arXiv subjects

Yue Shen

Publications and source records attributed to Yue Shen.

At least 37 records · Page 2Linked to original sources

arXiv2026

EHR-Complex: Benchmarking Medical Agents for Complex Clinical Reasoning

Clinical agents promise to democratize access to electronic health records (EHRs), yet existing benchmarks fail to reflect the complexity of practical EHR analysis, e.g., often operating on idealized, clean EHRs via static SQL generation rather than interactive execution. In this work, we introduce EHR-Complex, a large-scale benchmark designed for interactive clinical database reasoning. Built on the large MIMIC-IV substrate (365K patients, 31 tables, 500M+ records), EHR-Complex comprises about 52K tasks spanning six clinical intents, supporting both patient-level and population-level queries, where each task requires an agent to interact with a sandboxed environment by executing SQL queries or Python code. Notably, EHR-Complex considers the real-world SQL task complexity for longitudinal multi-table aggregation and compositional reasoning, resulting in 31.93 SQL structural components per query on average. Evaluation results on EHR-Complex reveal the clinical difficulty of these EHR reasoning scenarios, with the top-performing model achieving only 62.3% exact-match accuracy. Pass^k consistency drops below 50% for nearly all evaluated models at k=4, exposing broad stochastic fragility. A fine-grained analysis of more than 3,800 failed trajectories for representative LLMs reveals three dominant failure modes: SQL logic errors, medical-code lookup failures, and semantic misunderstandings. EHR-Complex provides a rigorous testbed for clinical agents and highlights remaining gaps in robust reasoning for large-scale EHR analysis.

cs.AI↗

arXiv2026

NEXUS: Abundance, Environments, and Spectral Diversity of Little Red Dots from the NIRSpec MSA Sample

We present a comprehensive study of Little Red Dots (LRDs) at 2.3 < z < 7.4 using NIRCam photometry and NIRSpec MSA/PRISM spectra from the ongoing NEXUS program. Photometric selection combining several commonly adopted methods yields a high completeness of about 85% for LRD selection over this redshift range and for a flux limit of F444W < 26. The overall purity is about 60%, with contamination from emission-line galaxies and normal active galactic nuclei (AGNs), as well as dwarf stars. Most (>90%) of the spectroscopically confirmed LRDs have robust broad-line detection. Our spectroscopic sample of 36 LRDs displays the full range of spectral diversity of LRDs. It includes objects with extreme Balmer breaks similar to the LRD "Cliff", as well as objects with moderately reddened rest-optical continua that can be fit with low-temperature blackbody components in the recent BH* model framework. The broad H$α$ emission is correlated with the continuum emission at 5100 Angstrom, suggesting common origins for these emission components; the narrow [O III] emission, however, is poorly correlated with the optical continuum. We do not find evidence of redshift evolution in these spectral properties. The space density of LRDs declines toward z about 2, opposite to the trend for normal AGNs, although low-luminosity LRDs at z about 2-4 may be more abundant than currently probed by ground-based searches. The clustering of LRDs suggests that they live in dark matter halos of several times $10^{11}\ h^{-1}$ solar masses, albeit with large uncertainties. Overall, these results are consistent with recent observations of LRDs and with the emerging picture of accreting SMBHs enshrouded in dense gas envelopes as the origin of LRDs.

astro-ph.GA↗

arXiv2026

SlimSearcher: Training Efficiency-Aware Web Agents via Adaptive Reward Gating

Deep research agents have demonstrated remarkable capabilities in complex information-seeking tasks, yet this power comes at a steep computational cost. Driven by accuracy-focused training paradigms, current models adopt brute-force strategies characterized by blind tool dependency and performative reasoning-generating long, redundant trajectories that are far from necessary for resolving these tasks, leading to wasteful tool calls and excessive token consumption. To overcome this efficiency trap, we propose SlimSearcher, a principled framework that pushes the Pareto frontier between accuracy and computational cost across both Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). In the SFT stage, SlimSearcher employs Pareto-efficient filtration to distill trajectories that are both successful and economical, guiding the model toward inherently efficiency-aware search behaviors. During RL, we introduce Adaptive Reward Gating, a dynamic reward-shaping mechanism that evaluates relative tool and token efficiency within a sampled cohort. By cascading these adaptive efficiency metrics with a strict correctness gate, our approach effectively avoids the brevity bias associated with absolute penalties and mitigates reward hacking. Extensive experiments on long-horizon benchmarks, including GAIA, BrowseComp, and XBenchDeepSearch, demonstrate that SlimSearcher reduces average tool-call rounds by 17%-58% while maintaining or improving accuracy.

cs.LG↗

arXiv2026

Varstrometry for Off-nucleus and Dual Subkiloparsec AGN (VODKA): Three Quadruply Lensed Quasars at Cosmic Noon in HST and JWST

We present results from imaging observations of three quadruply lensed quasars by Hubble Space Telescope (HST) and James Webb Space Telescope (JWST) at redshifts $z = 2.550$, 2.975, and 1.500. We model our targets assuming a singular isothermal ellipsoid mass profile and an elliptical SÃ\c{opyright}rsic profile for the lensing galaxies, and reconstruct the geometric configuration of each system with measured Einstein radii of 0.44$'$, 0.58$'$, and 0.49$'$. While no spectroscopic measurements are available for the lenses, we constrain the redshift of each lens to $0.5 < z < 1.2$, $1.0 < z < 1.5$, and $0.4 < z < 0.9$. For all three lenses, the best-fit light model yield a typical de Vaucouleurs $n_{\rm S\acute{e}rsic} \sim 4$ profile and an effective radius $R_e$ around $\sim 1.5 - 3.5$ kpc. We accordingly classify the three lenses as early-type galaxies at an intermediate to high redshift, a common type for strong lensing galaxies. Compared to other known quadruple lenses, the lensing galaxies in this work are at the lower end of the distribution of Einstein radii and upper end of the distribution of the lens redshifts. They represent an interesting quadrant of subarcsecond-separation lenses in the population of single-galaxy strong lensing which have been largely unexplored yet and will be great targets of interest in upcoming high-resolution lensing surveys.

astro-ph.GA↗

arXiv2026

A New Member of the Fast and Furious Family: A Relativistic and Time-Variable UV Outflow in a Luminous Quasar

We report the fastest quasar outflow first detected in the ultraviolet, via variable C IV and Si IV absorption at outflow velocities $-77,000$ km s$^{-1}$ to at least $-90,000$ km s$^{-1}$, in the radio-quiet quasar SDSS J231854.31+243954.2 (J2318). J2318 is a weak-lined quasar in the rest-frame ultraviolet, but Gemini GNIRS spectroscopy reveals an H$α$ redshift of $z=2.6781\pm0.0004$. A twenty-year photometric time series shows peak-to-peak variability of 0.5 mag in the $g$ band. The C IV outflow strengthened monotonically over three epochs spanning $\sim$2.2 rest-frame years. The existence of such a high-velocity outflow implies that models of quasar outflows must be able to either accelerate gas to $0.3c$ while still preserving C IV and Si IV ions, or enable the formation of C IV and Si IV ions in gas which has been accelerated to $0.3c$. Virial estimates reveal a black-hole mass of $1.65\times10^9~M_\odot$, which leads to an Eddington luminosity and Eddington ratio of $2.4\times10^{47}$ erg s$^{-1}$ and $0.45$, respectively. Using very conservative assumptions, the UV-absorbing outflow alone has an estimated mass loss of $>0.82~M_\odot~{\rm yr}^{-1}$ and a kinetic luminosity ratio $L_{kin}/L_{bol}\geq0.75$%. The lower limit is just above the threshold usually cited for significant feedback on the host galaxy. Comparison to PDS 456, the only other known quasar with a UV-absorbing outflow at $0.3c$, suggests that the true $\dot{M}$ and $L_{kin}/L_{bol}$ could be up to two orders of magnitude larger.

astro-ph.GA↗

arXiv2026

Varstrometry for Off-nucleus and Dual sub-Kpc AGN (VODKA): Radio Classification of High-Redshift Dual AGN Candidates with the Very Large Array

Dual active galactic nuclei (dual AGNs) are pairs of simultaneously accreting supermassive black holes in merging galaxies. We investigate dual AGNs to understand whether merger-induced accretion is a significant growth mechanism for supermassive black holes. Searching for such systems is favorable at close separations and high redshift (Cosmic Noon, $z \sim 2$) due to the expected combination of high galaxy merger rate and peak AGN activity which characterize this era of the Universe. The sample of nine dual AGN candidates is selected based on resolved optical dual detections with Gaia, whose angular separations are less than 1$' '$ and redshifts range between 1.5 and 2.8. Each pair is spatially coincident with a Sloan Digital Sky Survey quasar primary target. We aim to classify the secondary targets and other components in the radio regime using 2-band Very Large Array imaging (C and Ku-bands) to test for dual AGN presence. We identify two dual AGNs and three quadruply imaged gravitational lens AGNs, two out of which show evidence of radio flux anomaly. The two new dual AGNs add to the limited census of confirmed kpc-scale pairs at Cosmic Noon, while the radio flux anomalies in J0911+0550 and J1118+0745 provide independent evidence for substructure in their lensing potentials, consistent with microlensing on compact AGN emission regions. Besides one confirmed AGN-star pair, three candidates remain unclassified due to lack of radio detection for one or two components.

astro-ph.GA↗

arXiv2026

Understanding the Broad-line Region of Active Galactic Nuclei with Photoionization. II. Slim disks, Self-shadowing, and BLR sizes

Reverberation-mapping (RM) measurements have revealed that high-accretion-rate active galactic nuclei (AGNs) systematically lie below the canonical broad-line region (BLR) radius - optical continuum luminosity (R-L) relation, exhibiting shorter lags than predicted for fixed 5100Åluminosity. The physical origin of these offsets remains debated. We investigate how accretion-flow structure and BLR cloud properties affect the emissivity-weighted BLR radius using analytic slim-disk SEDs and photoionization calculations on a two-dimensional axisymmetric grid. As the accretion rate approaches and exceeds the Eddington limit, geometric thickening of the inner disk produces anisotropic illumination and self-shadowing, reducing ionizing flux seen by low-latitude BLR clouds and flattening the R-L relation at high L/LEdd. Self-shadowing at high accretion rates reproduces the observed R-L trend in the RM AGN sample reasonably well, but this effect alone is insufficient to explain the observed lag offset in low-mass ($\sim10^{7}M_\odot$) systems with high accretion rates. Motivated by accretion-disk density scalings, we further explore models in which the BLR gas density increases toward lower black hole mass or higher accretion rate. We find that an accretion-rate-dependent BLR density enhancement further improves agreement with observed RM data, where the BLR gas density increases by a factor of 3-5 for one dex increase in $\dot{m}$. Variations in BLR opening angles produce a less important effect on BLR sizes. These results demonstrate that self-consistent modeling of accretion disk SED, BLR illumination and photoionization, and gas density variations can fully explain the observed distribution of AGNs in the BLR size - optical luminosity plane. This framework provides a physically motivated link between accretion-flow structure and BLR observables across a broad range of black-hole properties.

astro-ph.GA↗

arXiv2026

LATTE: Forecasting Peer Anchored Preference Trajectories for Personalized LLM Generation

Personalized generation with frozen large language models requires a conditioning signal that is both compact and current. Existing personalization methods typically retrieve or summarize user histories in text, or compress them into static latent profiles and soft prompts. These approaches are efficient, but they treat a user's past behavior as an aggregate profile and therefore mix stable identity, recent drift, and item content in the same representation. We propose LAtent Trajectory Tracking and Extrapolation (LATTE), a framework that represents personalization as forecasting a peer anchored relative preference state. For each historical session, LATTE subtracts a time masked baseline formed from comparable users who responded to the same item, producing a state that measures how the target user differs from peers under a shared item context. A lightweight sequence predictor then forecasts the next state in this trajectory, and a State to Token Bridge injects the forecast into a frozen instruction tuned LLM through a single anchored soft token. We provide a latent factor analysis showing when peer anchoring cancels shared item variation and why temporal forecasting trades off stale averages against noisy recent states. Experiments on Amazon Reviews 2023 and MemoryCD show that LATTE consistently outperforms retrieval, summary memory, static latent profiles, difference aware latent profiles, and soft prompt compression baselines. On Amazon Reviews 2023, LATTE improves average ROUGE-L from 0.219 for a static latent profile and 0.245 for the strongest added latent compression baseline to 0.259. Additional pairwise comparisons and diagnostic analyses suggest that the improvement is mainly due to forecasting user-specific trajectory information, rather than merely adding a soft prompt interface.

cs.CL↗

arXiv2026

Changing-Look Active Galactic Nuclei in SDSS-V: Host-Galaxy Properties and Black-Hole Scaling Relations

Changing-look active galactic nuclei (CL-AGNs) exhibit dramatic spectral variability on unexpectedly short timescales, challenging standard accretion flow models. Despite growing samples, the physical drivers of this extreme variability, and the potential link to host-galaxy properties, remain unknown. Regardless of the underlying mechanism, the transition between AGN-dominated and host-dominated spectra offers a unique opportunity to study relations between AGNs and their hosts within the same objects. We present intermediate-resolution spectroscopy of 23 CL-AGNs identified by the Sloan Digital Sky Survey V (SDSS-V), obtained with the Very Large Telescope/X-shooter and Gemini-N/GMOS. An analysis of the Mgii emission line observed in the spectra demonstrates that the majority of these sources cannot be driven by variable obscuration. Our CL-AGNs roughly follow the M_BH-sigma_* and M_BH-M_* relations of inactive galaxies, with a median black hole-to-stellar mass ratio of 0.38%. We find no evidence that the stellar population properties of our CL-AGNs, including stellar mass, age, young stellar fraction, and star-formation rate, differ from those of type 2 AGNs in SDSS. These results suggest that CL-AGNs reside in typical AGN host galaxies and that their extreme variability is likely unrelated to host-galaxy environment, supporting the idea that CL-AGNs are not a distinct population, but rather represent a phase of normal AGN activity. This result, in turn, implies that CL-AGNs can serve as useful probes of the AGN-host connection, providing access to both AGN-dominated and host-dominated spectra of the same systems.

astro-ph.GA↗

arXiv2026

WebClipper: Efficient Evolution of Web Agents with Graph-based Trajectory Pruning

Deep Research systems based on web agents have shown strong potential in solving complex information-seeking tasks, yet their search efficiency remains underexplored. We observe that many state-of-the-art open-source web agents rely on long tool-call trajectories with cyclic reasoning loops and exploration of unproductive branches. To address this, we propose WebClipper, a framework that compresses web agent trajectories via graph-based pruning. Concretely, we model the agent's search process as a state graph and cast trajectory optimization as a minimum-necessary Directed Acyclic Graph (DAG) mining problem, yielding pruned trajectories that preserve essential reasoning while eliminating redundant steps. Continued training on these refined trajectories enables the agent to evolve toward more efficient search patterns and reduces tool-call rounds by about 20% while improving accuracy. Furthermore, we introduce a new metric called F-AE Score to measure the model's overall performance in balancing accuracy and efficiency. Experiments demonstrate that WebClipper compresses tool-call rounds under excellent performance, providing practical insight into balancing effectiveness and efficiency in web agent design.

cs.AI↗

arXiv2026

NEXUS: Quick Release Notes

NEXUS is a JWST Multi-Cycle (Cycles 3-5) GO Treasury imaging and spectroscopic survey around the North Ecliptic Pole during 2024-2028. It contains two overlapping tiers in depth and area coverage. The Wide tier ($\sim 400~{\rm arcmin}^2$) performs NIRCam/WFSS 2.4-5 $μ$m grism spectroscopy with three annual epochs over 3 years (final spectral continuum ${\rm S/N/pixel>3}$ at F444W $<22.2$), accompanied by NIRCam multi-band imaging in F090W, F115W, F150W, F200W, F356W and F444W. The Deep tier ($\sim 50~{\rm arcmin}^2$) performs high-multiplexing NIRSpec 0.54-5.5 $μ$m MOS/PRISM spectroscopy for ~10,000 targets in total, over 18 epochs with a 2-month cadence, along with F200W+F444W NIRCam imaging for each epoch. Parallel imaging observations with MIRI and additional NIRCam filters are also performed within the Wide and Deep tiers. The primary data covering the Deep tier (including NIRCam imaging, NIRSpec/MSA spectra, and vetted MSA spectroscopic redshifts) are released in regular Quick Data Releases to facilitate follow-up studies. This evolving document describes the MSA targeting information and observing status for each of the 18 Deep epochs, which started in May 2025 and continue on the regular 2-month cadence. We also describe the content and caveats of the quick release data and report selected cases of diverse scientific interests.

astro-ph.IM↗

arXiv2026

Extreme Galaxy-scale Outflows Are Frequent among Luminous Early Quasars

The existence of abundant post-starburst/quiescent galaxies just $\sim$1-2 Gyrs after the Big Bang challenges our current paradigm of galaxy evolution. Cosmological simulations suggest that quasar feedback is likely the most promising mechanism responsible for such rapid quenching. Here we report a high detection rate (6/27) of exceptionally fast and powerful galaxy-scale outflows traced by [O III] emission in z $\sim$ 5-6 luminous quasars as revealed by the James Webb Space Telescope (JWST), with velocity up to $\sim$8400 km s$^{-1}$ and order-of-magnitude kinetic energy outflow rates up to $\sim$260% the observed quasar bolometric luminosities. This fraction is $>$3.9 and $\sim$8.8 times of those in comparison samples at z $\sim$ 1.5-3.5 and z $<$ 1, respectively. These extreme outflows are comparable to or even faster than the most rapid [O III] outflows reported at z $\lesssim$ 3, and could reach the circumgalactic medium (CGM) or even the intergalactic medium (IGM). The average kinetic energy outflow rate of our sample is more than 2 dex higher than those of the lower-redshift comparison samples. The substantially higher frequency of outflows with energetics well above the threshold for negative feedback in our sample strongly suggests that quasar feedback plays a significant role in efficiently quenching/regulating early massive galaxies.

astro-ph.GA↗

arXiv2026

GroupRank: A Groupwise Paradigm for Effective and Efficient Passage Reranking with LLMs

Large Language Models (LLMs) have emerged as powerful tools for passage reranking in information retrieval, leveraging their superior reasoning capabilities to address the limitations of conventional models on complex queries. However, current LLM-based reranking paradigms are fundamentally constrained by an efficiency-accuracy trade-off: (1) pointwise methods are efficient but ignore inter-document comparison, yielding suboptimal accuracy; (2) listwise methods capture global context but suffer from context-window constraints and prohibitive inference latency. To address these issues, we propose GroupRank, a novel paradigm that balances flexibility and context awareness. To unlock the full potential of groupwise reranking, we propose an answer-free data synthesis pipeline that fuses local pointwise signals with global listwise rankings. These samples facilitate supervised fine-tuning and reinforcement learning, with the latter guided by a specialized group-ranking reward comprising ranking-utility and group-alignment. These complementary components synergistically optimize document ordering and score calibration to reflect intrinsic query-document relevance. Experimental results show GroupRank achieves a state-of-the-art 65.2 NDCG@10 on BRIGHT and surpasses baselines by 2.1 points on R2MED, while delivering a 6.4$\times$ inference speedup.

cs.IR↗

arXiv2026

CIV wind properties of the SDSS-V X-ray selected quasars: strong optical-to-UV emission is key regardless of X-ray strength

We present an investigation of the rest-frame optical/UV and X-ray properties for a sample of 3027 X-ray selected quasars between $1.5 \leq z \leq 3.5$ detected in the deepest Spectrum Roentgen Gamma/eROSITA data available and observed by the fifth iteration of the Sloan Digital Sky Survey (SDSS-V). We parametrize the CIV$\lambda1549$ emission line to infer the strength of accretion disc winds and perform X-ray spectral fitting. The X-ray spectral properties -- namely, the 2keV monochromatic luminosity (L$_\text{2keV}$) and spectral slope -- are not strongly correlated with wind strength. Despite this result, the X-ray selected sample is shifted towards lower CIV blueshifts and higher equivalent widths than the optically selected sample observed in previous SDSS surveys, and matching in optical luminosity, redshift, and Eddington ratio does not reduce these differences. We estimate the far-UV luminosity using the HeII$\lambda1640$ line luminosity and define the slopes between this and the 2500A monochromatic luminosity ($L_{2500}$) and L$_\text{2keV}$ ($α_\text{ouv}$ and $α_\text{uvx}$, respectively) in a similar manner to the familiar $α_\text{ox}$ parameter, which tracks the spectral slope between $L_{2500}$ and L$_\text{2keV}$. The quantity $α_\text{ouv}$ is more strongly correlated with wind strength in our sample than $α_\text{ox}$. We show that the correlation between $α_\text{ox}$ and wind strength is driven by the relationship between the optical luminosity and wind strength. Our results are consistent with a radiation line-driven wind, whereby the ionising far-UV photons must not over-ionise the gas. The hard X-ray photons are few enough in number to have a negligible effect on the ionisation state of the material.

astro-ph.GA↗

arXiv2026

DeepDISC-Euclid: Source Classification and Photometric Redshifts in Euclid Deep Field North With a Pixel-Level Deep Learning Approach

The first Euclid Quick Data Release (Q1) provides extensive imaging and spectroscopic data for hundreds of millions of photometric objects across several deep fields. Accurate classifications and photometric redshifts (photo-z) for these sources are crucial to maximizing the value of these data. In this work, we perform source classification and photo-z estimation for the Euclid Deep Field North (EDF-N) around the North Ecliptic Pole, using a deep learning framework (DeepDISC) that learns and infers using 9-band images simultaneously. We train three dedicated models for (1) source detection and classification, (2) galaxy photo-z, and (3) quasar photo-z. The Euclid Q1 input source catalog, and classifications and spectroscopic redshifts (spec-z) from the Dark Energy Spectroscopic Instrument Data Release 1 are adopted as our training data. DeepDISC source detection achieves overall completeness of ~93% and purity of ~80% if using the Euclid source catalog as the ground truth. Using a JWST source catalog within EDF-N as the reference, we estimate a true purity of ~ 90% for DeepDISC sources. About 99.2%, 99.0%, and 84.8% of stars, galaxies, and quasars, respectively, are correctly recovered with their spectroscopic classifications. The DeepDISC photo-zs show good agreement with spectroscopic redshifts, for both galaxies and quasars. Comparisons with other Euclid Q1 products demonstrate that DeepDISC provides comparable or improved performance in source detection/deblending, classification and photo-z, especially for quasars. These results demonstrate the potential of pixel-level deep learning approaches for large-scale sky surveys such as Euclid and Roman, which will continue to improve with better training labels. We release the full DeepDISC source catalog (~13 million objects) for EDF-N with classifications and photo-zs, including photo-z probability distributions.

astro-ph.GA↗

arXiv2026

SLSREC: Self-Supervised Contrastive Learning for Adaptive Fusion of Long- and Short-Term User Interests

User interests typically encompass both long-term preferences and short-term intentions, reflecting the dynamic nature of user behaviors across different timeframes. The uneven temporal distribution of user interactions highlights the evolving patterns of interests, making it challenging to accurately capture shifts in interests using comprehensive historical behaviors. To address this, we propose SLSRec, a novel Session-based model with the fusion of Long- and Short-term Recommendations that effectively captures the temporal dynamics of user interests by segmenting historical behaviors over time. Unlike conventional models that combine long- and short-term user interests into a single representation, compromising recommendation accuracy, SLSRec utilizes a self-supervised learning framework to disentangle these two types of interests. A contrastive learning strategy is introduced to ensure accurate calibration of long- and short-term interest representations. Additionally, an attention-based fusion network is designed to adaptively aggregate interest representations, optimizing their integration to enhance recommendation performance. Extensive experiments on three public benchmark datasets demonstrate that SLSRec consistently outperforms state-of-the-art models while exhibiting superior robustness across various scenarios.We will release all source code upon acceptance.

cs.IR↗

arXiv2026

DIVER: A Multi-Stage Approach for Reasoning-intensive Information Retrieval

Retrieval-augmented generation has achieved strong performance on knowledge-intensive tasks where query-document relevance can be identified through direct lexical or semantic matches. However, many real-world queries involve abstract reasoning, analogical thinking, or multi-step inference, which existing retrievers often struggle to capture. To address this challenge, we present DIVER, a retrieval pipeline designed for reasoning-intensive information retrieval. It consists of four components. The document preprocessing stage enhances readability and preserves content by cleaning noisy texts and segmenting long documents. The query expansion stage leverages large language models to iteratively refine user queries with explicit reasoning and evidence from retrieved documents. The retrieval stage employs a model fine-tuned on synthetic data spanning medical and mathematical domains, along with hard negatives, enabling effective handling of reasoning-intensive queries. Finally, the reranking stage combines pointwise and listwise strategies to produce both fine-grained and globally consistent rankings. On the BRIGHT benchmark, DIVER achieves state-of-the-art nDCG@10 scores of 46.8 overall and 31.9 on original queries, consistently outperforming competitive reasoning-aware models. These results demonstrate the effectiveness of reasoning-aware retrieval strategies in complex real-world tasks.

cs.IR↗

arXiv2026

A large population of over-massive black hole quasars at z=0.3-0.8 revealed by eROSITA

In most galaxies, the central black hole accounts for no more than a percent of the total mass in stars. Recently, however, extremely over-massive black holes with ratios of 10% have been reported in dwarf galaxies at z<1 and at cosmic dawn (z>5.5) by JWST. Both findings have been interpreted as signatures of the still mysterious origins of super-massive black holes, such that most of the black hole mass was built at birth rather than through black hole accretion. Here we show that among evolved galaxies over-massive black holes are also present, indicating that overmassive BHs are not a signature unique to black hole formation channels. The first large-area sky survey of the eROSITA X-ray telescope on board SpectrRG identified 200 quasars by their luminous hard X-ray radiation. These signpost rapidly growing black holes. Complementary optical spectroscopy from the Sloan Digital Sky Survey and archival UV to IR photometric data combined with galaxy-quasar decomposition techniques allow us unbiased estimates of cosmological distances, black hole masses and host galaxy stellar masses. We securely identify a sample of over-massive black holes with BH-to-host ratios of more than 5%, which may have undergone exponential accretion spurts lasting about a billion years. Our survey identified a high space density of at least 4/Gpc^3 of overmassive black holes near cosmic noon. This indicates an accretion channel disconnected from the stellar population that cause strong deviations from galaxy scaling relations. This channel is currently not part of galaxy evolution models. The identified channel, if applicable also for the first billion years of cosmic time, can explain JWST AGN without requiring them to signify imprints of black hole seeding mechanism.

astro-ph.HE↗