arXiv subjects

Shu Wang

Publications and source records attributed to Shu Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Understanding the Mechanism of Altruism in Large Language Models

Altruism is fundamental to human societies, fostering cooperation and social cohesion. Recent studies suggest that large language models (LLMs) can display human-like prosocial behavior, but the internal computations that produce such behavior remain poorly understood. We investigate the mechanisms underlying LLM altruism using sparse autoencoders (SAEs). In a standard Dictator Game, minimal-pair prompts that differ only in social stance (generous versus selfish) induce large, economically meaningful shifts in allocations. Leveraging this contrast, we identify a set of SAE features (0.024% of all features across the model's layers) whose activations are strongly associated with the behavioral shift. To interpret these features, we use benchmark tasks motivated by dual-process theories to classify a subset as primarily heuristic (System 1) or primarily deliberative (System 2). Causal interventions validate their functional role: activation patching and continuous steering of this feature direction reliably shift allocation distributions, with System 2 features exerting a more proximal influence on the model's final output than System 1 features. The same steering direction generalizes across multiple social-preference games. Together, these results enhance our understanding of artificial cognition by translating altruistic behaviors into identifiable network states and provide a framework for aligning LLM behavior with human values, thereby informing more transparent and value-aligned deployment.

econ.GN↗

arXiv2026

Interstellar Object 3I/ATLAS Observed from Mars by China's Tianwen-1 Spacecraft

China's Tianwen-1 Mars orbiter successfully imaged the third interstellar object, 3I/ATLAS, during its close encounter with Mars using the onboard HiRIC CMOS camera. This is China's first deep-space observation of an astronomical object. These observations constitute the first imaging of this object from a vantage point significantly out of its orbital plane, providing a unique constraint on dust dynamics. Three observing epochs between 2025 September 30 and October 3 reveal clear changes in coma and tail morphology driven by the rapidly evolving viewing geometry. Comparison with Finson-Probstein dust dynamical models indicates that the coma is dominated by large grains with solar radiation pressure parameter $β\approx 10^{-3} $ - $10^{-2}$, corresponding to grain sizes of a few 100s $μ$m. The extent of the sunward coma implies dust ejection velocities of $3$ - $10$ m s$^{-1}$. Despite the morphological evolution, the azimuthally averaged surface brightness profile remains nearly unchanged through the three epochs, transitioning from a radial slope near -1 close to the nucleus to slightly steeper than -1.5 at larger cometocentric distances, consistent with steady-state dust outflow accelerated by solar radiation pressure. Photometry yields an average $Afρ\sim (2.0\pm0.2)\times10^4$ cm and a corresponding dust mass loss rate of $\dot{M} \sim 10^3$ kg s$^{-1}$.

astro-ph.EP↗

arXiv2026

HiF-VLA: Hindsight, Insight and Foresight through Motion Representation for Vision-Language-Action Models

Vision-Language-Action (VLA) models have recently enabled robotic manipulation by grounding visual and linguistic cues into actions. However, most VLAs assume the Markov property, relying only on the current observation and thus suffering from temporal myopia that degrades long-horizon coherence. In this work, we view motion as a more compact and informative representation of temporal context and world dynamics, capturing inter-state changes while filtering static pixel-level noise. From this perspective, HiF-VLA equips a motion-centric world model for the VLA, enabling agents to reason about temporal dynamics for future evolution during action generation. Building on this idea, we propose HiF-VLA (Hindsight, Insight, and Foresight for VLAs), a unified framework that leverages motion for bidirectional temporal reasoning. HiF-VLA encodes past dynamics through hindsight priors, anticipates future motion via foresight reasoning, and integrates both through a hindsight-modulated joint expert to enable a ''think-while-acting'' paradigm for long-horizon manipulation. As a result, HiF-VLA surpasses strong baselines on LIBERO-Long and CALVIN ABC-D benchmarks, while incurring negligible additional inference latency. Furthermore, HiF-VLA achieves substantial improvements in real-world long-horizon manipulation tasks, demonstrating its broad effectiveness in practical robotic settings.

cs.RO↗

arXiv2026

MemMachine: A Ground-Truth-Preserving Memory System for Personalized AI Agents

Large Language Model (LLM) agents require persistent memory to maintain personalization, factual continuity, and long-horizon reasoning, yet standard context-window and retrieval-augmented generation (RAG) pipelines degrade over multi-session interactions. We present MemMachine, an open-source memory system that integrates short-term, long-term episodic, and profile memory within a ground-truth-preserving architecture that stores entire conversational episodes and reduces lossy LLM-based extraction. MemMachine uses contextualized retrieval that expands nucleus matches with surrounding context, improving recall when relevant evidence spans multiple dialogue turns. Across benchmarks, MemMachine achieves strong accuracy-efficiency tradeoffs: on LoCoMo it reaches 0.9169 using gpt4.1-mini; on LongMemEvalS (ICLR 2025), a six-dimension ablation yields 93.0 percent accuracy, with retrieval-stage optimizations -- retrieval depth tuning (+4.2 percent), context formatting (+2.0 percent), search prompt design (+1.8 percent), and query bias correction (+1.4 percent) -- outperforming ingestion-stage gains such as sentence chunking (+0.8 percent). GPT-5-mini exceeds GPT-5 by 2.6 percent when paired with optimized prompts, making it the most cost-efficient setup. Compared to Mem0, MemMachine uses roughly 80 percent fewer input tokens under matched conditions. A companion Retrieval Agent adaptively routes queries among direct retrieval, parallel decomposition, or iterative chain-of-query strategies, achieving 93.2 percent on HotpotQA-hard and 92.6 percent on WikiMultiHop under randomized-noise conditions. These results show that preserving episodic ground truth while layering adaptive retrieval yields robust, efficient long-term memory for personalized LLM agents.

cs.AI↗

arXiv2026

Adapting SAM to Nuclei Instance Segmentation and Classification via Cooperative Fine-Grained Refinement

Nuclei instance segmentation is critical in computational pathology for cancer diagnosis and prognosis. Recently, the Segment Anything Model has demonstrated exceptional performance in various segmentation tasks, leveraging its rich priors and powerful global context modeling capabilities derived from large-scale pre-training on natural images. However, directly applying SAM to the medical imaging domain faces significant limitations: it lacks sufficient perception of the local structural features that are crucial for nuclei segmentation, and full fine-tuning for downstream tasks requires substantial computational costs. To efficiently transfer SAM's robust prior knowledge to nuclei instance segmentation while supplementing its task-aware local perception, we propose a parameter-efficient fine-tuning framework, named Cooperative Fine-Grained Refinement of SAM, consisting of three core components: 1) a Multi-scale Adaptive Local-aware Adapter, which enables effective capability transfer by augmenting the frozen SAM backbone with minimal parameters and instilling a powerful perception of local structures through dynamically generated, multi-scale convolutional kernels; 2) a Hierarchical Modulated Fusion Module, which dynamically aggregates multi-level encoder features to preserve fine-grained spatial details; and 3) a Boundary-Guided Mask Refinement, which integrates multi-context boundary cues with semantic features through explicit supervision, producing a boundary-focused signal to refine initial mask predictions for sharper delineation. These three components work cooperatively to enhance local perception, preserve spatial details, and refine boundaries, enabling SAM to perform accurate nuclei instance segmentation directly.

cs.CV↗

arXiv2026

The Dependence of the Extinction Coefficient on Reddening for Galactic Cepheids

Cepheids are fundamental distance indicators, playing a crucial role not only in the cosmic distance ladder but also in mapping the structure, kinematics, and extinction properties of the Milky Way. Using high-precision photometry and parallaxes from $Gaia$ Data Release 3, we identify a significant anti-correlation between the $G$-band extinction coefficient and reddening for Galactic Cepheids, quantified as $R_G = 1.918 \pm 0.060 - (0.106 \pm 0.022)\,E(G_{\mathrm{BP}} - G_{\mathrm{RP}})$. We propose that this anti-correlation arises from the combination of the non-linear effects inherent to the broad $Gaia$ bands and the $R_V$ variations caused by diverse interstellar medium. Adopting a fixed $R_G$ would not only lead to an overestimation of the metallicity dependence of Cepheid luminosities, but also systematically underestimate the distances to highly reddened Cepheids. Moreover, the strong reddening dependence of $R_G$ makes Wesenheit function based on it unsuitable for highly reddened Cepheids, since the definition of Wesenheit magnitudes requires a fixed extinction coefficient. In contrast, infrared-based distances, being less affected by non-linear effects and insensitive to $R_V$, provide the most reliable Cepheid distances at present. This work emphasizes the importance of accurately determining $R_V$ for Galactic Cepheids and accounting for non-linear effects in distance measurements, particularly in the optical bands.

astro-ph.GA↗

arXiv2026

EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing

Instruction-based image editing has advanced rapidly, yet reliable and interpretable evaluation remains a bottleneck. Current protocols either (i) depend on paired reference images, resulting in limited coverage and inheriting biases from prior generative models or (ii) rely solely on zero-shot vision language models (VLMs), whose prompt-based assessments of instruction following, content consistency, and visual quality are often imprecise. To address this, we introduce EdiVal, an automated and fine-grained evaluation framework grounded in an object-centric perspective, designed to assess not only standard single-turn but also multi-turn instruction-based editing with precision. Given an input image, EdiVal first decomposes it into semantically meaningful objects, then synthesizes diverse, context-aware editing instructions while dynamically updating object pools across turns. These two stages enable two novel object centric metrics tailored for multi turn evaluation and one global metric of visual quality: 1) EdiVal-IF, which measures instruction following by combining open vocabulary object detectors for symbolic checks with VLMs for semantic verification on detector guided crops; 2) EdiVal-CC, which evaluates content consistency by calculating semantic similarity of unchanged objects and background using the evolving object pools; and 3) EdiVal-VQ, which quantifies changes in overall visual quality with human preference models. Instantiating this pipeline, we build EdiVal Bench, a multi-turn editing benchmark covering 9 instruction types and 16 state-of-the-art editing models, spanning in-context, flow-matching, and diffusion paradigms. We demonstrate that EdiVal can be used to identify existing failure modes, thereby informing the development of the next generation of editing models.

cs.CV↗

arXiv2026

The extinction distances for over a thousand planetary nebulae with Gaia measurements

Although Gaia has identified the central stars of planetary nebulae (CSPNe) for about 70% of known Galactic planetary nebulae (PNe), reliable distance estimates remain incomplete, with fewer than one quarter having accurate parallaxes. Meanwhile, classical extinction-distance samples include only about 70 objects, corresponding to 1.8% of the Galactic PN population. We aim to construct a large and homogeneous catalogue of PN distances by refining extinction-distance measurements with Gaia DR3, providing an independent complement to CSPN parallax-based distances. We develop a Gaia-based extinction-distance method by combining an improved blue-edge approach with an extinction-jump model. PN distances are derived from stellar extinction jumps in line-of-sight extinction-distance profiles and are further constrained by comparisons with published distances, the spatial distribution of stars relative to the PN centre, and the PN radius-distance relation. We obtain distances for 1,066 PNe with a median relative uncertainty of 13%, with about 87% of the sample having uncertainties below 20%. The catalogue includes 765 PNe whose CSPN parallaxes have uncertainties greater than 20% and 128 PNe without CSPN parallaxes. This method complements CSPN parallax-based approaches and extends the traditional extinction-based method to higher Galactic latitudes. For PNe with discrepant literature distances, it helps identify the more reliable estimates and assess CSPN identifications. We find a likely misidentification of the reported CSPN for Fr2-36 and analyse 33 PNe with two CSPN candidates, suggesting improved identifications for 15 objects. This catalogue represents the largest homogeneous set of extinction-based PN distances to date and provides a robust benchmark for studies of Galactic structure, PN populations, and interstellar extinction.

astro-ph.GA↗

arXiv2026

VecFormer: Towards Efficient and Generalizable Graph Transformer with Graph Token Attention

Graph Transformer has demonstrated impressive capabilities in the field of graph representation learning. However, existing approaches face two critical challenges: (1) most models suffer from exponentially increasing computational complexity, making it difficult to scale to large graphs; (2) attention mechanisms based on node-level operations limit the flexibility of the model and result in poor generalization performance in out-of-distribution (OOD) scenarios. To address these issues, we propose \textbf{VecFormer} (the \textbf{Vec}tor Quantized Graph Trans\textbf{former}), an efficient and highly generalizable model for node classification, particularly under OOD settings. VecFormer adopts a two-stage training paradigm. In the first stage, two codebooks are used to reconstruct the node features and the graph structure, aiming to learn the rich semantic \texttt{Graph Codes}. In the second stage, attention mechanisms are performed at the \texttt{Graph Token} level based on the transformed cross codebook, reducing computational complexity while enhancing the model's generalization capability. Extensive experiments on datasets of various sizes demonstrate that VecFormer outperforms the existing Graph Transformer in both performance and speed.

cs.LG↗

arXiv2026

Integrated Exploration and Sequential Manipulation on Scene Graph with LLM-based Situated Replanning

In partially known environments, robots must combine exploration to gather information with task planning for efficient execution. To address this challenge, we propose EPoG, an Exploration-based sequential manipulation Planning framework on Scene Graphs. EPoG integrates a graph-based global planner with a Large Language Model (LLM)-based situated local planner, continuously updating a belief graph using observations and LLM predictions to represent known and unknown objects. Action sequences are generated by computing graph edit operations between the goal and belief graphs, ordered by temporal dependencies and movement costs. This approach seamlessly combines exploration and sequential manipulation planning. In ablation studies across 46 realistic household scenes and 5 long-horizon daily object transportation tasks, EPoG achieved a success rate of 91.3%, reducing travel distance by 36.1% on average. Furthermore, a physical mobile manipulator successfully executed complex tasks in unknown and dynamic environments, demonstrating EPoG's potential for real-world applications.

cs.RO↗

arXiv2026

TS-PEFT: Unveiling Token-Level Redundancy in Parameter-Efficient Fine-Tuning

Current Parameter-Efficient Fine-Tuning (PEFT) methods typically operate under an implicit assumption: Once a target module is selected, every token passing through it contributes equally to the downstream task and requires a parameter update. In this paper, we challenge this convention by revealing a pervasive token-level redundancy in the fine-tuning of large models (LMs). We propose TS-PEFT, a theoretical framework utilizing proximal optimization that acts as a dynamic probe to identify token-level redundancy during the fine-tuning process. Extensive experiments demonstrate that indiscriminately updating all tokens is not only computationally superfluous but often introduces optimization noise. Surprisingly, by discarding 30%-70% of token updates, TS-PEFT consistently matches or exceeds the performance of dense baselines such as LoRA, DoRA. Our in-depth analysis shows that the learned token-level sparsity is a superior indicator of module importance compared to traditional weight criteria, providing a novel data-driven perspective on the intrinsic adaptation mechanism of LMs.

cs.CL↗

arXiv2026

Discovery of an Extremely Luminous Type II Cepheid in the Andromeda Giant Stellar Stream: Evidence for a Hierarchical Triple with an Inner Binary Merger

We report the discovery of LAMOST J0041+3948, the most luminous post-AGB Type II Cepheid (TIIC) known, located in the Andromeda Giant Stellar Stream. Its spectral energy distribution (SED) exhibits a strong near-infrared excess, indicating the presence of a circumbinary dusty disk and hence binarity. SED fitting yields an effective temperature of $T_{\rm eff}=6738_{-262}^{+234}\,$K and a post-AGB luminosity of $\log(L/L_{\odot})=4.32_{-0.08}^{+0.07}$. Comparison with theoretical evolutionary tracks suggests a ~$2.0$-$4.0\,M_{\odot}$ progenitor when accounting for a possible scattered-light contribution. ZTF Light curves reveal a pulsation period of 89d that lies close to the period-luminosity relation for long-period RV Tauri stars. Follow-up spectroscopy reveals clear $s$-process enrichment and signatures consistent with an accretion disk around the companion. The inferred progenitor is significantly younger and more massive than a typical stream member, suggesting that an additional mechanism such as a stellar merger is required. We propose a formation channel in which the present post-AGB binary descends from a hierarchical triple system. In this scenario, the inner binary merged after the system was displaced to its current location by the galaxy merger event, and the resulting massive merger remnant subsequently evolved into the extremely luminous post-AGB star observed today.

astro-ph.SR↗

arXiv2025

Variation of the 2175 Å extinction feature in Andromeda galaxy

Extinction curves contain key information on interstellar dust composition and size distribution, with the 2175 Å bump being the most prominent feature. We analyze 20 sightlines toward M31 using HST/STIS UV spectroscopy combined with multi-band photometry to characterize this feature. The extinction curves show substantial diversity, from MW-like shapes to flatter profiles with $R_V$ reaching up to $\sim5.8$. The strength of the 2175 Å feature varies widely, including two sightlines where the bump is essentially absent. The bump central wavelength spans a broader range than previously reported, while its width remains consistent with earlier studies. A moderate positive correlation is found between bump strength ($c_3$) and width ($γ$). We derive an average UV extinction curve toward M31 with $R_V \approx 3.53$. These results provide new constraints on dust properties and their spatial variations in this galaxy.

astro-ph.GA↗

arXiv2025

BookRAG: A Hierarchical Structure-aware Index-based Approach for Retrieval-Augmented Generation on Complex Documents

As an effective method to boost the performance of Large Language Models (LLMs) on the question answering (QA) task, Retrieval-Augmented Generation (RAG), which queries highly relevant information from external complex documents, has attracted tremendous attention from both industry and academia. Existing RAG approaches often focus on general documents, and they overlook the fact that many real-world documents (such as books, booklets, handbooks, etc.) have a hierarchical structure, which organizes their content from different granularity levels, leading to poor performance for the QA task. To address these limitations, we introduce BookRAG, a novel RAG approach targeted for documents with a hierarchical structure, which exploits logical hierarchies and traces entity relations to query the highly relevant information. Specifically, we build a novel index structure, called BookIndex, by extracting a hierarchical tree from the document, which serves as the role of its table of contents, using a graph to capture the intricate relationships between entities, and mapping entities to tree nodes. Leveraging the BookIndex, we then propose an agent-based query method inspired by the Information Foraging Theory, which dynamically classifies queries and employs a tailored retrieval workflow. Extensive experiments on three widely adopted benchmarks demonstrate that BookRAG achieves state-of-the-art performance, significantly outperforming baselines in both retrieval recall and QA accuracy while maintaining competitive efficiency.

cs.IR↗

arXiv2025

Scaling Relations of the Dusty Torus with Luminosity and the Broad-Line Region

We measure and compare the size of the dusty torus with active galactic nucleus (AGN) luminosity and the size of the broad-line region (BLR), using a sample of 182 AGNs with the best H$β$ lag measurements. After correcting for accretion-disk contamination, torus sizes are determined from the time lags of the Wide-field Infrared Survey Explorer W1 and W2 band light curves relative to the optical band variability based on the interpolated cross-correlation function (ICCF) analysis and the Multiple and Inhomogeneous Component Analysis. We find that the torus size from the W1-band (W2-band) tightly correlates with the 5100~Å continuum luminosity with an intrinsic scatter of 0.15-0.16 dex and the best-fit slope of $0.35 \pm 0.03$ ($0.33 \pm 0.03$), which is clearly shallower than the expected 0.5 slope from the sublimation radius-luminosity relation. We find a moderate negative trend that higher Eddington AGNs tend to have smaller torus sizes than expected from the best-fit, suggesting the Eddington ratio plays a role in flattening the torus size-luminosity relation. By comparing the torus size with the H$β$ reverberation time lag for a subsample of 67 AGNs, we find that the torus size is a factor of $\sim 10$ and $\sim 14$ larger than the BLR size, respectively for W1 and W2 bands. The torus size based on the W1 (W2) band correlates with the BLR size with the best-fit slope of $1.28 \pm 0.16$ ($1.10 \pm 0.15$), which is comparable but slightly steeper than a linear correlation.

astro-ph.GA↗

arXiv2025

PathFix: Automated Program Repair with Expected Path

Automated program repair (APR) techniques are effective in fixing inevitable defects in software, enhancing development efficiency and software robustness. However, due to the difficulty of generating precise specifications, existing APR methods face two main challenges: generating too many plausible patch candidates and overfitting them to partial test cases. To tackle these challenges, we introduce a new APR method named PathFix, which leverages path-sensitive constraints extracted from correct execution paths to generate patches for repairing buggy code. It is based on one observation: if a buggy program is repairable, at least one expected path is supposed to replace the fault path in the patched program. PathFix operates in four main steps. First, it traces fault paths reaching the fault output in the buggy program. Second, it derives expected paths by analyzing the desired correct output on the control flow graph, where an expected path defines how a feasible patch leads to the correct execution. Third, PathFix generates and evaluates patches by solving state constraints along the expected path. Fourth, we validate the correctness of the generated patch. To further enhance repair performance and mitigate scalability issues introduced by path-sensitive analysis, we integrate a large language model (LLM) into our framework. Experimental results show that PathFix outperforms existing solutions, particularly in handling complex program structures such as loops and recursion.

cs.SE↗

arXiv2025

Spectroscopic Reverberation Mapping for SARM: The Case of Mrk 1048 and Mrk 618

Robust extragalactic distance measurements are crucial for resolving the persistent discrepancy in the value of the Hubble constant (H$_0$)). Active Galactic Nuclei (AGNs), through their compact and variable broad-line regions (BLRs), enable the determination of geometric distances when reverberation mapping (RM) is combined with spectroastrometry(SA). We report results from a spectroscopic RM campaign (October 2022 to March 2023) targeting two narrow-line Seyfert 1 galaxies, Mrk 1048 and Mrk 618, using 3.6-m DOT and 2-m HCT. High-cadence spectro-photometric monitoring was carried out using onboard instruments such as ADFOSC, HFOSC, and TANSPEC, resulting in well-sampled continuum and emission line light curves. The observed fractional variability ($F_{\mathrm{var}}$) ranged from 4% to 14% across the $g$-band, H$β$, and H$α$ light curves. The time lags were measured using the interpolated cross-correlation function (ICCF), PyI$^{2}$CCF, and \textsc{JAVELIN} methods. In the rest frame, the ICCF analysis yields H$β$ lags of $10.5^{+2.6}_{-4.2}$ days for Mrk 1048 and $10.2^{+3.4}_{-2.9}$ days for Mrk 618, while the corresponding H$α$ lags are $18.7^{+5.3}_{-5.4}$ and $14.4^{+4.6}_{-10.5}$ days, respectively. The emission-line widths, measured from the rms spectra using $σ_{\mathrm{line}}$, give virial black hole mass estimates of $6.3^{+2.0}_{-2.1} \times 10^7\,M_\odot$ for Mrk 1048 and $1.2^{+0.4}_{-0.6} \times 10^7\,M_\odot$ for Mrk 618. These results will serve as a basis for absolute geometric distance calibration when combined with VLTI/GRAVITY spectro-astrometric measurements, thereby contributing to the development of AGNs as standardizable cosmological probes.

astro-ph.GA↗

arXiv2025

Classification for 969 double-mode RR Lyrae stars from Zwicky Transient Facility

RR Lyrae (RRL) variable stars are cornerstone distance indicators. In particular, double-mode RR Lyrae (RRd) stars enable period--luminosity relations (PLRs) that are less sensitive to metallicity, reducing systematic biases in distance measurements. However, their utility has been limited by a global sample of only $\sim$3,000 objects. We develop an automated RRd-screening pipeline and apply it to a cross-matched sample between the Gaia DR3 RRL catalog and ZTF DR22 time-series photometry. The workflow combines Lomb--Scargle period searches, iterative pre-whitening, period-ratio constraints that suppress $\sim$1-day sampling aliases, and amplitude-based quality cuts, enabling large-scale RRd star screening. We produce two ZTF-based catalogs: (i) 39,322 reliable single-mode RRL (40.5\% of the cross-matched set) and (ii) 969 RRd stars. Among the RRd stars, 614 objects are newly identified, substantially enlarging this previously scarce sample; the catalog achieves an estimated completeness of 47.7\%. The PLR derived from the newly discovered RRd stars agrees with the LMC-based relation, though with larger uncertainties. Incorporating these stars will help tighten the RRd PLR and improve distance measurements. Looking ahead, systematic RRd searches with upcoming surveys such as the Legacy Survey of Space and Time (LSST) and the China Space Station Telescope (CSST) should further extend high-accuracy distances across the Local Group and strengthen their cosmological applications.

astro-ph.SR↗