally approached using dynamic programming (DP) algorithms that find an optimal sequential path. This paper introduces UniOTalign, a novel framework that recasts alignment from a fundamentally different perspective: global matching via Optimal Transport (OT). Instead of finding a path, UniOTalign computes an optimal flow or transport plan between two proteins, which are represented as distributions of residues in a high-dimensional feature space. We leverage pre-trained Protein Language Models (PLMs) to generate rich, context-aware embeddings for each residue. The core of our method is the Fused Unbalanced Gromov-Wasserstein (FUGW) distance, which finds a correspondence that simultaneously minimizes feature dissimilarity and preserves the internal geometric structure of the sequences. This approach naturally handles sequences of different lengths and is particularly powerful for aligning proteins with nonsequential similarities, such as domain shuffling or circular permutations, which are challenging for traditional DP methods. UniOTalign therefore offers a new, mathematically principled, global matching paradigm for protein alignment, moving beyond the limitations of path-finding algorithms.

q-bio.QM↗

arXiv2025

Cosmic Ray Perpendicular Superdiffusion and Parallel Mirror Diffusion in a Partially Ionized and Turbulent Medium

Understanding cosmic ray (CR) diffusion in a partially ionized medium is both crucial and challenging. In this study, we investigate CR perpendicular superdiffusion and parallel transport in turbulent, partially ionized media using high-resolution 3D two-fluid simulations that treat ions and neutrals separately. We examine the influence of neutral-ion decoupling and the associated damping of turbulence on CR propagation in both transonic and supersonic conditions. Our simulations demonstrate that neutral-ion decoupling significantly damps velocity and magnetic field fluctuations at small scales, producing spectral slopes steeper than those of Kolmogorov and Burgers scaling. In supersonic turbulence, large-scale shock motion is not subject to damping and generates small-scale density enhancements. Moreover, the damping of magnetic field fluctuations substantially decreases pitch-angle scattering, which, however, only slightly affects the CR parallel mean free path $λ_\|$, due to the nonresonant mirror interactions of CRs. In the direction perpendicular to the mean magnetic field, we identify two regimes of the perpendicular superdiffusion of CRs: a diffusive regime ($λ_\| L_{\rm inj}$), with perpendicular separation scaling as $t^{3/2}$. At initially large pitch angles, the effects of magnetic mirroring-naturally arising in magnetohydrodynamic turbulence-become significant, enhancing the confinement of CRs and resulting in $λ_\|<L_{\rm inj}$, despite the damping effect. These results imply that large-pitch-angle CRs can be well confined in the cold ISM, such as molecular clouds.

astro-ph.GA↗

arXiv2025

Reinforced Embodied Planning with Verifiable Reward for Real-World Robotic Manipulation

Enabling robots to execute long-horizon manipulation tasks from free-form language instructions remains a fundamental challenge in embodied AI. While vision-language models (VLMs) have shown promise as high-level planners, their deployment in the real world is hindered by two gaps: (i) the scarcity of large-scale, sequential manipulation data that couples natural language with multi-step action plans, and (ii) the absence of dense, interpretable rewards for fine-tuning VLMs on planning objectives. To address these issues, we propose REVER, a framework that empowers VLMs to generate and validate long-horizon manipulation plans from natural language instructions in real-world scenarios. Under REVER we train and release RoboFarseer, a VLM incentivized to emit chain-of-thought that perform temporal and spatial reasoning, ensuring physically plausible and logically coherent plans. To obtain training data, we leverage the Universal Manipulation Interface framework to capture hardware-agnostic demonstrations of atomic skills. An automated annotation engine converts each demonstration into vision-instruction-plan triplet. We introduce a verifiable reward that scores the generated plan by its ordered bipartite matching overlap with the ground-truth skill sequence. At run time, the fine-tuned VLM functions both as a planner and as a monitor, verifying step-wise completion. RoboFarseer matches or exceeds the performance of proprietary models that are orders of magnitude larger, while on open-ended planning it surpasses the best baseline by more than 40%. In real-world, long-horizon tasks, the complete system boosts overall success by roughly 60% compared with the same low-level controller without the planner. We will open-source both the dataset and the trained model upon publication.

cs.RO↗

arXiv2025

Rate-Distortion Optimized Communication for Collaborative Perception

Collaborative perception emphasizes enhancing environmental understanding by enabling multiple agents to share visual information with limited bandwidth resources. While prior work has explored the empirical trade-off between task performance and communication volume, a significant gap remains in the theoretical foundation. To fill this gap, we draw on information theory and introduce a pragmatic rate-distortion theory for multi-agent collaboration, specifically formulated to analyze performance-communication trade-off in goal-oriented multi-agent systems. This theory concretizes two key conditions for designing optimal communication strategies: supplying pragmatically relevant information and transmitting redundancy-less messages. Guided by these two conditions, we propose RDcomm, a communication-efficient collaborative perception framework that introduces two key innovations: i) task entropy discrete coding, which assigns features with task-relevant codeword-lengths to maximize the efficiency in supplying pragmatic information; ii) mutual-information-driven message selection, which utilizes mutual information neural estimation to approach the optimal redundancy-less condition. Experiments on 3D object detection and BEV segmentation demonstrate that RDcomm achieves state-of-the-art accuracy on DAIR-V2X and OPV2V, while reducing communication volume by up to 108 times. The code will be released.

cs.CV↗

arXiv2025

High-Precision and High-Efficiency Trajectory Tracking for Excavators Based on Closed-Loop Dynamics

The complex nonlinear dynamics of hydraulic excavators, such as time delays and control coupling, pose significant challenges to achieving high-precision trajectory tracking. Traditional control methods often fall short in such applications due to their inability to effectively handle these nonlinearities, while commonly used learning-based methods require extensive interactions with the environment, leading to inefficiency. To address these issues, we introduce EfficientTrack, a trajectory tracking method that integrates model-based learning to manage nonlinear dynamics and leverages closed-loop dynamics to improve learning efficiency, ultimately minimizing tracking errors. We validate our method through comprehensive experiments both in simulation and on a real-world excavator. Comparative experiments in simulation demonstrate that our method outperforms existing learning-based approaches, achieving the highest tracking precision and smoothness with the fewest interactions. Real-world experiments further show that our method remains effective under load conditions and possesses the ability for continual learning, highlighting its practical applicability. For implementation details and source code, please refer to https://github.com/ZiqingZou/EfficientTrack.

cs.RO↗

arXiv2025

Over One Order of Magnitude Enhancement in Hole Mobility of 2D III-V Semiconductors through Valence Band Edge Shift

Two-dimensional (2D) semiconductors show great potential to sustain Moore's law in the era of ultra-scaled electronics. However, their scalable applications are severely constrained by low hole mobility. In this work, we take 2D-GaAs as a prototype of III-V semiconductors to investigate the effects of quantum anharmonicity (QA) on hole transport, employing the stochastic self-consistent harmonic approximation assisted by the machine learning potential. It is found that the room-temperature hole mobility of 2D-GaAs is reduced by $\sim$44% as the QA effects are incorporated, which is attributed to the enhanced electron-phonon scattering from the out-of-plane acoustic polarization. The valence band edge shift (VBES) strategy is proposed to increase the hole mobility by $\sim$1600% at room temperature, which can be realized by 1% biaxial compressive strain. The electron-phonon scattering rate is dramatically decreased due to the full filtering of the original interband electron-phonon scattering channels that existed in the flat hole pocket. The VBES strategy can be further extended to other 2D III-V semiconductors to promote their hole mobilities.

cond-mat.mtrl-sci↗

arXiv2025

Prioritizing Recurrent Services

We study optimal scheduling in multi-class queueing systems with reentrance, where jobs may return for additional service after completion. Such reentrance creates feedback loops that fundamentally alter congestion dynamics and challenge classical scheduling results. We model two distinct dimensions of the reentrance behavior, the probability of return and the speed of return, and show that their product, the effective return rate, is the key statistic that governs optimal priorities. Our main result establishes a dichotomy: when the effective return rate of the smaller job class (the class with lower expected total workload) is lower, a fixed priority rule is optimal; when it is higher, fixed rules are suboptimal and the optimal policy must be state dependent. This characterization clarifies how reentrance changes the externalities that jobs impose on one another and provides structural guidance for designing scheduling policies.

math.OC↗

arXiv2025

pyFAST: A Modular PyTorch Framework for Time Series Modeling with Multi-source and Sparse Data

Modern time series analysis demands frameworks that are flexible, efficient, and extensible. However, many existing Python libraries exhibit limitations in modularity and in their native support for irregular, multi-source, or sparse data. We introduce pyFAST, a research-oriented PyTorch framework that explicitly decouples data processing from model computation, fostering a cleaner separation of concerns and facilitating rapid experimentation. Its data engine is engineered for complex scenarios, supporting multi-source loading, protein sequence handling, efficient sequence- and patch-level padding, dynamic normalization, and mask-based modeling for both imputation and forecasting. pyFAST integrates LLM-inspired architectures for the alignment-free fusion of sparse data sources and offers native sparse metrics, specialized loss functions, and flexible exogenous data fusion. Training utilities include batch-based streaming aggregation for evaluation and device synergy to maximize computational efficiency. A comprehensive suite of classical and deep learning models (Linears, CNNs, RNNs, Transformers, and GNNs) is provided within a modular architecture that encourages extension. Released under the MIT license at GitHub, pyFAST provides a compact yet powerful platform for advancing time series research and applications.

cs.LG↗

arXiv2025

IDU: Incremental Dynamic Update of Existing 3D Virtual Environments with New Imagery Data

For simulation and training purposes, military organizations have made substantial investments in developing high-resolution 3D virtual environments through extensive imaging and 3D scanning. However, the dynamic nature of battlefield conditions-where objects may appear or vanish over time-makes frequent full-scale updates both time-consuming and costly. In response, we introduce the Incremental Dynamic Update (IDU) pipeline, which efficiently updates existing 3D reconstructions, such as 3D Gaussian Splatting (3DGS), with only a small set of newly acquired images. Our approach starts with camera pose estimation to align new images with the existing 3D model, followed by change detection to pinpoint modifications in the scene. A 3D generative AI model is then used to create high-quality 3D assets of the new elements, which are seamlessly integrated into the existing 3D model. The IDU pipeline incorporates human guidance to ensure high accuracy in object identification and placement, with each update focusing on a single new object at a time. Experimental results confirm that our proposed IDU pipeline significantly reduces update time and labor, offering a cost-effective and targeted solution for maintaining up-to-date 3D models in rapidly evolving military scenarios.

cs.CV↗

arXiv2025

Lie-RMSD: A Gradient-Based Framework for Protein Structural Alignment using Lie Algebra

The comparison of protein structures is a fundamental task in computational biology, crucial for understanding protein function, evolution, and for drug design. While analytical methods like the Kabsch algorithm provide an exact, closed-form solution for minimizing the Root Mean Square Deviation (RMSD) between two sets of corresponding atoms, their application is limited to this specific metric. The rise of deep learning and automatic differentiation frameworks offers a new, more flexible paradigm for such optimization problems. We present Lie-RMSD, a novel, fully differentiable framework for protein structural alignment. Our method represents the rigid-body transformation (rotation and translation) as a 6-dimensional vector in the Lie algebra se(3) of the special Euclidean group SE(3). This representation allows the RMSD to be formulated as a loss function that can be directly minimized by modern gradient-based optimizers. We benchmarked our framework by aligning two allosteric conformations of Adenylate Kinase (PDB IDs: 4AKE and 1AKE). We demonstrate that a suite of standard optimizers (SGD, Adam, AdamW, and Sophia) can robustly converge to the global minimum, achieving precision effectively identical to the analytical Kabsch algorithm. This work validates the accuracy of the Lie algebra-based gradient descent approach and establishes a robust foundation for its extension to more sophisticated and biologically relevant scoring functions where no analytical solutions exist.

q-bio.QM↗

arXiv2025

Chemical templates of the Central Molecular Zone. Shock and protostellar object signatures under Galactic Center conditions

(Abridged) The Central Molecular Zone (CMZ) of the Milky Way exhibits extreme conditions, including high gas densities, elevated temperatures, enhanced cosmic-ray ionization rates, and large-scale dynamics. Large-scale molecular surveys reveal increasing chemical and physical complexity in the CMZ. A key step to interpreting the molecular richness found in the CMZ is to build chemical templates tailored to its diverse conditions. The combined impact of high ionization, elevated temperatures, and dense gas remains insufficiently explored for observable tracers. In this study, we utilized UCLCHEM, a gas-grain time-dependent chemical model, to link physical conditions with their corresponding molecular signatures and identify key tracers of temperature, density, ionization, and shock activity. We ran a grid of models of shocks and protostellar objects representative of typical CMZ conditions, focusing on twenty-four species, including complex organic molecules. Shocked and protostellar environments show distinct evolutionary timescales ($\lesssim 10^4$ vs. $\gtrsim 10^4$ years), with 300 K emerging as a key temperature threshold for chemical differentiation. We find that cosmic-ray ionization and temperature are the main drivers of chemical trends. HCO$^+$, H$_2$CO, and CH$_3$SH trace ionization, while HCO, HCO$^+$, CH$_3$SH, CH$_3$NCO, and HCOOCH$_3$ show consistent abundance contrasts between shocks and protostellar regions over similar temperature ranges. While our models underpredict some complex organics in shocks, they reproduce observed trends for most species, supporting scenarios involving recurring shocks in Galactic Center clouds and enhanced ionization towards Sgr B2(N2). Future work should assess the role of shock recurrence and metallicity in shaping chemistry.

astro-ph.GA↗

arXiv2025

Communication-Efficient Multi-Agent 3D Detection via Hybrid Collaboration

Collaborative 3D detection can substantially boost detection performance by allowing agents to exchange complementary information. It inherently results in a fundamental trade-off between detection performance and communication bandwidth. To tackle this bottleneck issue, we propose a novel hybrid collaboration that adaptively integrates two types of communication messages: perceptual outputs, which are compact, and raw observations, which offer richer information. This approach focuses on two key aspects: i) integrating complementary information from two message types and ii) prioritizing the most critical data within each type. By adaptively selecting the most critical set of messages, it ensures optimal perceptual information and adaptability, effectively meeting the demands of diverse communication scenarios.Building on this hybrid collaboration, we present \texttt{HyComm}, a communication-efficient LiDAR-based collaborative 3D detection system. \texttt{HyComm} boasts two main benefits: i) it facilitates adaptable compression rates for messages, addressing various communication requirements, and ii) it uses standardized data formats for messages. This ensures they are independent of specific detection models, fostering adaptability across different agent configurations. To evaluate HyComm, we conduct experiments on both real-world and simulation datasets: DAIR-V2X and OPV2V. HyComm consistently outperforms previous methods and achieves a superior performance-bandwidth trade-off regardless of whether agents use the same or varied detection models. It achieves a lower communication volume of more than 2,006$\times$ and still outperforms Where2comm on DAIR-V2X in terms of AP50. The related code will be released.

cs.CV↗

arXiv2025

Statistics of Gas Density, Velocity, and Magnetic Fields in Cool-Core Galaxy Clusters

Understanding turbulence within the Intracluster Medium (ICM) of galaxy clusters is pivotal for comprehending their evolution and dynamics. Employing 3D magnetohydrodynamic (MHD) simulations of galaxy cluster mergers, we examine the statistical properties of gas density, magnetic fields, and velocity, particularly emphasizing the central regions spanning 400 kpc. The simulations are designed to resemble massive cool-core clusters such as Perseus, while varying the initial plasma $β$ values (100, 200, and 500). Our findings indicate that while the statistical histogram distributions of gas density and velocity appear similar across different $β$ scenarios, their spatial distributions and morphological patterns exhibit noticeable differences. Through the application of the second-order structure function, we identified a scaling relation in velocity fluctuations, characterized by a slope of 1/2 and predominantly dominated by solenoidal components. Furthermore, our analysis reveals a pronounced anisotropy in both velocity and magnetic field fluctuations, with more significant fluctuations along the direction perpendicular to the magnetic fields. This anisotropy is scale-dependent, becoming more pronounced at smaller scales, and exhibits a decreasing trend in scenarios where the magnetic field is relatively weak, particularly at $β=500$. This suggests that the anisotropic nature of these fluctuations is predominantly regulated by the magnetic fields. Additionally, we test the efficacy of the Synchrotron Intensity Gradient (SIG) method for tracing magnetic fields in these environments. The SIG shows a global agreement with the magnetic field across all three $β$ scenarios, confirming the SIG's insensitivity to the medium's magnetization level.

astro-ph.GA↗

arXiv2025

Probing Turbulence, Gravity, Supernovae, and Magnetic Field Effects with the 6D Kinematics of Young Stars in Milky Way Star-Forming Regions

The dynamics of star forming gas can be affected by many physical processes, such as turbulence, gravity, supernova explosions, and magnetic fields. In this paper, we investigate several nearby star forming regions (Orion, Upper Sco, Taurus, and Perseus) for kinematic imprints of these influences on the newly formed stars. Using Gaia DR3 astrometry and APOGEE DR17 radial velocities, we compute first-order velocity structure functions (VSFs) of young stars in galactic Cartesian coordinates in both 6D (3D positions and 3D velocities) and 4D (3D positions and each 1D velocity) to identify signatures of turbulence and anisotropic motion. We also construct 3D and 1D radial velocity profiles to identify coherent expansion trends, and compare stellar proper motions to plane-of-sky magnetic field orientations in Taurus and Perseus. We find that the VSFs are mildly anisotropic, with slightly different amplitudes, slopes, or features in different directions in several groups, but in general, they are all consistent with Larson's Relation at intermediate length scales, especially in less compact groups. In several cases, the VSFs exhibit features suggestive of local energy injection from supernovae. Radial velocity profiles reveal clear anisotropic expansion in multiple groups, with the most extreme cases corresponding to those with the most anisotropic VSFs. In Perseus, we find that the motions of young stars are preferentially perpendicular to the local magnetic field. We find multiple, overlapping causes in each group for the observed kinematics. Our findings support that young stars remember more than just the turbulent state of their natal clouds.

astro-ph.GA↗

arXiv2025

Dimer-Enhanced Optimization: A First-Order Approach to Escaping Saddle Points in Neural Network Training

First-order optimization methods, such as SGD and Adam, are widely used for training large-scale deep neural networks due to their computational efficiency and robust performance. However, relying solely on gradient information, these methods often struggle to navigate complex loss landscapes with flat regions, plateaus, and saddle points. Second-order methods, which use curvature information from the Hessian matrix, can address these challenges but are computationally infeasible for large models. The Dimer method, a first-order technique that constructs two closely spaced points to probe the local geometry of a potential energy surface, efficiently estimates curvature using only gradient information. Inspired by its use in molecular dynamics simulations for locating saddle points, we propose Dimer-Enhanced Optimization (DEO), a novel framework to escape saddle points in neural network training. DEO adapts the Dimer method to explore a broader region of the loss landscape, approximating the Hessian's smallest eigenvector without computing the full matrix. By periodically projecting the gradient onto the subspace orthogonal to the minimum curvature direction, DEO guides the optimizer away from saddle points and flat regions, enhancing training efficiency with non-stepwise updates. Preliminary experiments on a Transformer toy model show DEO achieves competitive performance compared to standard first-order methods, improving navigation of complex loss landscapes. Our work repurposes physics-inspired, first-order curvature estimation to enhance neural network training in high-dimensional spaces.

cs.LG↗

arXiv2025

Magnetic field of a ring-like shape molecular cloud

We present a detailed study of the magnetic field structure in the G111 molecular cloud, a ring-like filamentary cloud within the NGC 7538 region. We utilized interstellar dust polarization from the Planck telescope to trace large-scale field orientations, starlight extinction polarization from the Kanata telescope to probe the cloud's magnetic field after foreground subtraction, and velocity gradients derived from CO isotopologues, observed with the IRAP 30m telescope, to examine dense regions. Our results reveal a coherent yet spatially varying magnetic field within G111. We correct the significant foreground dust contamination through careful subtraction. We observe a global alignment of the magnetic field with density structures suggesting that the field is dynamically important in shaping the cloud. The curved magnetic field along the dense regions, coinciding with mid-infrared emission in WISE data, indicates shock compression, likely driven by stellar feedback or supernova remnants. Our findings support a scenario where G111's morphology results from turbulent shock-driven compression. The interplay between magnetic fields and external forces is crucial in shaping and maintaining the structure of the molecular cloud. Future high-resolution observations will be essential to further constrain the magnetic field's role in cloud evolution.

astro-ph.GA↗

arXiv2025

Characterizing 3D Magnetic Fields and Turbulence in H I Clouds

3D Galactic magnetic fields are critical for understanding the interstellar medium, Galactic foreground polarization, and the propagation of ultra-high-energy cosmic rays. Leveraging recent theoretical insights into anisotropic magnetohydrodynamic (MHD) turbulence, we introduce a deep learning framework to predict the full 3D magnetic field structure-including the plane-of-sky (POS) position angle, line-of-sight (LOS) inclination, magnetic field strength, sonic Mach number ($M_s$), and Alfvén Mach number ($M_A$)-from spectroscopic H~I observations. The deep learning model is trained on synthetic H~I emission data generated from multiphase 3D MHD simulations. We then apply the trained model to observational data from the Commensal Radio Astronomy FAST Survey, presenting maps of 3D magnetic field orientation, magnetic field strength, $M_s$, and $M_A$ for two H~I clouds, a low-velocity cloud (LVC) and an intermediate-velocity cloud (IVC), which overlap in the POS yet reside at different LOS distances. The deep-learning-predicted POS magnetic field position angles align closely with those determined using the velocity gradient technique, whose integrated results are consistent with independent measurements from Planck 353~GHz polarization data. This study demonstrates the potential of deep learning approaches as powerful tools for modeling the 3D distributions of 3D Galactic magnetic fields and turbulence properties throughout the Galaxy.

astro-ph.GA↗