arXiv subjects

Yue Hu

Publications and source records attributed to Yue Hu.

At least 73 records · Page 4Linked to original sources

arXiv2026

ALMA Central molecular zone Exploration Survey (ACES) V: CS(2-1), SO(2_3-1_2), CH3CHO(5_1,4-4_1,3), HC3N(11-10), and H40a lines data

We present data from the ALMA Central Molecular Zone Exploration Survey (ACES) Large Program, which provides broad spectral-line and 3 mm continuum coverage of the Central Molecular Zone (CMZ) at a spatial resolution of 0.1 pc. The survey delivers homogeneous, wide-field mosaics that enable direct comparisons of the physical and chemical conditions across diverse environments in the Galactic center. In this data release paper, we present the CS(2-1), SO(2_3-1_2), CH3CHO(5_1,4-4_1,3), HC3N(11-10), and H40a lines observed simultaneously within two broad spectral windows. These lines reveal pronounced spatial and chemical variations across the CMZ, tracing distinct components of molecular gas, shock-affected regions, and ionized structures. The high angular resolution and multi-line capability of the ACES dataset make it a powerful resource for future studies of gas dynamics, star formation activity, and the physical connection between the CMZ and Sgr A*.

astro-ph.GA↗

arXiv2026

Co-designing a Social Robot for Newcomer Children's Cultural and Language Learning

Newcomer children face barriers in acquiring the host country's language and literacy programs are often constrained by limited staffing, mixed-proficiency cohorts, and short contact time. While Socially Assistive Robots (SARs) show promise in education, their use in these socio-emotionally sensitive settings remains underexplored. This research presents a co-design study with program tutors and coordinators, to explore the design space for a social robot, Maple. We contribute (1) a domain summary outlining four recurring challenges, (2) a discussion on cultural orientation and community belonging with robots, (3) an expert-grounded discussion of the perceived role of an SAR in cultural and language learning, and (4) preliminary design guidelines for integrating an SAR into a classroom. These expert-grounded insights lay the foundation for iterative design and evaluation with newcomer children and their families.

cs.RO↗

arXiv2026

ALMA Central Molecular Zone Exploration Survey (ACES)-IV. Data of the two intermediate-width spectral windows

We release the intermediate-width spectral window data from the ALMA Central Molecular Zone Exploration Survey (ACES) Large Program, which covers SiO(2-1), SO(2_2-1_1), H13CO+(1-0), H13CN(1-0), HN13C(1-0), and HC15N (1-0), among other molecular line transitions, with an angular resolution of ~2 arcsec and a velocity resolution of 1.7 km s-1 . The full cubes of the two spectral windows as well as the key data products will be available to the community. We also present the integrated brightness, peak brightness, centroid velocity, and Galactic longitude-velocity maps of the six lines. We briefly discuss morphological correlations between the continuum and the molecular line emission, and brightness ratios between pairs of isotopologue or isotopomer lines. We highlight features and trends in the data that will be followed up in upcoming ACES science papers.

astro-ph.GA↗

arXiv2026

Photonic timing-engineered solitons for dual-microcomb metrology

Dissipative microcavity solitons offer a route to integrate comb-based metrology systems on photonic chips. However, integrated solitons generally lack agile control of comb parameters, particularly pulse timing control, hindering their application in quantum-limited metrology. Here we introduce dynamical soliton trapping to enable optically timing-engineered microcombs (OTEM). By injecting an auxiliary laser to anchor one of the microcomb lines, we create a potential well to trap and steer the soliton. Thus, soliton timing can be engineered by phase modulating the injected laser. Theory and measurement reveal the fast response bandwidth of the OTEM, which enabled a soliton slew rate of 31.3 ps/$μ$s, surpassing existing time-programmable fiber laser combs by more than two orders of magnitude. Leveraging the timing scan, we used a single OTEM for single-pixel and parallel ranging by retrieving the phase of the multi-heterodyne beat spectrum. Picometer-scale ranging precision was achieved, establishing new record for optical absolute ranging. Our work can transform timing-engineering of microcavity solitons and endow integrated dual-microcomb metrology systems with enhanced precision.

physics.optics↗

arXiv2026

AntibodyDesignBFN: High-Fidelity Fixed-Backbone Antibody Design via Discrete Bayesian Flow Networks

The computational design of antibodies with high specificity and affinity is a cornerstone of modern therapeutic development. While deep generative models have demonstrated potential, they often struggle to balance high-fidelity geometric conditioning with the discrete nature of amino acid sequences. In this work, we present AntibodyDesignBFN, a novel framework for fixed-backbone antibody design based on Discrete Bayesian Flow Networks (BFN). Unlike standard diffusion models, BFNs operate on a continuous probability simplex, enabling a fully differentiable generative process that seamlessly integrates geometric gradients. By combining a lightweight Geometric Transformer with Invariant Point Attention (IPA) and a resource-efficient training strategy, our model establishes a new state-of-the-art. Evaluations on a rigorous 2025 temporal test set (43 complexes) demonstrate that AntibodyDesignBFN achieves an unprecedented Amino Acid Recovery(AAR) of 67.8%, significantly outperforming leading graph-based baselines. Furthermore, the model is highly efficient, enabling millisecond-scale inference on consumer-grade hardware. AntibodyDesignBFN thus offers a powerful, accessible, and mathematically robust framework for next generation antibody engineering. Code and model checkpoints are available at https://github.com/YueHuLab/AntibodyDesignBFN and https://huggingface.co/YueHuLab/AntibodyDesignBFN.

q-bio.QM↗

arXiv2026

ALMA Central Molecular Zone Exploration Survey (ACES) II: 3mm continuum images

The ALMA Central Molecular Zone Exploration Survey, ACES, has mapped $\gtrsim1000$ square arcminutes at 3 mm toward the center of our Galaxy. ACES provides the first large-scale, high-resolution ($\sim2.5$") view of the central $\sim200$ parsecs of the Milky Way. In this work, we describe the continuum data processing and present the continuum data products. In the combined mosaic of 45 individual ALMA mosaics, the typical RMS noise achieved is $\sim0.1$ mJy per $\sim2.5$" beam, though there is a tail of substantially higher noise toward regions with bright continuum structure, especially around Sgr A* and Sgr B2. In-band spectral indices are measurable for a small fraction of the brightest and most compact sources, enabling distinction between dust-dominated and free-free- or synchrotron-dominated sources. To recover emission on large angular scales, we present the GBT MUSTANG-2 Three millimeter Extended Nucleus Survey (TENS), a new 10"resolution survey of the CMZ, which we combine with the ACES image by feathering. To demonstrate the quality and reliability of the ACES data, we compare to previously-published ALMA data obtained with higher resolution and sensitivity, finding overall good agreement with past results, but some disagreement toward the brightest sources.

astro-ph.GA↗

arXiv2026

ALMA Central molecular zone Exploration Survey (ACES) III: Molecular line data reduction and HNCO and HCO$^{+}$ data

The ALMA Central molecular zone Exploration Survey (ACES) large program has observed the inner ~ 200 pc of the Milky Way at 3 mm (Band 3) using ALMA's 12m, 7m, and Total Power arrays. With an angular resolution of ~ 2", ACES provides a contiguous, multi-scale view of the Central Molecular Zone (CMZ) via the dust continuum and a suite of molecular lines. We present an overview of the molecular line data processing for ACES and describe the first data release. We showcase the HNCO (4-3) and HCO$^{+}$ (1-0) data, which were targeted at high spectral resolution (0.2 km s$^{-1}$) to trace the kinematics of the molecular gas in the CMZ. The HNCO and HCO$^{+}$ maps are compared with previous single-dish CMZ surveys and discrete ALMA observations of CMZ clouds to demonstrate the quality of the data. We highlight the ubiquity of parsec-scale, linear absorption features traced by HCO$^{+}$. Their origin is unknown, and ACES provides the first opportunity to study these enigmatic features throughout the CMZ. We release the HNCO and HCO$^{+}$ cubes for all 45 ACES fields, along with the full cube mosaics which combine all fields into a contiguous mosaic of the CMZ. We additionally provide advanced products of these full mosaics, including integrated and peak intensity, noise, and position-velocity maps. These products provide substantial legacy value for the community, offering an unparalleled view of the physical and kinematic structure of the dense gas in the CMZ.

astro-ph.GA↗

arXiv2026

ALMA Central Molecular Zone Exploration Survey (ACES) I: Overview

The mass flows and energy cycles within the inner regions of galaxies exert a powerful influence on the evolution of the galaxy population. The centre of the Milky Way is the only galactic nucleus for which it is possible to resolve the physical mechanisms that drive these cycles, namely star formation and feedback, while also tracing global (>100 pc) processes which determine where and when star formation and feedback occur. We present an overview of ACES, the 'Atacama Large Millimeter/submillimeter Array (ALMA) CMZ Exploration Survey', a ~1.5" angular resolution, 0.2-3 km/s spectral resolution ALMA Band 3 (85-102 GHz), survey of the 'Central Molecular Zone' (CMZ) -- the inner-100 pc of the Galaxy (l = 359.4 deg to 0.8 deg). ACES spectral setup is tuned to observe optimal tracers of the physical, chemical, and kinematic conditions in over 70 spectral features (e.g. HCO+, HNCO, SiO, H40alpha, complex molecules) of the gas in the CMZ, to derive the properties of all potentially star-forming Galactic Centre gas, from global scales (100 pc) to dense ~0.05 pc structures that are expected to host individual star-forming cores, down to sub-sonic (<0.4 km/s) velocity resolution. In this overview paper, we provide the scientific justification for the ACES survey, explain the choice of observational setup, and describe the data legacy products. Finally, we show some of the initial ACES data which highlight the power of ACES' combination of high angular resolution, unprecedented spatial dynamic range, sensitivity, spectral resolution and spectral bandwidth as an illustration of how ACES aims to understand how global processes set the location, intensity, and timescales for star formation and feedback in the CMZ.

astro-ph.GA↗

arXiv2026

FTimeXer: Frequency-aware Time-series Transformer with Exogenous variables for Robust Carbon Footprint Forecasting

Accurate and up-to-date forecasting of the power grid's carbon footprint is crucial for effective product carbon footprint (PCF) accounting and informed decarbonization decisions. However, the carbon intensity of the grid exhibits high non-stationarity, and existing methods often struggle to effectively leverage periodic and oscillatory patterns. Furthermore, these methods tend to perform poorly when confronted with irregular exogenous inputs, such as missing data or misalignment. To tackle these challenges, we propose FTimeXer, a frequency-aware time-series Transformer designed with a robust training scheme that accommodates exogenous factors. FTimeXer features an Fast Fourier Transform (FFT)-driven frequency branch combined with gated time-frequency fusion, allowing it to capture multi-scale periodicity effectively. It also employs stochastic exogenous masking in conjunction with consistency regularization, which helps reduce spurious correlations and enhance stability. Experiments conducted on three real-world datasets show consistent improvements over strong baselines. As a result, these enhancements lead to more reliable forecasts of grid carbon factors, which are essential for effective PCF accounting and informed decision-making regarding decarbonization.

cs.LG↗

arXiv2026

Mode Energy Partition in Partially Ionized Compressible MHD Turbulence

We investigate how neutral-ion collisional damping modifies the spectral properties and energy partition of compressible MHD turbulence using a suite of 3D two-fluid simulations. By systematically varying the neutral-ion coupling strength and decomposing the turbulent velocity field into Alfvén, slow, and fast (polarization) modes, we quantify how each mode responds to the transition from strong to weak coupling. In the strong-coupling regime, the Alfvén and slow modes follow nearly Kolmogorov $k^{-5/3}$ spectra and dominate the kinetic energy budget, while fast modes exhibit a steeper spectrum and contribute $\sim$10\% of the total energy. As the coupling weakens and neutral-ion damping becomes significant, all mode spectra steepen, approaching a dissipation-dominated $k^{-4}$ spectrum, except that the slope mode's spectrum parallel to the mean magnetic field has a power-law slope shallower than -4. While the total kinetic energy is reduced in the weak coupling regime, the slow-mode energy fraction increases substantially toward small scales, whereas the Alfvén-mode fraction decreases correspondingly. In contrast, the fast-mode energy fraction remains largely insensitive to coupling strength. These results demonstrate that partial ionization not only steepens the turbulent spectra but also reshapes the mode energy distribution, enhancing the relative importance of the slow mode while suppressing Alfvén mode in the damping regime. Our findings have important implications for turbulence-driven processes in the partially ionized interstellar medium, including cosmic-ray transport and acceleration.

astro-ph.GA↗

arXiv2026

Deconstructing Positional Information: From Attention Logits to Training Biases

Positional encodings enable Transformers to incorporate sequential information, yet their theoretical understanding remains limited to two properties: distance attenuation and translation invariance. Because natural language lacks purely positional data, the interplay between positional and semantic information is still underexplored. We address this gap by deconstructing the attention-logit computation and providing a structured analysis of positional encodings, categorizing them into additive and multiplicative forms. The differing properties of these forms lead to distinct mechanisms for capturing positional information. To probe this difference, we design a synthetic task that explicitly requires strong integration of positional and semantic cues. As predicted, multiplicative encodings achieve a clear performance advantage on this task. Moreover, our evaluation reveals a hidden training bias: an information aggregation effect in shallow layers that we term the single-head deposit pattern. Through ablation studies and theoretical analysis, we proved that this phenomenon is inherent in multiplicative encodings. These findings deepen the understanding of positional encodings and call for further study of their training dynamics.

cs.LG↗

arXiv2026

SWE-Dev: Evaluating and Training Autonomous Feature-Driven Software Development

Large Language Models (LLMs) have shown strong capability in diverse software engineering tasks. However, feature-driven development, a highly prevalent real-world task that involves developing new functionalities for large, existing codebases, remains underexplored. We therefore introduce SWE-Dev, the first large-scale dataset (with 14,000 training and 500 test samples) designed to evaluate and train autonomous coding systems on real-world end-to-end feature-driven software development tasks. To ensure verifiable and diverse training, SWE-Dev uniquely provides all instances with a runnable environment and its developer-authored executable unit tests. This collection not only provides high-quality data for Supervised Fine-Tuning (SFT), but also enables Reinforcement Learning (RL) by delivering accurate reward signals from executable unit tests. We evaluated SWE-Dev across 17 base LLMs, 10 reasoning-focused LLMs, 10 multi-agent systems, and 8 tool-augmented LLM agents. Results show substantial headroom: the best single-turn model reaches only 22.51\% Pass@1 on the hard split, while OpenHands agents improve to 56.44\% but still leave many tasks unsolved. Code is available here https://github.com/DorothyDUUU/SWE-Dev.

cs.SE↗

arXiv2026

NeVStereo: A NeRF-Driven NVS-Stereo Architecture for High-Fidelity 3D Tasks

In modern dense 3D reconstruction, feed-forward systems (e.g., VGGT, pi3) focus on end-to-end matching and geometry prediction but do not explicitly output the novel view synthesis (NVS). Neural rendering-based approaches offer high-fidelity NVS and detailed geometry from posed images, yet they typically assume fixed camera poses and can be sensitive to pose errors. As a result, it remains non-trivial to obtain a single framework that can offer accurate poses, reliable depth, high-quality rendering, and accurate 3D surfaces from casually captured views. We present NeVStereo, a NeRF-driven NVS-stereo architecture that aims to jointly deliver camera poses, multi-view depth, novel view synthesis, and surface reconstruction from multi-view RGB-only inputs. NeVStereo combines NeRF-based NVS for stereo-friendly renderings, confidence-guided multi-view depth estimation, NeRF-coupled bundle adjustment for pose refinement, and an iterative refinement stage that updates both depth and the radiance field to improve geometric consistency. This design mitigated the common NeRF-based issues such as surface stacking, artifacts, and pose-depth coupling. Across indoor, outdoor, tabletop, and aerial benchmarks, our experiments indicate that NeVStereo achieves consistently strong zero-shot performance, with up to 36% lower depth error, 10.4% improved pose accuracy, 4.5% higher NVS fidelity, and state-of-the-art mesh quality (F1 91.93%, Chamfer 4.35 mm) compared to existing prestigious methods.

cs.CV↗

arXiv2026

Metrology-grade mid-infrared spectroscopy for multi-dimensional perception

The mid-infrared spectral window is essential for molecular fingerprinting and atmospheric sensing, yet unlocking its full potential is currently constrained by a fundamental instrumental trade-off: existing systems cannot simultaneously deliver broad bandwidth, high photon flux, and metrological frequency fidelity. Here, we resolve this bottleneck by demonstrating a metrology-grade spectroscopic system based on difference frequency generation, driven by widely tunable, near-infrared diode lasers traceable to atomic standards. Our system achieves continuous tunability across the 3-3.7 $μ$m atmospheric window and delivers output power exceeding 45 mW with an absolute frequency accuracy of 7.2 MHz. We harness this convergence to overcome a critical barrier in integrated photonics, unambiguously identifying and eliminating hydrogen-induced absorption in silicon nitride microresonators to achieve an 88-fold reduction in optical loss. We further reveal multi-phonon absorption in the silica cladding as the fundamental limit to mid-infrared integrated photonics. Finally, we demonstrate the system's versatility through scattering-resilient LiDAR capable of penetrating optically dense fog, and dual-modality sensing that simultaneously retrieves target distance and chemical composition. By unifying the rigor of frequency metrology with the versatility of broadband sensing, this architecture establishes a new paradigm for multi-dimensional perception in complex environments.

physics.optics↗

arXiv2026

NegoCollab: A Common Representation Negotiation Approach for Heterogeneous Collaborative Perception

Collaborative perception improves task performance by expanding the perception range through information sharing among agents. . Immutable heterogeneity poses a significant challenge in collaborative perception, as participating agents may employ different and fixed perception models. This leads to domain gaps in the intermediate features shared among agents, consequently degrading collaborative performance. Aligning the features of all agents to a common representation can eliminate domain gaps with low training cost. However, in existing methods, the common representation is designated as the representation of a specific agent, making it difficult for agents with significant domain discrepancies from this specific agent to achieve proper alignment. This paper proposes NegoCollab, a heterogeneous collaboration method based on the negotiated common representation. It introduces a negotiator during training to derive the common representation from the local representations of each modality's agent, effectively reducing the inherent domain gap with the various local representations. In NegoCollab, the mutual transformation of features between the local representation space and the common representation space is achieved by a pair of sender and receiver. To better align local representations to the common representation containing multimodal information, we introduce structural alignment loss and pragmatic alignment loss in addition to the distribution alignment loss to supervise the training. This enables the knowledge in the common representation to be fully distilled into the sender.

cs.CV↗

arXiv2026

Beyond Accuracy: A Cognitive Load Framework for Mapping the Capability Boundaries of Tool-use Agents

The ability of Large Language Models (LLMs) to use external tools unlocks powerful real-world interactions, making rigorous evaluation essential. However, current benchmarks primarily report final accuracy, revealing what models can do but obscuring the cognitive bottlenecks that define their true capability boundaries. To move from simple performance scoring to a diagnostic tool, we introduce a framework grounded in Cognitive Load Theory. Our framework deconstructs task complexity into two quantifiable components: Intrinsic Load, the inherent structural complexity of the solution path, formalized with a novel Tool Interaction Graph; and Extraneous Load, the difficulty arising from ambiguous task presentation. To enable controlled experiments, we construct ToolLoad-Bench, the first benchmark with parametrically adjustable cognitive load. Our evaluation reveals distinct performance cliffs as cognitive load increases, allowing us to precisely map each model's capability boundary. We validate that our framework's predictions are highly calibrated with empirical results, establishing a principled methodology for understanding an agent's limits and a practical foundation for building more efficient systems.

cs.CL↗

arXiv2026

PEARL: Plan Exploration and Adaptive Reinforcement Learning for Multihop Tool Use

Large Language Models show great potential with external tools, but face significant challenges in complex, multi-turn tool invocation. They often exhibit weak planning, tool hallucination, erroneous parameter generation, and struggle with robust interaction. To tackle these issues, we present PEARL, a novel framework to enhance LLM planning and execution for sophisticated tool use. PEARL adopts a two-stage approach: an offline phase where the agent explores tools to learn valid usage patterns and failure conditions, and an online reinforcement learning phase. In the online phase, a dedicated Planner is trained via group Relative Policy Optimization (GRPO) with a carefully designed reward function that provides distinct signals for planning quality. Experiments on the ToolHop and T-Eval benchmarks show PEARL significantly outperforms existing methods, achieving a new state-of-the-art success rate of \textbf{56.5\%} on ToolHop while maintaining a low invocation error rate. Our work marks a key advance in addressing the complex planning challenges of tool use, contributing to the development of more robust and reliable LLM-based agents.

cs.CL↗

arXiv2026

Mutagenesis screen to map the functions of parameters of Large Language Models

Large Language Models (LLMs) have significantly advanced artificial intelligence, excelling in numerous tasks. Although the functionality of a model is inherently tied to its parameters, a systematic method for exploring the connections between the parameters and the functionality are lacking. Models sharing similar structure and parameter counts exhibit significant performance disparities across various tasks, prompting investigations into the varying patterns that govern their performance. We adopted a mutagenesis screen approach inspired by the methods used in biological studies, to investigate Llama2-7b and Zephyr. This technique involved mutating elements within the models' matrices to their maximum or minimum values to examine the relationship between model parameters and their functionalities. Our research uncovered multiple levels of fine structures within both models. Many matrices showed a mixture of maximum and minimum mutations following mutagenesis, but others were predominantly sensitive to one type. Notably, mutations that produced phenotypes, especially those with severe outcomes, tended to cluster along axes. Additionally, the location of maximum and minimum mutations often displayed a complementary pattern on matrix in both models, with the Gate matrix showing a unique two-dimensional asymmetry after rearrangement. In Zephyr, certain mutations consistently resulted in poetic or conversational rather than descriptive outputs. These "writer" mutations grouped according to the high-frequency initial word of the output, with a marked tendency to share the row coordinate even when they are in different matrices. Our findings affirm that the mutagenesis screen is an effective tool for deciphering the complexities of large language models and identifying unexpected ways to expand their potential, providing deeper insights into the foundational aspects of AI systems.

cs.AI↗