arXiv subjects

Chao Liu

Publications and source records attributed to Chao Liu.

At least 73 records · Page 4Linked to original sources

arXiv2026

A Tale of Two Origins: In-Situ versus Accreted Nitrogen-Rich Field Stars in the MW

Spectroscopic surveys have identified significant numbers of metal-poor nitrogen-rich (N-rich) field stars. These stars are strong candidates for escapees from globular clusters (GCs), as their distinctive nitrogen enhancement mirrors the chemical patterns observed in some of the members of GCs. As part of the effort to characterize their chemodynamical properties, we derived abundances for up to 25 elements in a sample of 33 N-rich field giant stars (18 of them are studied for the first time) using high-resolution optical spectroscopy. We confirm their elevated abundances of N, Na, and Al, strongly supporting a GC origin. Given that Galactic GCs themselves formed within diverse progenitor galaxies, we sought to identify the ancestral systems of these N-rich field stars. By analyzing their dynamical parameters, we separated the sample into high-energy (HE) and low-energy (LE) groups. The HE group exhibits lower [α/Fe] and enhanced r-process abundances compared to the LE group. This indicates that the HE stars likely escaped from GCs accreted from massive dwarf galaxies (e.g., Gaia-Sausage-Enceladus), while the LE stars probably originated from in-situ GCs. We also find that the chemical pattern of these N-rich stars with [Fe/H] {\lessapprox} -1.0 are similar to the high-redshift ''N-emitters''. Furthermore, orbital integrations revealed a close encounter between one N-rich field star and the globular cluster NGC 6235. Our work demonstrates the potential of using chemodynamical analyses to trace Galactic assembly through chemical peculiar stars, while highlighting that larger samples and more precise data in the future are crucial to establish definitive origins.

astro-ph.GA↗

arXiv2026

Detection and classification of astronomical sources with Astro-RetinaNet in crowded stellar fields

Upcoming next-generation sky surveys will detect large number of faint objects with magnitudes larger than 25. When objects are crowded within a limited a field of view, blending becomes unavoidable. Blending leads to the omission of many sources during photometry in these fields, which cause an underestimates of tens of percent in crowded fields, and remains a major challenge for existing source-extraction techniques. Although artificial neural networks had shown promising results in the detection and classification in wide-field surveys, they often fail with severely blended stars. We developed a robust deep learning model, Astro-RetinaNet, based on the Retinanet algorithm to detect and classify blended sources in single-band astronomical images. After training and evaluating the performance of our network on simulated images, we find precision of 0.96, 0.89,0.70, 0.50,0.75 for single star, 2-star, 3-star, 4-star and 5-or-more star blending cases, respectively, with star number density $\sim$22000 stars per $\rm arcmin^2$. We compare our method's detection capability and completeness both on CSST simulated NGC 2298 images and HST observed M31 images. In crowded and non-crowded stellar fields of simulated NGC 2298, our results show that the model can recover $82\%$ and $95\%$ sources respectively at magnitude ($i$ band) of 25, while for SExtractor and Photutils the completeness reduces to $20\%, 59\%$ and $60\%, 88\%$ respectively. In the M31 case, as faint as 27 magnitude ($F814W$) in a crowded field, Astro-RetinaNet detects 2,224 sources, significantly outperforming Photutils and SExtractor by factors of 3.4 and 7.1, respectively.

astro-ph.IM↗

arXiv2026

A Vertically Orientated Dark Matter Halo Marks a Flip of the Galactic Disk

Unveiling the 3D shape of the Milky Way's dark-matter halo is critical to understanding its formation history. We created an innovative dynamical model with minimal assumptions on the internal dynamical structures and accommodates a highly flexible triaxial DM halo. By applying the method to 6D phase-space data of K-giant stars from LAMOST + Gaia, we robustly determine the 3D dark-matter distribution of the Milky Way out to approximately $50$ kpc. We discover a triaxial, nearly oblate dark-matter halo with $q_{\rm DM} = Z/X= 0.92\pm0.08$, $p_{\rm DM} = Y/X= 0.8\pm0.2$ averagely within 50 kpc, where $Z$ axis is defined perpendicular to the stellar disk. The axes ratio $q_{\rm DM} > p_{\rm DM}$ is strongly preferred; the long-intermediate axis plane of the dark-matter halo is unexpectedly vertical to the Galactic disk, yet aligned with the `plane of satellites'. This striking configuration suggests that the Galactic disk (and the inner halo) has flipped, likely torqued by minor mergers, from an original alignment with the outer dark-matter halo and satellite plane, as supported by Milky Way analogues from Auriga and TNG50. By allowing $q_{\rm DM}(r)$ and $p_{\rm DM}(r)$ vary with radii, we find tentative evidence that the dark-matter halo is twisted, that it agrees alignment with the disk in the inner regions and transitions to a vertical orientation at $r\gtrsim 20$ kpc, supporting the disk flip scenario prediction. Such disk reorientation is non-trivial yet its physical mechanism is straightforward to comprehend and naturally originates a vertical satellite plane. Our findings offer a unified framework that links dark-matter halo orientation, satellite alignment, and disk evolution, reinforcing the internal consistency of the Milky Way in $Λ$CDM model.

astro-ph.GA↗

arXiv2026

Exploring the formation mechanisms of tidal structures in globular clusters of extragalactic origin

Tidal structures around globular clusters provide valuable insights into cluster evolution and the hierarchical assembly of the Milky Way. Using wide-field imaging data from the DESI Legacy Survey combined with a color-magnitude matched-filter technique, we perform a systematic analysis of extra-tidal features in 28 Galactic globular clusters of likely extragalactic origin, representing the largest homogeneous sample examined in this context to date. The clusters display diverse morphologies: 12 exhibit tidal tails, 9 show diffuse envelopes, and 7 reveal no clear extra-tidal features. Notably, we report the first detection of an extended tidal structure around the Sagittarius-associated cluster Terzan 7. To explore the underlying drivers, we compare intrinsic properties, orbital dynamics, and possible accretion associations across morphological groups. From the parameter distributions, complemented by Kolmogorov-Smirnov tests, we find that total mass, escape velocity, concentration, tidal filling factor, pericentric radius, eccentricity and radial angle in action-angle coordinates are all likely group-sensitive parameters. These results suggest that both internal cluster properties and orbital configurations play important roles in shaping extra-tidal morphologies. In addition, the cluster's accretion history shows no clear correlation with the presence of tidal features, indicating that it is not a direct driver of outer structure formation. Overall, the diversity of tidal structures is unlikely to be governed by a single factor, but instead reflects the interplay between internal dynamical evolution and the external Galactic environment. This study provides the most comprehensive constraints so far on the physical processes driving extra-tidal structures in accreted globular clusters.

astro-ph.GA↗

arXiv2026

Strong Singleton-Like Bounds, Quasi-Perfect Codes and Distance-Optimal Codes in the Sum-Rank Metric

Codes in the sum-rank metric have received many attentions in recent years, since they have wide applications in the multishot network coding, the space-time coding and the distributed storage. In this paper, by constructing covering codes in the sum-rank metric from covering codes in the Hamming metric, we derive new upper bounds on sizes, the covering radii and the block length functions of codes in the sum-rank metric. As applications, we present several strong Singleton-like bounds that are tighter than the classical Singleton-like bound when block lengths are large. In addition, we give the explicit constructions of the distance-optimal sum-rank codes of matrix sizes $s\times s$ and $2\times 2$ with minimum sum-rank distance four respectively by using cyclic codes in the Hamming metric. More importantly, we present an infinite families of quasi-perfect $q$-ary sum-rank codes with matrix sizes $2\times m$. Furthermore, we construct almost MSRD codes with larger block lengths and demonstrate how the Plotkin sum can be used to give more distance-optimal sum-rank codes.

cs.IT↗

arXiv2026

Selecting Optimal Stellar Calibration Fields for the CSST Imaging Survey

The Chinese Space Station Survey Telescope (CSST) will perform a decade-long high-precision wide-field imaging survey that relies on rigorous on-orbit calibration. This necessitates stable celestial benchmark fields to maintain photometric and astrometric consistency throughout the mission lifetime. We establish comprehensive selection criteria including observational visibility, stellar number density, bright-star contamination, and interstellar dust extinction. Using the CSST Observation Strategy Analysis Tool (COSAT) and all-sky dust maps from Planck and SFD, we constrain eligible regions to the ranges of ecliptic latitude $ |β| > 50^\circ$ and galactic latitude $|b| > 15^\circ$. From an initial sample of 29 candidate clusters meeting these spatial constraints, six globular clusters (M13, M92, NGC 104, NGC 362, NGC 1261, and NGC 1851) are identified as optimal calibration fields, fulfilling all the critical criteria. These selected clusters are recommended as optimal calibration field candidates for CSST's on-orbit calibration program, and are fundamental to achieving unprecedented photometric precision in CSST's space-based survey.

astro-ph.SR↗

arXiv2026

Flare: Anomaly Diagnostics for Divergent LLM Training in GPU Clusters of Thousand-Plus Scale

The rapid proliferation of large language models has driven the need for efficient GPU training clusters. However, it is challenging due to the frequent occurrence of training anomalies. Since existing diagnostic tools are narrowly tailored to specific issues, there are gaps in their ability to address anomalies spanning the entire training stack. In response, we introduce Flare, a diagnostic framework designed for distributed LLM training at scale. Flare first integrates a lightweight tracing daemon for full-stack and backend-extensible tracing. Additionally, it features a diagnostic engine that automatically diagnoses anomalies, with a focus on performance regressions. The deployment of Flare across 6,000 GPUs has demonstrated significant improvements in pinpointing deficiencies in real-world scenarios, with continuous operation for over eight months.

cs.OS↗

arXiv2026

Spectral Disentanglement and Enhancement: A Dual-domain Contrastive Framework for Representation Learning

Large-scale multimodal contrastive learning has recently achieved impressive success in learning rich and transferable representations, yet it remains fundamentally limited by the uniform treatment of feature dimensions and the neglect of the intrinsic spectral structure of the learned features. Empirical evidence indicates that high-dimensional embeddings tend to collapse into narrow cones, concentrating task-relevant semantics in a small subspace, while the majority of dimensions remain occupied by noise and spurious correlations. Such spectral imbalance and entanglement undermine model generalization. We propose Spectral Disentanglement and Enhancement (SDE), a novel framework that bridges the gap between the geometry of the embedded spaces and their spectral properties. Our approach leverages singular value decomposition to adaptively partition feature dimensions into strong signals that capture task-critical semantics, weak signals that reflect ancillary correlations, and noise representing irrelevant perturbations. A curriculum-based spectral enhancement strategy is then applied, selectively amplifying informative components with theoretical guarantees on training stability. Building upon the enhanced features, we further introduce a dual-domain contrastive loss that jointly optimizes alignment in both the feature and spectral spaces, effectively integrating spectral regularization into the training process and encouraging richer, more robust representations. Extensive experiments on large-scale multimodal benchmarks demonstrate that SDE consistently improves representation robustness and generalization, outperforming state-of-the-art methods. SDE integrates seamlessly with existing contrastive pipelines, offering an effective solution for multimodal representation learning.

cs.LG↗

arXiv2026

Variations in the Milky Way's Stellar Mass Function at [Fe/H] < -1

We present the first determination of the Galactic stellar mass function (MF) for low-mass stars (0.2-0.5 M_sun) at metallicities [Fe/H] < -1. A sample of ~53,000 stars was selected as metal-poor on the basis of both their halo-like orbits and their spectroscopic [Fe/H] from Gaia DR3 BP/RP (XP) spectra. These metallicity estimates for low-mass stars were enabled by calibrating Gaia XP spectra with stellar parameters from SDSS-V. For -1.5 < [Fe/H] < -1, we find that the MF below 0.5 M_sun exhibits a "bottom-heavy" power-law slope of alpha ~ -1.6. We tentatively find that at even lower metallicities, the MF becomes very bottom-light, with a near-flat power-law slope of alpha ~ 0 that implies a severe deficit of low-mass stars. This metallicity-dependent variation is insensitive to the adopted stellar evolution model. These results show that the Galactic low-mass MF is not universal, with variations in the metal-poor regime. A further calibration of XP metallicities in the regime of M < 0.5 M_sun and [Fe/H] < -1.5 will be essential to verify these tentative low-metallicity trends.

astro-ph.GA↗

arXiv2026

On the Adversarial Robustness of 3D Large Vision-Language Models

3D Vision-Language Models (VLMs), such as PointLLM and GPT4Point, have shown strong reasoning and generalization abilities in 3D understanding tasks. However, their adversarial robustness remains largely unexplored. Prior work in 2D VLMs has shown that the integration of visual inputs significantly increases vulnerability to adversarial attacks, making these models easier to manipulate into generating toxic or misleading outputs. In this paper, we investigate whether incorporating 3D vision similarly compromises the robustness of 3D VLMs. To this end, we present the first systematic study of adversarial robustness in point-based 3D VLMs. We propose two complementary attack strategies: \textit{Vision Attack}, which perturbs the visual token features produced by the 3D encoder and projector to assess the robustness of vision-language alignment; and \textit{Caption Attack}, which directly manipulates output token sequences to evaluate end-to-end system robustness. Each attack includes both untargeted and targeted variants to measure general vulnerability and susceptibility to controlled manipulation. Our experiments reveal that 3D VLMs exhibit significant adversarial vulnerabilities under untargeted attacks, while demonstrating greater resilience against targeted attacks aimed at forcing specific harmful outputs, compared to their 2D counterparts. These findings highlight the importance of improving the adversarial robustness of 3D VLMs, especially as they are deployed in safety-critical applications.

cs.CV↗

arXiv2026

Decision-Aware Semantic State Synchronization in Compute-First Networking

In Compute-First Networking (CFN), an Access Point (AP) makes task offloading decisions based on resource state information reported by a Service Node (SN). A fundamental challenge arises from the trade-off between update overhead and decision accuracy: Frequent state updates consume limited network resources, while infrequent updates lead to stale state views and degraded task performance, especially under high system load. Existing approaches based on periodic updates or Age of Information (AoI) mainly focus on temporal freshness and often overlook whether a state change is actually relevant to offloading decisions. This paper proposes SenseCFN, a decision-aware state synchronization framework for CFN. Instead of synchronizing raw resource states, SenseCFN focuses on identifying state changes that are likely to alter offloading decisions. To this end, we introduce a lightweight semantic state representation that captures decision-relevant system characteristics, along with a Semantic Deviation Index (SDI) to quantify the impact of state shifts on decision outcomes. Based on SDI, the SN triggers updates only when significant decision-impacting changes are detected. Meanwhile, the AP performs offloading decisions using cached semantic states with explicit awareness of potential staleness. The update and offloading policies are jointly optimized using a centralized training with distributed execution (CTDE) approach. Simulation results show that SenseCFN maintains a task success rate of up to 99.6% in saturation-prone scenarios, outperforming baseline methods by more than 25%, while reducing status update frequency by approximately 70% to 96%. These results indicate that decision-aware state synchronization provides an effective and practical alternative to purely time-based update strategies in CFN.

cs.NI↗

arXiv2026

Evidence-Grounded Multi-Agent Planning Support for Urban Carbon Governance via RAG

Urban carbon governance requires planners to integrate heterogeneous evidence -- emission inventories, statistical yearbooks, policy texts, technical measures, and academic findings -- into actionable, cross-departmental plans. Large Language Models (LLMs) can assist planning workflows, yet their factual reliability and evidential traceability remain critical barriers in professional use. This paper presents an evidence-grounded multi-agent planning support system for urban carbon governance built upon standard text-based Retrieval-Augmented Generation (RAG) (without GraphRAG). We align the system with the typical planning workflow by decomposing tasks into four specialized agents: (i) evidence Q\&A for fact checking and compliance queries, (ii) emission status assessment for diagnostic analysis, (iii) planning recommendation for generating multi-sector governance pathways, and (iv) report integration for producing planning-style deliverables. We evaluate the system in two task families: factual retrieval and comprehensive planning generation. On factual retrieval tasks, introducing RAG increases the average score from below 6 to above 90, and dramatically improves key-field extraction (e.g., region and numeric values near 100\% detection). A real-city case study (Ningbo, China) demonstrates end-to-end report generation with strong relevance, coverage, and coherence in expert review, while also highlighting boundary inconsistencies across data sources as a practical limitation.

cs.CY↗

arXiv2025

AKG kernel Agent: A Multi-Agent Framework for Cross-Platform Kernel Synthesis

Modern AI models demand high-performance computation kernels. The growing complexity of LLMs, multimodal architectures, and recommendation systems, combined with techniques like sparsity and quantization, creates significant computational challenges. Moreover, frequent hardware updates and diverse chip architectures further complicate this landscape, requiring tailored kernel implementations for each platform. However, manual optimization cannot keep pace with these demands, creating a critical bottleneck in AI system development. Recent advances in LLM code generation capabilities have opened new possibilities for automating kernel development. In this work, we propose AKG kernel agent (AI-driven Kernel Generator), a multi-agent system that automates kernel generation, migration, and performance tuning. AKG kernel agent is designed to support multiple domain-specific languages (DSLs), including Triton, TileLang, CPP, and CUDA-C, enabling it to target different hardware backends while maintaining correctness and portability. The system's modular design allows rapid integration of new DSLs and hardware targets. When evaluated on KernelBench using Triton DSL across GPU and NPU backends, AKG kernel agent achieves an average speedup of 1.46$\times$ over PyTorch Eager baselines implementations, demonstrating its effectiveness in accelerating kernel development for modern AI workloads.

cs.AI↗

arXiv2025

Planet Across Space and Time (PAST). VII. The origin and tidal evolution of hot Jupiters constrained by a broken age-frequency relation

The discovery of hot Jupiters has challenged the classical planet formation theory. Although various formation mechanisms have been proposed, the dominant channel and relative contributions remain unclear. Furthermore, hot Jupiters offer a unique opportunity to test tidal theory and measure the fundamental tidal quality factor, which is yet to be well-constrained. In this work, based on a hot Jupiter sample around single Sun-like stars with kinematic properties, {we find that the declining trend of their frequency is broken with a ridge at about 2 Gyr, providing direct evidence that hot Jupiters are formed with multiple origins of different timescales. By fitting with the theoretical expectations, we provide a constraint of tidal factor for Sun-like stars, which aligns well with the detected number of hot Jupiters with orbital decay. Moreover, we simultaneously constrain the relative importance of different channels: although the majority of hot Jupiters are formed early, within several tenths of Gyr via 'Early' models (e.g., in-situ formation, disk migration, planet-planet scattering and Kozai-Lidov interaction), a significant portion (about 40%) should be formed late on a relatively long timescale extending up to several Gyr mainly via the secular chaos mechanism, further supported by the obliquity distribution of 'late-arrived' hot Jupiters. Our findings provide a unified framework that reconciles hot Jupiter demographics and long-term evolution with multichannel formation.

astro-ph.EP↗

arXiv2025

AndesVL Technical Report: An Efficient Mobile-side Multimodal Large Language Model

In recent years, while cloud-based MLLMs such as QwenVL, InternVL, GPT-4o, Gemini, and Claude Sonnet have demonstrated outstanding performance with enormous model sizes reaching hundreds of billions of parameters, they significantly surpass the limitations in memory, power consumption, and computing capacity of edge devices such as mobile phones. This paper introduces AndesVL, a suite of mobile-side MLLMs with 0.6B to 4B parameters based on Qwen3's LLM and various visual encoders. We comprehensively outline the model architectures, training pipeline, and training data of AndesVL, which achieves first-tier performance across a wide range of open-source benchmarks, including fields such as text-rich image understanding, reasoning and math, multi-image comprehension, general VQA, hallucination mitigation, multilingual understanding, and GUI-related tasks when compared with state-of-the-art models of a similar scale. Furthermore, we introduce a 1+N LoRA architecture alongside a Quantization-Aware LoRA Fine-Tuning (QALFT) framework to facilitate efficient task adaptation and model compression during mobile-side deployment of AndesVL. Moreover, utilizing our cache eviction algorithm -- OKV -- along with customized speculative decoding and compression strategies, we achieve a 6.7x peak decoding speedup ratio, up to 30.9% memory reduction, and 1.8 bits-per-weight when deploying AndesVL-4B on MediaTek Dimensity 9500 chips. We release all models on https://huggingface.co/OPPOer.

cs.CV↗

arXiv2025

Existence and bounds of nonlinear singularity-free cosmological solutions in a string-inspired gravity

We provide a rigorous proof for the existence of homogeneous, isotropic and globally singularity-free cosmological solutions in Einstein-dilaton-Gauss-Bonnet (EdGB) gravity with exponential coupling. While numerical studies suggested such solutions exist, a formal proof remained elusive. By employing a novel ``power identity method'' and overcoming significant challenges posed by the strong nonlinearities of the exponential coupling, which are not present in the quadratic coupling analyzed in our companion paper \cite{he2025proofssingularityfreesolutionsscalarization}, we establish a FLRW solution valid for all time $t\in(-\infty,+\infty)$, where the Hubble parameter remains positive and vanishes asymptotically, while the scalar field evolves monotonically. This result align with numerical simulations and offer a firm mathematical foundation for singularity-free cosmology in a string-inspired setting.

math.AP↗

arXiv2025

FANoise: Singular Value-Adaptive Noise Modulation for Robust Multimodal Representation Learning

Representation learning is fundamental to modern machine learning, powering applications such as text retrieval and multimodal understanding. However, learning robust and generalizable representations remains challenging. While prior work has demonstrated that active noise injection, a form of data augmentation, can enhance encoding performance, most existing methods rely on heuristic or static noise, overlooking the dynamic nature of feature distributions during training. In this work, we systematically study the role of noise in representation learning from both gradient-based and feature distribution perspectives, using InfoNCE loss as a representative example. Focusing on multimodal representation learning, we propose FANoise, a novel feature-adaptive noise injection strategy. By leveraging the dynamics of contrastive learning, FANoise effectively mitigates the negative impacts of noise while preserving its benefits. Under this theoretically grounded framework, comprehensive experiments demonstrate that FANoise consistently improves overall performance on multimodal tasks across various base VLM models.

cs.LG↗

arXiv2025

Stellar Parameters of BOSS M dwarfs in SDSS-V DR19

We utilized the Stellar LAbel Machine (SLAM), a data-driven model based on Support Vector Regression, to derive stellar parameters ([Fe/H], $T_{\rm eff}$, and $\log{g}$) for SDSS-V M dwarfs using low-resolution optical spectra (R$\sim$2000) obtained with the BOSS spectrographs. These parameters are calibrated using LAMOST F, G or K dwarf companions ([Fe/H]), and APOGEE Net ($T_{\rm eff}$ and $\log{g}$), respectively. Comparisons of SLAM predicted [Fe/H] values between two components of M+M dwarfs wide binaries show no bias but with a scatter of 0.11 dex. Further comparisons with two other works, which also calibrated the [Fe/H] of M dwarfs by using the F/G/K companions, reveal biases of -0.06$\pm$0.16 dex and 0.02$\pm$0.14 dex, respectively. The SLAM-derived effective temperatures agree well with the temperature which is calibrated by using interferometric angular diameters (bias: -27$\pm$92 K) and those of the LAMOST (bias: -34$\pm$65 K), but are systematically lower than those from an empirical relationship between the color index and $T_{\rm eff}$ by 146$\pm$45 K. The SLAM surface gravity aligns well with those of LAMOST (bias: -0.01$\pm$0.07 dex) and those derived from the stellar mass and radius (bias: -0.04$\pm$0.09 dex). Finally, we investigated a bias in [Fe/H] between SLAM and APOGEE ASPCAP. It depends on ASPCAP's [Fe/H] and $T_{\rm eff}$, we provide an equation to correct the ASPCAP metallicities.

astro-ph.SR↗