arXiv subjects

Shu Wang

Publications and source records attributed to Shu Wang.

At least 19 recordsLinked to original sources

arXiv2026

A JWST/NIRSpec Catalog of Low-Mass Candidates toward NGC 2024 from Multi-component Spectral Modeling

Young low-mass objects in crowded, highly obscured star-forming regions are difficult to characterize. Deriving reliable physical parameters requires separating photospheric emission from complex extinction and circumstellar emission. We apply a multi-component spectral-modeling framework to JWST/NIRSpec PRISM spectra toward NGC 2024, obtaining adopted fits for 78 sources. The framework incorporates atmospheric models, dust and ice opacities, and optional thermal emission. Most spectra favor additional large-H$_2$O-ice opacity over the baseline small-grain model, while a smaller subset requires additional gray opacity, potentially arising from large refractory grains. Some spectra also require a thermal component to reproduce the long-wavelength continuum. The parameters inferred with this framework, together with the NGC 2024 distance, yield radii and bolometric luminosities. Comparison with evolutionary models for young low-mass and substellar objects provides mass estimates and identifies 46 candidate members, 27 sources classified as background, and five uncertain sources. Candidate members show lower fitted surface gravities than likely background stars. Of the candidate members, 23 lie above the youngest available isochrones, plausibly reflecting the cluster's youth and uncertainties in model initial conditions. The resulting catalog provides a uniform characterization of the NGC 2024 sample and establishes a framework for treating complex extinction and circumstellar emission in JWST spectra of other obscured star-forming regions.

astro-ph.SR↗

arXiv2026

Age-Resolved Vertical Motions in the Milky Way Warp: A Localized Enhancement in Intermediate-Age Populations

Determining how the Milky Way warp evolves provides constraints on the formation and dynamical evolution of galactic disks. Using Apache Point Observatory Galactic Evolution Experiment Data Release 19 (APOGEE DR19) spectroscopy, we estimate ages for 222,355 red giants. With Gaia astrometry and geometric distances, we analyze 168,824 of these stars in five age bins spanning $0$--$10$ Gyr. The warp-related radial vertical-velocity ($V_Z$) profile of young red giants ($0$--$2$ Gyr) is broadly consistent with those of classical Cepheids and young open clusters. Over $9\leq R<12$ kpc, we find that the mean $V_Z$ values of the intermediate-age red giants ($2$--$4$ and $4$--$6$ Gyr) are $1.48\pm0.46$ and $1.49\pm0.52$ km s$^{-1}$ higher than that of the young population, respectively. By contrast, at $R\simeq12$--$13$ kpc, the $V_Z$ profiles of the $0$--$6$ Gyr red giants are consistent within their uncertainties. The intermediate-age enhancement persists across our robustness tests and may reflect smaller effective kinematic warp onset radii for the intermediate-age populations. The observed velocity differences may arise from the combined effects of earlier perturbations and the present-day warp, or from differences in vertical structure among stellar populations.

astro-ph.GA↗

arXiv2026

Efficient estimation of structural VARs with higher-order moments

This paper provides a practical standard for efficient implementation of GMM estimation of structural vector autoregressions identified via higher-order moments. We establish conditions under which the structural estimator is adaptive to reduced-form estimation and the conventional weighting is asymptotically optimal. Despite this optimality, its feasible implementation can be unstable because the covariance of the moments involves many higher-order co-moments. We propose exploiting the same identifying assumption for a structured, bias-corrected covariance estimation under both mutual independence and the weaker mean independence, which also allows for common volatility. The resulting estimator is both asymptotically and computationally efficient. Monte Carlo experiments show substantial gains over conventional GMM and performance comparable to likelihood-based estimators under independent shocks. Under common volatility, the mean-independence estimator remains close to the truth while competing estimators are biased. An oil-market application illustrates the practical gains of the approach.

econ.EM↗

arXiv2026

Testing strong exogeneity in proxy SVARs with internal synthetic proxies

Proxy SVARs often rely on a single credible proxy, leaving proxy exogeneity untestable. Narrative arguments used to justify proxy validity imply a stronger form of exogeneity. We formalize the strong exogeneity assumption and propose a corresponding test based on internal synthetic proxies. The approach uses the external proxy to locate the target-shock direction in the reduced-form innovations and synthesize the candidate shock into an additional internal synthetic proxies, which allows to test the joint exogeneity of the original and internal synthetic proxy. We derive the large-sample theory for the ISP test and show that the identification strength is governed by a direct measure of the target shock's non-Gaussianity. We study weak identification under asymptotic Gaussianity and develop an identification-robust version of the test. Monte Carlo experiments demonstrate the favorable power of the test when the proxy has a low signal-to-noise ratio. Applications to monetary policy and oil supply news shocks find no evidence against the exogeneity of the benchmark proxies.

econ.EM↗

arXiv2026

Understanding the Mechanism of Altruism in Large Language Models

Altruism is fundamental to human societies, fostering cooperation and social cohesion. While recent studies suggest that large language models (LLMs) can display human-like prosocial behavior, the underlying mechanism that produces such behavior remains unclear. We leverage sparse autoencoders (SAEs) to identify a set of features that contribute to an LLM's altruistic choices in the Dictator Game. We show that intervening on these features substantially shifts the model's allocations and produces similar effects in several other social decision scenarios. Together, these results link the model's prosocial output to internal features that can be identified and causally manipulated, providing a transparent approach to studying the computational basis of decision-making in LLMs.

econ.GN↗

arXiv2026

BenchChecker: Assessing the Credibility of Bug-Fixing Benchmarks for LLMs

While Large Language Models (LLMs) show promise for Automatic Program Repair (APR), existing benchmarks suffer from data contamination, in which test samples appear in training sets, artificially inflating performance. To address this, we propose BenchChecker, a framework compatible with both open-source and closed-source models that supports more credible evaluation by detecting such contamination. Grounded in the observation that persistent patches likely remain in recent repository snapshots used for training, BenchChecker employs a forensics-based two-stage detection process using only textual model outputs together with public repository history. The first stage conducts a repository presence test by probing the model with code prefixes to infer training inclusion; the second stage performs a patch presence test to verify patch persistence via similarity alignment and test-suite verification. Evaluations using StarCoder-15B, StarCoder2-15B, and CodeGen2-16B trained on The Stack v2 demonstrate that BenchChecker achieves robust detection with AUC scores around 0.9 across six benchmarks. Furthermore, applying BenchChecker to SWE-bench Verified and Multi-SWE-bench reveals that filtering contaminated samples reduces the reported resolution rates of most LLMs by more than 20% relative to the original rates on medium-difficulty tasks.

cs.SE↗

arXiv2026

AgentFly: Scaling Agentic Reinforcement Learning with Unified Resource System

Methods to build LLM agents have evolved from prompt engineering and supervised finetuning to agentic reinforcement learning (agentic RL). However, agentic RL remains bottlenecked by its surrounding systems: agents must interact with heterogeneous environments, such as sandboxes, model services, and external APIs. Their allocation, reuse, and lifecycle dominate rollout cost and cap the scale at which training becomes practical. In this work, we present AgentFly, an agentic RL framework built with a unified resource layer that treats each of these environments as a distinct, typed resource scheduled through one engine, with per-tool acquisition for multi-turn reuse, asynchronous backpressure, and rollout versus global-scoped lifecycles. AgentFly adopts a four-layer design: (I) agent layer that abstracts the agent, tool, and reward concepts, decomposing agentic RL into defining agents, tools, and reward functions; (II) rollout layer that composes these into agent loops and computes rewards; (III) context layer that organizes rollouts, injects contextual information, and arranges resources; and (IV) a low-level resource layer that performs resource management. We provide a suite of prebuilt tools and environments, demonstrate successful agent training across multiple tasks and models, and report the first controlled cross-framework throughput comparison against agentic RL frameworks.

cs.AI↗

arXiv2026

The Silk Nebula: A Newly Discovered Giant [OIII] Emission Filament

We report [O\,\textsc{iii}] narrow-band imaging of an extremely large-scale nebula, the Silk Nebula, discovered by the SWONS (Super Wide-Field [O\,\textsc{iii}] Narrow-band Survey), together with multi-fiber medium-resolution spectroscopic data covering the H$α$, [N\,\textsc{ii}], [S\,\textsc{ii}], and [O\,\textsc{iii}] emission lines, acquired with the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) as part of the Medium-Resolution Spectroscopic Survey of Nebulae (MRS-N). Based on 2,684 high-quality spectra, we measured its physical parameters, including relative line intensities, radial velocities (RVs), and full widths at half-maximum (FWHMs). The [S\,\textsc{ii}]/H$α$ ratio of the nebula significantly exceeds the shock-excitation threshold of 0.4. Its location in BPT diagnostic diagrams and spatial morphology of [O\,\textsc{iii}] both support the optical emission is dominated by shock-heated gas. The oxygen abundance is near solar and spatially uniform, consistent with a shock propagating into the ambient interstellar medium. Because [O\,\textsc{iii}] is the only emission line that clearly traces the nebular morphology, we base our distance estimate exclusively on the [O\,\textsc{iii}] kinematics, obtaining a fiducial distance of $\sim$160 pc. At this distance, the nebula has a physical extent of $\sim$28 pc along the north-south direction. We suggest that the Silk Nebula is a large [O\,\textsc{iii}] filament possibly associated with a supernova remnant, although a diffuse-ionized-gas (DIG) or ionized-bubble origin cannot be ruled out with the present data, and one of the largest angular-scale [O\,\textsc{iii}] emission structures documented to date. Future deep X-ray observations and shock-modeling efforts are required to firmly establish its nature.

astro-ph.GA↗

arXiv2026

Gaseous SiO and Silicate Clouds in Three Benchmark Early-L Dwarfs

Tracing silicon from gaseous SiO into silicate condensates provides a direct view of the condensation process that forms silicate clouds in L-dwarf atmospheres. Here we present evidence for gaseous SiO in three benchmark early-L dwarfs, HD 89744B, GJ 1048B, and LSPM J0632+5053B, using JWST/NIRSpec and MIRI spectroscopy. Atmospheric retrievals with and without SiO under multiple cloud prescriptions identify the strongest SiO signatures at 4.0--4.2 and 7.7--8.2 $μ$m, with the local likelihood diagnostic reaching $4.6σ$. In the physical cloud retrievals, the 8 $μ$m band probes lower pressures than the 4 $μ$m overtone, reaching atmospheric layers where equilibrium SiO is more strongly depleted by silicate condensation. All three objects also show compact Fe opacity at pressures of a few bar. Their inferred gas-phase C/O ratios are consistent with expectations from the host-star compositions after accounting for refractory oxygen sequestration. Within the adopted Mie treatment of amorphous silicate grains, the silicate mineralogies favor Mg$_2$SiO$_4$ for HD 89744B and a mixed Mg$_2$SiO$_4$+MgSiO$_3$ cloud for GJ 1048B. Extending this framework to cooler L dwarfs will trace the coupled evolution of gaseous SiO depletion and silicate-cloud formation along the L-dwarf sequence.

astro-ph.SR↗

arXiv2026

RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification

Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification. Existing oversampling methods generate synthetic samples to rebalance class distributions; however, they often produce large numbers of low-quality candidates that distort decision boundaries or introduce artifacts, leading to overfitting and degraded generalization. In this work, we introduce \textbf{RUBRIC}, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem. RUBRIC ranks candidates using a realism-utility trade-off: realism is estimated via a neural density-ratio discriminator from each candidate's resemblance to real minority samples, while utility captures proximity to the decision boundary through a concave, margin-based scoring function $g_-(t)=-\log(1+e^{-t/τ})$. The discriminator uses the same architecture and training protocol on every benchmark and is fit independently to that dataset's real minority class versus its synthetic pool. We show that, under mild regularity conditions, the proposed filtering framework monotonically tightens the generalization bound for margin-based classifiers by jointly reducing distribution shift and suppressing near-negative tail contributions. Through extensive experiments on standard public imbalanced-classification benchmarks, we demonstrate that RUBRIC boosts minority-class recall while preserving overall discriminative ability across multiple data generators. Sensitivity analyses in $λ$ and the selection budget $K$ further characterize performance trade-offs oriented toward ranking quality.

cs.LG↗

arXiv2026

A Recurrent Infrared Modulation in the Warm Post-impact Debris Disk NGC 2547 ID8

Extreme debris disks are signposts of violent rocky collisions, but it remains unclear whether their mid-infrared variability traces transient dust clouds or longer-lived post-impact reservoirs. We analyze a 19 yr mid-infrared record of the young solar analog NGC 2547 ID8, combining four spectra from 2007-2024 with 3-5 micron monitoring through 2026. The optically thin mineral surface remains compositionally stable: broad fits spanning multiple grain sizes give 550-577 K, and a common-basis decomposition with T_surf fixed at 560 K is dominated by 1 micron olivine/Mg-olivine and 0.1 micron forsterite. The W2-equivalent 4.5 micron excess, traced across six recurrent high states, is instead driven mainly by changes in the hot-base continuum component and defines a 956.8 d infrared recurrence timescale with cycle-to-cycle variations in timing and amplitude. This combination supports the interpretation of ID8 as an early post-impact stage in which material on eccentric orbits repeatedly interacts with a warm rocky reservoir, preserving the impact's mineral fingerprint while the interaction geometry evolves. Material on eccentric orbits can persist for decades, and possibly longer, while interacting with debris disks. Longer-baseline, higher-cadence infrared monitoring of ID8 will further constrain the stability of its recurrence timescale and the detailed evolution of its infrared variability.

astro-ph.EP↗

arXiv2026

A Uniform Census of Variable Stars in M101 Based on HST/ACS Photometry

We present a homogeneous HST/ACS F555W/F814W time-series census of variable stars in two M101 disk fields, using a uniform multi-band reduction and reproducible variable-selection workflow for nearby SN-host galaxies. Supplementary archival F555W epochs are used only to refine selected long-period solutions. Candidates are selected with Lomb-Scargle period searches, Fourier light-curve modeling, two-band consistency statistics, artificial-star-test photometric corrections, and mode-aware Cepheid refinement. The final catalog contains 1417 variable-star candidates, including 1102 secure fundamental-mode and 78 secure first-overtone classical Cepheids. Of these secure Cepheids, 420 are newly identified relative to the Shappee & Stanek catalog. The catalog also contains luminous candidates outside the classical Cepheid locus. We release the catalog and source-level two-band light curves for studies of Cepheid physics, stellar variability, and distance-scale applications.

astro-ph.SR↗

arXiv2026

The Intermediate-Mass Black Hole Reverberation Mapping Project: Scientific Overview and Sample Characteristics

Recent discoveries with the James Webb Space Telescope of massive black holes at high redshift have highlighted fundamental questions about black hole seed formation and the coevolution of black holes with their host galaxies. Because the initial seed population cannot yet be observed directly, nearby intermediate-mass black holes provide a complementary fossil record of black hole formation and early growth. Motivated by this opportunity, we present the Intermediate-Mass Black Hole Reverberation Mapping (IMBH-RM) project and construct a homogeneous Sloan Digital Sky Survey sample of active broad-line IMBHs by uniformly reanalyzing literature candidates with consistent spectral decomposition and black hole mass estimation. Our sample contains 192 reliable IMBH candidates at $z\lesssim0.3$ with $\log(M_{\rm BH}/M_\odot)<6$, including four particularly compelling sources with $\log(M_{\rm BH}/M_\odot)<5$. The primary goal of IMBH-RM is to obtain reliable black hole masses from direct measurements and characteristic sizes of the broad-line region and accretion disk for a carefully selected subsample. These measurements will provide robust low-mass anchors for calibrating single-epoch black hole mass estimates and extending black hole--galaxy scaling relations into the IMBH regime. By building a statistically meaningful reverberation-mapped sample spanning $10^4-10^6\,M_\odot$, we aim to constrain the local IMBH mass distribution and place observational constraints on competing black hole seed formation scenarios. The future Multi-Channel Imager aboard the Chinese Space-station Survey Telescope provides a particularly promising platform for achieving these goals.

astro-ph.GA↗

arXiv2026

Differential Reddening and Extinction Law Analyses of Galactic Open Clusters

Extinction significantly affects open cluster parameters and their use in studies of Galactic structure, yet homogeneous large sample measurements of open cluster extinction properties remain limited. Using Gaia-era open cluster member samples combined with multi-band photometry and stellar parameters, we derive color excesses of member stars and provide the homogeneous characterization of the mean reddening, differential reddening, and color excess ratio (CER) at the cluster scale. Differential reddening increases systematically with mean reddening, with highly reddened clusters near the Galactic plane showing stronger extinction variations. Star-by-star reddening corrections narrow color--magnitude diagram (CMD) sequences in 369 of 435 clusters (85%) with reliable CMD-width measurements, and cluster color excess maps reveal small-scale extinction structures. The median CER is compatible with the standard diffuse interstellar medium extinction curve, while the broad CER distribution and its large-scale variations across the Galactic disk likely reflect differences in the dominant dust environments sampled along different Galactic sight lines.

astro-ph.GA↗

arXiv2026

Se-DPO: Self-Evolving Token Credit for Direct Preference Optimization

Direct Preference Optimization (DPO) aggregates token-level log-probability ratios via uniform summation, implicitly treating all tokens as contributing equally to the preference signal. However, the contribution of individual tokens to the preference signal varies. We introduce token credit, which modulates each token's KL regularization based on its contribution to the preference outcome. We derive that effective token credit is proportional to the magnitude of each token's implicit reward, and observe that this quantity evolves substantially during training. This implies that static token credit becomes increasingly misaligned as training progresses. In this work, we propose Se-DPO (Self-Evolving Token Credit for DPO), a live mechanism that derives token credit from the model's own evolving internal signals during DPO training. Since the reward signal varies in reliability across positions, Se-DPO calibrates token credit based on both the strength and the confidence of each token's contribution. Se-DPO requires no external models, adding only a lightweight calibration network with minimal computational overhead. Experiments show that Se-DPO improves over DPO by up to 9.8 points on AlpacaEval~2 and 12.2 points on Arena-Hard.

cs.CL↗

arXiv2026

To Keep or Not to Keep: Learning KV Cache Retention in Disaggregated LLM Serving Systems

Disaggregated \textsc{LLM} serving separates prefill and decode across node pools, placing a network fabric between \textsc{KV}-cache computation and consumption. This breaks a key assumption of classical cache policies: a miss no longer incurs only local recomputation, but also network transfer to the decode node. Likewise, admitting a block to the global cache requires an additional transfer, so poor retention decisions waste both memory and bandwidth. We present \textbf{KVLearn}, a learning-based framework for cost-aware \textsc{KV}-cache retention in disaggregated serving. KVLearn includes: (i) a lightweight Prefix Reuse Predictor (\textsc{PRP}) that estimates reuse probability from structural and temporal prefix features without modifying model weights; (ii) a Cost-Aware Retention Score (\textsc{CARS}) that converts reuse probability into keep/admit decisions using recomputation, transfer, and storage costs; and (iii) an Adaptive Threshold Controller (\textsc{ATC}) that adjusts admission thresholds online from observed hit rates and memory pressure. We integrate KVLearn into a globally disaggregated serving topology and evaluate it on text and multimodal workloads, including large image/video-derived \textsc{KV} blocks with heterogeneous reuse patterns. KVLearn reduces end-to-end time-to-first-token by up to 56% versus No-Cache, 38% versus LRU-Pool, and 33% versus Mooncake-style baselines, while cutting inter-node \textsc{KV} transfer volume by up to 53% versus LRU-Pool. On MM-Session, throughput remains within ${\sim}5%$ of oracle. Code is available at https://github.com/FastLM/KVLearn.

cs.LG↗

arXiv2026

JWST Reveals a Galaxy-Wide Association of Red Supergiants with OB Stars in NGC 5584

The spatial relationship between young and evolved massive-star tracers offers a means of probing recent star formation across galactic disks. We combine JWST/NIRCam, $Swift$/UVOT UVW2, and GALEX FUV/NUV images to investigate the spatial association between red-supergiant (RSG) candidates and UV-selected star-forming complexes (SFCs) in the spiral galaxy NGC 5584. We identify 5,310 RSG candidates using colour--magnitude criteria and define 106 UV-selected SFCs from the UVOT/UVW2 morphology, with their FUV and NUV luminosities measured independently using the GALEX images. We find that 82% of the UV-selected SFCs overlap significant RSG overdensities, indicating a galaxy-wide statistical association between UV-bright SFCs and the evolved massive-star population. Among the 93 SFCs with reliable JWST coverage, the UV-$β$-corrected FUV and NUV luminosities correlate tightly with the JWST/NIRCam luminosities, with Pearson correlation coefficients of $r=0.962$--$0.971$. MIST isochrone comparisons for 43 RSG-rich SFCs indicate a characteristic age of $\log(\mathrm{age/yr})\simeq 7.16$. These results show that RSG overdensities and UV-selected SFCs trace related but distinct phases of recent massive-star formation. Their spatial correspondence is statistical rather than one-to-one and likely reflects differences in stellar age, dust extinction, spatial resolution, aperture coverage, and local star-formation history.

astro-ph.SR↗

arXiv2026

A-SR: Self-Evolving Agentic LLMs for Symbolic Regression via Hierarchical Coordination

Symbolic regression aims to discover closed-form equations from data, but existing LLM-guided methods often rely on a unified proposal loop that compresses heterogeneous search failures into a scalar score and a single prompt. We propose A-SR, a self-evolving agentic framework that shifts the control unit from expression edits to role-conditioned evidence views. A-SR coordinates formula discovery through routing among coordination protocols, an online evaluator-reward role policy, and state-routed process memory. During search, evaluator feedback characterizes reliability and productivity, updates role-level utilities, and routes elite motifs, failure traces, and validity diagnostics to different agents. The framework self-evolves at two timescales: within a run, it adapts the search process without updating LLM parameters; across runs, recorded trajectories can be distilled into open-source LLMs as role-conditioned proposal priors. Averaged over the four LSR-Synth scientific domains in LLM-SRBench, A-SR improves Acc@0.01 over baselines from 25.79% to 48.30% with Llama3.1-8B, while A-SR-LoRA improves the corresponding Qwen3-4B result from 24.58% to 38.29%. On four real-world scientific discovery tasks, A-SR obtains the best in-distribution or out-of-distribution normalized mean squared error on 7 of 8 reported metrics.

cs.CL↗