Software Engineering
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 6 November
LLM agents increasingly operate databases on their own: they generate queries and run them against live data. Two checks are commonly used to judge such an agent - whether it produces a valid query, and whether it reaches the right final answer - but neither tells us whether the agent knows what its query will actually do before it runs. We argue that this consequence prediction ability must be evaluated in its own right, and we build MongoBench to measure it: given a collection of documents and a MongoDB find() filter, a model must return the exact set of matching _ids, with ground truth from a real mongod server. The 300 items each compose three or four documented matching pitfalls and are proposed symmetrically by three vendors' models, so results stratify by item source. Evaluating four models - frontier systems from three vendors and one open-weight reasoner - accuracy ranges from 70.1% to 92.1%, a 22-point spread. Each proposer mispredicts 10-31% of the items it wrote itself, so producing a query does not imply predicting its effect. A failure-mode analysis shows errors are better predicted by which semantic pitfall a query exercises - range comparisons and null - than by surface measures such as document count or the number of composed rules. Prepending a verified rule sheet helps unevenly; executing the original filter against the engine yields reliable match sets, but delegation must be checked: on 28% of items (averaged over three runs) one model's executed filters never passed a canonical check against the intended query, and it answered about 22 of them wrong, while the other three passed on every item. We release the benchmark, the mongod oracle, and the evaluation code with all prompts at https://github.com/research-repo-bai/MongoBench.
Combinatorics
Ore-degree conditions for matchings in uniform hypergraphs
Let $\sigma_r(\mathcal{H})$ be the minimum sum of vertex degrees over all nonedges of an $r$-uniform hypergraph $\mathcal{H}$ on $n$ vertices. Balogh, Palmer and Raeisi conjectured that $n>2rs$ and $\sigma_r(\mathcal{H})>r[\binom{n-1}{r-1}-\binom{n-s}{r-1}]$ imply the existence of a matching of size $s$ in $\mathcal{H}$. We prove that, for every fixed $r\ge3$ and all sufficiently large $s$, $\sigma_r(\mathcal{H})>r[\binom{n-1}{r-1}-\binom{n-s}{r-1}]$ forces an $s$-matching whenever $n\ge rs+\lfloor(r-1)^{r-1}/r^{r-2}\rfloor$. A complete $r$-graph on $rs-1$ vertices together with $\lfloor(r-1)^{r-1}/r^{r-2}\rfloor$ isolated vertices shows that the vertex bound is sharp for all sufficiently large $s$. By adding new vertices to $\mathcal{H}$, retaining all its edges, and adding every $r$-set containing at least one new vertex as an edge, we also deduce that $n\ge rs+C$ suffices under the same degree condition for every $s\ge1$, where $C$ is a finite constant depending only on $r$. For $r=3$, we provide the explicit sufficient order $n\ge\lceil(7s+11)/2\rceil$ for $s\ge6$, together with bounds for $2\le s\le5$, establishing the conjecture in this uniformity. For $s=2$, we improve the known sufficient order $2r+2$ to $2r+1$ for every $r\ge3$ and show that this order is best possible.
Read the full abstract
Geometric Topology
Cable Knot Counterexamples to the All-Knot Volume Conjecture
Let $E$ be the figure-eight knot and let $V$ be the hyperbolic volume of its exterior. We study the reduced colored Jones polynomial at $q=e^{2\pi i/N}$ for the $(p,w)$-cable of the connected sum of $d\ge 2$ copies of $E$, where $w\ge 2$ and $p,w$ are coprime. The normalized logarithmic growth rate converges to the JSJ volume $dV$ along colors coprime to $2w$, and its upper limit along colors sharing a divisor with $w$ is at most $V$. Thus the $(1,2)$-cable of $E\#E$ violates the full-sequence volume limit for knots. The $(1,3)$-cable of the same companion shows that convergence can fail even along all odd colors. The separation comes from the multiples of $N$ in the exact cabling sum and a uniform first-derivative estimate for the other companion colors.
Read the full abstract
High Energy Physics (Theory)
Analogue Schwinger Pair Production in Acoustic Black Holes
We show that an analogue Schwinger pair production can be realized in Reissner-Nordström-like acoustic black holes constructed from a Bose-Einstein condensate subject to a synthetic vector potential. Upon Kaluza-Klein reduction of the five-dimensional acoustic geometry, the higher-dimensional phonon modes appear as effectively charged phonons in the resulting four-dimensional spacetime, in which the electric field experienced by the charged phonons arises from the component of the synthetic vector potential along the compactified direction. Then by appropriately choosing the background flow velocity and the synthetic vector potential, the reduced acoustic metric can take the form of a Reissner-Nordström-like black hole. From this metric we obtain the mean number of spontaneously produced phonon pairs in both the extremal and near extremal cases, and show that in the extremal limit this mean number takes the form of the Schwinger formula. Moreover, charged-phonon pair production not only admits the same effective thermal interpretation as Schwinger pair production of real charged particles, but can also be interpreted as the analogue Hawking radiation and superradiance from the five-dimensional perspective. Finally, we make estimates of pair production using experimental parameters in Bose-Einstein condensate, providing theoretical guidance for future laboratory detection.
Read the full abstract
Computation and Language
Does Reasoning Improve Psychological Depth in Large Language Models?
Benchmark gains from reasoning-enhanced large language models encourage the assumption that these models should broadly perform better on complex generative tasks, but this remains underexplored in open-ended, creative domains shaped by nuances of reader experience. We tested this assumption through the lens of psychological depth in short stories. We compared DeepSeek-R1 vs. DeepSeek-V3, our closest available contrast for reasoning-oriented post-training, and GPT-5 vs. GPT-4o, through 60 blinded pairwise evaluations in which seven human annotators and an ensemble LLM-as-a-Judge rated prompt-matched story outputs. Human evaluations showed no universal reasoning advantage: GPT-5 was slightly preferred over GPT-4o (60.0-62.9% across two reasoning-effort settings), whereas DeepSeek-R1 fell behind DeepSeek-V3 (42.9%). Although this pattern was marked by low inter-annotator agreement (Krippendorff's $\alpha = 0.070$), our annotation-quality checks suggested genuine evaluative subjectivity, plausibly reflecting distinct annotator decision schemes rather than annotation failure. In contrast, the LLM-as-a-Judge showed a strong pro-reasoning bias (98.3% of pairs vs. 55.2% of human votes), consistent across its five configurations and partly tracking sentence length. These findings suggest that, at the frontier, reasoning-related quality differences are small enough to be dominated by reader subjectivity, and that LLM-as-a-Judge evaluators can obscure this subjectivity in nuanced creative-generation tasks.
Read the full abstract
Game Theory
Fair Selection of Generalized Nash Equilibria: A Comparability Perspective
Generalized Nash equilibrium (GNE) problems model self-interested agents coupled through their costs and shared constraints, and arise throughout control applications that allocate a scarce common resource. Such problems typically admit a continuum of equilibria, so any implementation must select one. This selection is not neutral: it decides how the cost of the shared constraint is distributed among the agents, a decision often expected to be fair. The variational GNE (v-GNE), which charges all agents a uniform shadow price, is the conventional choice and is frequently advocated as fair. Through the lens of comparability from social choice theory, we show that the GNE set itself requires no inter-agent comparability of costs, whereas the v-GNE implicitly assumes cardinal unit comparability, one of the stronger notions. We then introduce the fair GNE (f-GNE), which selects an equilibrium by explicitly optimizing a fairness metric matched to the comparability at hand, and give constructive routes to compute it for single-constraint and linear-quadratic games. An electric-vehicle charging game serves as a running example throughout this self-contained tutorial.
Read the full abstract
Robotics
Robust Readout of Cross-Point Resistive Arrays via Time-Domain Optimization
Cross-point resistive arrays are attractive for high-density tactile sensing because they reduce wiring complexity. However, conventional frequency-based readout methods can become inaccurate in robotic sensing environments, where long interconnects and measurement and environmental noise degrade the observed signals. We propose a time-domain optimization method for robust readout of cross-point resistive arrays. The method determines resistance values by fitting a time-domain observation model to the measured waveform over the interval, rather than relying solely on spectral amplitudes. By exploiting the full waveform, the proposed method improves robustness under noisy observation conditions. In simulations of a 16 x 16 array, the proposed method reduced the mean readout error by 20% compared with the baseline method. Furthermore, validation over 1,000 trials showed that the proposed method achieves more robust estimation than the conventional method, reducing the mean error and standard deviation by 24.6% and 77.6%, respectively. The improvement was also confirmed in cross-point array simulations with resistance distributions emulating tactile sensor responses.
Read the full abstract
Computer Vision
ProtoCouncil: Multi-Encoder Anatomical Prototypes for Prostate Cancer Detection in Micro-Ultrasound
Purpose: Micro-ultrasound ($\mu$US) is a promising modality for prostate cancer (PCa) detection, but identifying suspicious tissue depends heavily on clinical experience. Models trained on sparse, noisy core-level pathology offer little insight into their decisions and can fail under domain shift. Clinical use needs a traceable decision process and a flag for uncertain predictions. Methods: We introduce ProtoCouncil, in which three foundation encoders (MedSAM, DINOv3 and V-JEPA 2.1) learn case-based prototypes at three anatomical levels: whole gland, needle track and gland sectors. Prototypes are projected onto real training-core regions, and label-free fusion keeps every class logit as a bias plus an exact sum of prototype similarity contributions. Encoder disagreement triggers selective prediction. Results: In five-fold patient-level cross-validation (6,588 cores, 693 patients), ProtoCouncil reaches 82.9% AUROC for clinically significant PCa (csPCa, Grade Group $\geq$3), 4.3 points above the best previously published method. In zero-shot evaluation on two external cohorts, ProtoCouncil achieves higher all-cancer and csPCa AUROC than all previous methods. Nearest same-class exemplars match test-core tumor involvement better than random same-class cases. Externally, csPCa accuracy falls from 79% when encoders agree to 27% when two oppose the fused call. Abstaining on disagreement raises balanced accuracy by 7 and 15 points, while retaining 58% and 54% of the cores. Conclusion: ProtoCouncil grounds core-level predictions in exact similarity-based evidence from real, anatomically indexed training cases, which makes its reasoning interpretable. It achieves the highest csPCa AUROC across three cohorts and identifies unreliable predictions through encoder disagreement.
Read the full abstract
Sound
Real-Time Short-Response Selection for Spoken Dialogue Systems
To realize spoken dialogue systems that can interact naturally with users, it is important for the system to respond to user utterances at appropriate timings. To achieve such natural turn-taking, the system needs not only to predict appropriate response timings but also to mask delays caused by language generation and speech synthesis so that they are not perceived by the user. We propose a real-time short-response selection task that can be integrated with existing turn-taking models to mask generation delay. To address this task, the proposed model learns the mapping between dialogue context and short responses using contrastive learning. When a turn-taking event is detected, the system selects a response from a set of pre-generated candidates. The selected response is played while the main response is being generated. Experimental results show that the proposed method learns to select short responses based on dialogue context, outperforming random and LLM-based baselines. Moreover, integrating response selection maintained turn-taking prediction performance and added minimal processing overhead.
Read the full abstract
General Relativity and Quantum Cosmology
The Weyl Tensor for Gravitational-Wave Causal Sets: Thickness of Geodesic Analogs and Geodesic Deviation
Causal set theory proposes that spacetime is replaced by a partially ordered set (a directed acyclic graph) at the smallest scales. This graph is postulated to produce all the features of continuous geometry above some discreteness scale. We develop an approach to sprinkling causal sets into a linearized plane gravitational wave spacetime, and report on the feasibility of using the analog of geodesic deviation to find a component of the Weyl tensor for such causal sets. We take the collection of all longest chains between a pair of elements to be the discrete analog of the timelike geodesic between them and find through numerical simulations in Minkowski spacetime of various dimensions that the spread or "thickness" of the collection increases with chain length. We estimate, assuming an ideal element-to-element spatial distance estimator and a wave of amplitude $h_0 = 10^{-2}$, that at least $N_{\mathrm{total}} \gtrsim \mathcal{O}(10^8)$ elements are needed to overcome the noise introduced by the thickness and detect the curvature. This is beyond the $\mathcal{O}(10^6)$ elements we can currently simulate. Future work may show ways of extracting more information from the sprinkled causal set and hence lowering the estimate for $N_{\mathrm{total}}$.
Read the full abstract
Computer Vision
OmniLatent: Unified Latent for 3D Part Modeling
3D part decomposition is a central capability for editable asset creation, yet existing workflows typically segment a completed mesh as a post-processing step, making the result sensitive to geometric artifacts, texture ambiguity, and limited part-level annotations. We present OmniLatent, a unified latent modeling framework for 3D part decomposition. Instead of predicting part labels in isolation, OmniLatent learns a shared latent space that jointly encodes geometry, deformation, appearance, and binary part IDs, allowing data-rich geometric and visual priors to support data-scarce part learning. Given a single input image, the base model generates geometry, deformation, texture, and surface part IDs in one pass, producing a part-aware 3D asset without a separate segmentation stage. We further tune the pretrained model for two extensions: a point-prompt adapter that maps arbitrary 3D point prompts with binary part IDs into latent tokens for full segmentation, refinement, and part subdivision, and an amodal part completion model that completes partial surface masks into a coherent full-object decomposition through dynamic per-part token allocation, cross-attention to the given geometry latent, and alternating part-local and global attention. Extensive experiments show that OmniLatent produces semantically meaningful part segmentations with precise boundaries, supports interactive part editing, and completes partial part observations while preserving global object consistency.
Read the full abstract
Multiagent Systems
From Emergence to Reversibility
Multi-agent systems become capable through institutional mechanisms, such as rules that route information to agents that act on it. These mechanisms are also natural places to restrict the system. Yet as agents learn, share memory, delegate, and repair failures, the collective can find other paths to the same effects. A restriction on the original mechanism can then keep operating but no longer stops them, while routine performance and compliance still passes. We propose Reversible Multi-agent Systems as a research agenda: collectives that grow while an authorized supervisor can still restrict specified effects, keep them restricted as agents adapt, and reconstitute useful cooperation under renewed authority. What is reversed is the ability to produce a specified effect; restrictions must hold even if learned knowledge persists. The goal is to preserve effective intervention, not a particular mechanism: at every step of growth, some authorized intervention must keep a specified prohibited effect rare while maintaining required service. We identify three research problems: tracking what capabilities depend on, handing off control before an old mechanism loses its effect, and separating legitimate repair from unauthorized reconstruction. We outline how to evaluate control across growth paths, back control claims with evidence, and allocate the rights to restrict and restore.
Read the full abstract
Numerical Analysis
Equidistribution of Solution Gradients in Physics-Informed Neural Networks via Coordinate Transformations
Physics-Informed Neural Networks (PINNs) often struggle to approximate solutions with sharp gradients due to optimization difficulties and spectral bias. Motivated by the principle of gradient equidistribution from the stretching approach in grid generation, which redistributes solution gradients to reduce functional complexity, we propose Coordinate Transformation PINNs (CTPINNs), which incorporate learnable coordinate transformations to map the physical domain into a computational domain where the physical solutions become approximately affine, making them easier for neural networks to learn. We propose two constructions of the transformation: an explicit construction based on Majorants and an implicit construction based on invertable neural networks. Experimentally, the explicit construction can improve accuracy for PDEs with available explicit coordinate transformations, and the implicit construction can be combined with other advanced PINN models as an efficient plug-in module to improve performance across a range of challenging PDE benchmarks.
Read the full abstract
Machine Learning
StreamTab: Training-Free Temporal Link Prediction with Tabular Foundation Models
Temporal link prediction uses past interactions to predict whether two nodes will interact in the future. Temporal graph neural networks achieve strong results on this task, but they are trained separately for each graph. How a model should use interaction history also depends on the candidates it is compared with. A past interaction can distinguish a recurring link from a random pair that has never interacted, but when the negative candidates have also interacted before, other signals are needed. We propose StreamTab, a training-free framework built on a frozen tabular foundation model pretrained only on synthetic tables. StreamTab describes each candidate pair by statistics of its past interactions and shows the model recent observed interactions and sampled negative pairs as labeled examples, from which the model scores new candidates. Its central design, regime conditioning, samples the negative examples according to how the candidates are sampled, so the examples show the distinction the model must make. As new interactions arrive, StreamTab updates the statistics and refreshes the examples while the model parameters stay fixed. On thirteen datasets against 21 temporal graph baselines, StreamTab achieves the highest average precision on twelve datasets under historical sampling and eleven under inductive sampling, and ranks first or stays within one point of the best on seven datasets under random sampling. With the same features and model, regime conditioning raises average precision over examples with random negatives by a median of 12.6 points under historical sampling.
Read the full abstract
Geometric Topology
Zariski dense hyperbolic 3-manifold groups in $\mathrm{SL}(4,\mathbb{Z})$
We show that $\mathrm{SL}(4,\mathbb{Z})$ contains an infinite family of Zariski-dense subgroups isomorphic to the fundamental group of a fixed closed hyperbolic $3$-manifold, which are pairwise noncommensurable up to conjugacy.
Optimization and Control
Resolvent-Diagonal Energy of Isospectral Jacobi Matrices
We study the minimum squared resolvent diagonal among Jacobi matrices with prescribed eigenvalues. For the zeros of $P_n^{(\alpha,\beta)}$, we construct a canonical realization and bound its asymptotic cost against the optimum. Its rescaled inverse converges to the Green operator of a reflection-symmetric singular Schrödinger operator with Bessel-zero spectrum. A Green-coordinate identity places this construction and all finite realizations in the same density framework. Mass-constrained spectral problems then give computable lower bounds. The bounds meet at $\alpha=1/2$, with optimal energy $2/15$, and have first-order contact with derivative $-13/75$. We classify the constrained extremal profiles and the three endpoint geometries of the canonical density. Two polynomial families with the same limiting edge spectrum have different limiting optimal energies. For Nesterov momentum $(s-1)/(s+1)$, the optimal leading endpoint Hessian-drift gain among nonnegative-link Jacobi realizations is $\sqrt{2}$ times the original gain. An explicit isospectral flow strictly improves the standard one-sided realization for $\alpha>-1/2$, $\alpha\ne1/2$.
Read the full abstract
Dynamical Systems
Asymptotic stability of high-index saddle dynamics
High-index saddle dynamics (HiSD) has been extensively adopted to compute saddle points of arbitrary index, with various applications such as constructing solution landscapes of complex systems. In this work, we prove the asymptotic stability of various HiSD methods, including the idealized, perturbed, and coupled HiSD methods, as well as their momentum-accelerated variants. The analysis is Jacobian-free and thus compensates for the linear stability analysis in the literature. The results not only provide specific characterizations for the convergence rate and the basin of attraction, but also reveal intrinsic correlations among dynamics in the coupled HiSD. Theoretical results are substantiated by numerical experiments.
Read the full abstract
Machine Learning
LeanDojo-v2: A Comprehensive Library for AI-Assisted Theorem Proving in Lean
Building on recent progress in machine learning for formal theorem proving in Lean, we introduce LeanDojo-v2: an end-to-end framework that unifies repository data extraction, LLM fine-tuning, proof search, AI autocompletion, and LLM-assisted theorem proving in a single software library. In addition, we provide Lean4Code, a lightweight IDE that integrates the various features of LeanDojo-v2 through a simple, mathematician-friendly UI. These tools will enable mathematicians and researchers to easily fine-tune language models on domain-specific Lean codebases and experiment with LLM-assisted theorem proving without a complex setup.
Read the full abstract
Computer Vision
Grounding-Aware Judge Reward Reinforcement Learning for Out-of-Domain Face Anti-Spoofing
Face Anti-Spoofing (FAS) must distinguish genuine faces from presentation attacks. A unified Multimodal Large Language Model (MLLM) is an attractive solution: it detects physical attacks (print, replay, mask) in one model and produces interpretable forensic evidence via Chain-of-Thought (CoT). A central obstacle to post-training such a model with reinforcement learning is that FAS reasoning has no rule-verifiable ground truth: only the binary verdict is checkable, while the reasoning chain cannot be verified by any hand-crafted rule. Optimizing the verdict alone is therefore a sparse, hackable signal. The model can reach the correct answer through a shortcut such as a false-rejection bias, which does not generalize. We address this with a grounding-aware judge reward. A 32B vision-language judge scores each reasoning chain against the image on evidence anchoring, logical coherence, and domain knowledge, turning the image itself into the verification signal. The reward is applied in a Group Relative Policy Optimization (GRPO) stage after CoT supervised fine-tuning. It is the judge, rather than the RL objective, that supplies the process supervision a verdict-only reward cannot. Our study starts from a model that already performs well In-Domain (ID) after CoT-based supervised fine-tuning and asks how to improve its Out-of-Domain (OOD) generalization. We find that grounding-aware rewards are an effective means to that end. We validate the method on OOD generalization, where the reward reduces ACER on two physical benchmarks (LCC-FASD 0.288 to 0.273, MSU-MFSD 0.431 to 0.409) while preserving ID accuracy (CelebA-Spoof 0.033 to 0.035). Against ResNet-50/101 and CLIP-ViT-B/16 baselines, the MLLM dominates ID and matches or surpasses the strongest baseline OOD. To our knowledge this is the first use of LLM-as-judge process supervision for FAS.
Read the full abstract
Machine Learning
When Can You Trust Neural Critics? Auditing Mutual Information Estimates of Higher-Order Epistasis in Protein Language Models
Mutual information (MI) estimation is used to probe what neural representations encode, including in protein language models, but variational MI estimators have variance that scales at least exponentially in the true MI. We audit a prior neural-MI study of order-resolved epistasis in ESM-2 on the four-site GB1 landscape. Run-to-run differences show no systematic bias and a magnitude comparable to ordinary cross-seed retraining variability, and the evaluation-pairing scheme shifts scores substantially. At the original dimensionality, two of three alternative estimator families stabilize the estimate across seeds and one does not. In a synthetic calibration at the representation dimensionalities of two other protein language models (a Gaussian channel with known MI; no model representations), all three families recover materially less signal than required, despite correct signal ordering and stable seeds. A control critic given the true signal direction, which also reduces its input to one dimension, recovers 93.3% and 98.7% of the analytic MI at gain 1.5 (at 5,120 and 4,096 dimensions), so the signal is recoverable with privileged information, but the control does not identify why blind estimation fails. We package the checks into a reusable diagnostic protocol and conclude that even a stabilized estimator is not currently sufficient for order-resolved dependence claims made under blind search at this dimensionality and sample size.
Read the full abstract
Robotics
When to Trust the BIM: Identifiability-Aware LiDAR-Inertial–BIM Fusion under As-Built Deviations
Building information models (BIM) can provide a global geometric reference for robot localization with LiDAR-inertial odometry (LIO). However, as-built deviations from the as-designed model mean that a disagreement between a scan and the BIM may arise from a pose error, an offset wall surface (an as-built deviation or an unmodeled object in front of the wall), or a shifted room, and in some geometries these explanations produce identical observations. In this paper, we formalize this attribution problem as fault identification over three fault classes and derive a geometric criterion for when attribution is possible. Building on it, we propose an identifiability-aware method that reports an attribution only when the fault is attributable; otherwise, it retains the candidates, defers BIM updates and BIM-based pose corrections in the ambiguous directions, and recovers once new references arrive. In the pre-registered two-dimensional study, on cases whose fault is detectable but not attributable, the method's conditional error rate was 0.83%, against 31.67% for a baseline with uncertainty-based abstention at the same answer rate. On the 5,600 attributable cases, the method reported a unique attribution 5,574 times, with no incorrect attributions. In paired fault injections in a laboratory room with its IFC model and on three floors of the SLABIM dataset, it raised no off-target attribution (0 of 168 conditions in the laboratory; 0 of 1,302 on SLABIM) and attributed all eight laboratory faults predicted to be attributable, whereas both reproduced baselines raised off-target structural flags in every setting.
Read the full abstract
Astrophysics of Galaxies
Phosphorus-rich star formation in giant molecular clouds hosting primordial globular clusters
Recent observational studies have identified phosphorus-rich (P-rich) stars ([P/Fe]>1.0) in both Galactic globular clusters (GCs) including three confirmed cases and the field population. Using analytical models of GC formation, we investigate a scenario in which P-rich stars form from pristine gas in GC-forming giant molecular clouds (GMCs) polluted by oxygen-neon (ONe) novae within and surrounding the GMCs. Within this framework, we model the chemical abundance patterns of P-rich stars using diverse chemical yield sets for ONe novae while varying the dilution factor (Fdil), defined as the ratio of pristine gas mass to total ONe nova ejecta mass. The principal results of our models are as follows: Across a wide range of Fdil, most models display positive correlations between [P/Fe], [N/Fe], [O/Fe], [Al/Fe]$, [Si/Fe], [S/Fe], and [Si/S]. The [Ce/Fe] ratios in P-rich stars can reach values up to 1 if the GMCs are additionally polluted by interacting low-mass asymptotic giant branch stars.P-rich star formation is highly unlikely in environments with a low interaction frequency between GMCs and ONe novae. P-rich stars exhibit anomalously low 14N/15N, 16O/17O, and 16O/18O isotopic ratios, offering a clear observational diagnostic to test this scenario. Assuming that P-rich Galactic field stars originate from GCs, we discuss the origin of their chemical abundance patterns. We suggest that the existence of P-rich stars provides direct evidence that GC-forming GMCs were polluted by pre-existing stellar populations.
Read the full abstract
Human-Computer Interaction
"Despite her age, Margaret was eager to make a change": Unpacking Aging Representations and Stereotypes in LLM-generated Stories
As large language models (LLMs) increasingly generate everyday stories, their portrayals can shape expectations about aging. Prior work has largely examined age stereotypes in short responses, leaving unanswered whose lives models default to imagining and how age shapes characters' roles in complete narratives. We analyze 594 stories generated by Gemini, GPT, and Llama using text embeddings, LLM-assisted coding, and qualitative analysis. When age was unspecified, stories typically featured younger adults, with middle-aged protagonists largely absent. Older adults were often portrayed through loss, familiar routines, and help from younger characters. Even positive stories sometimes treated older adults' achievements as surprising or portrayed others making decisions on their behalf. We suggest evaluating not only the tone of generated stories, but also who appears, what opportunities they receive, and who makes decisions. Involving older adults in these evaluations can guide writing tools toward more varied, less stereotyped portrayals of later life.
Read the full abstract
Computation and Language
Failure Profiling and Reachable Trajectory Selection for Reasoning Distillation
Training smaller language models on reasoning solutions generated by a larger teacher model is the prevailing approach for transferring mathematical reasoning capabilities. However, the student tends to overfit the output distribution of the teacher and generalizes poorly when its own behavior at inference time deviates from the training data. Recent methods attempt to bridge this gap by incorporating error data, yet the corrective trajectories still reflect the reasoning patterns of the teacher, which the student cannot reliably reproduce, leaving the distribution mismatch largely unresolved. We propose a method, Failure Profiling and Reachable Trajectory Selection (FPRT), built on the insight that effective training data should be jointly determined by what the student fails on and what it can realistically learn. Our method first profiles the failure behavior of the student on each problem through repeated sampling, characterizing it in terms of error rate and error consistency to reveal qualitatively distinct failure modes. Each mode demands a fundamentally different training signal. The diagnosed mode then guides candidate generation from the teacher via distinct prompting strategies tailored to each failure type. Among the correct candidates, our method selects the one whose initial reasoning steps after the point of divergence are most reproducible by the student. These first corrective steps constitute the primary bottleneck, while the remaining steps follow an already corrected context and can therefore be continued more readily. This contrasts with standard reranking approaches, which score the trajectory as a whole and thereby dilute the signal at the critical reasoning transition. Experiments demonstrate that our method achieves the SOTA performance across the five benchmarks on both student backbones.
Read the full abstract
Robotics
DobotTrack: Target-Grounded Vision-Language-Action Learning for Embodied Person Tracking
Language-guided embodied person tracking requires a mobile agent to identify a target from a referring description and consistently follow the target under egocentric observations. We present DobotTrack, a target-grounded vision-language-action (VLA) framework that integrates auxiliary grounding supervision with efficient temporal modeling. We systematically investigate five auxiliary tasks—person segmentation, bounding-box grounding, visual-spatial classification (VSC), future DINO feature prediction, and future-frame prediction—and adopt complementary mask and bounding-box supervision to enhance target delineation and localization. We introduce Q-identity, a target-conditioned historical visual compression module that leverages target-grounding predictions to selectively preserve identity-discriminative appearance features and spatial cues in compact historical tokens. On EVT-Bench, our 0.8B-parameter model achieves an average success rate of 77.9%, exceeding OmTrackVLA by 16.9 percentage points across the three tasks. We further construct a real-world person-tracking dataset collected using a quadruped robot, comprising 126 episodes and 596,323 mask-annotated frames. The target masks enable automatic derivation of bounding boxes and visual-spatial labels encoding horizontal position and apparent image scale, providing additional grounding supervision without separate manual annotation of these labels. Together, our framework and annotation pipeline support learning target-grounded tracking policies from simulated and real-world data. Code is available at https://github.com/Dobot-Team/dobot_track.
Read the full abstract
Information Retrieval
MV-IVF: Multi-Vector Retrieval via Multi-Vector Clustering
Multi-vector retrieval represents each query and document as a set of embeddings and scores relevance via fine-grained set-to-set interactions such as MaxSim (Chamfer similarity). This expressiveness improves retrieval quality over single-vector representations, but makes search substantially more expensive. Existing systems address this challenge primarily through proxy representations or filter-and-rerank pipelines that reduce candidate generation to single-vector search. We study the complementary direction of building a native IVF-style index directly over multi-vector data. We introduce MV-IVF, an inverted-file index for multi-vector retrieval built on MV-Lloyd, a new Lloyd-style algorithm for clustering point clouds under a Chamfer objective. Our key observation is that once point clouds are assigned to clusters, the center-update step reduces to standard single-vector clustering over the flattened embeddings in each cluster. This enables efficient construction of a hierarchical IVF tree with center point clouds at internal nodes. At query time, MV-IVF performs best-first beam search over the tree to identify promising leaf clusters, then scores the documents in those leaves. We further accelerate search using low-bit quantization, query compression, and cache-efficient batched search. We implement MV-IVF in MVSIC, an optimized C++ library and unified evaluation framework for multi-vector retrieval that includes several major approaches. Across BEIR and LoTTE text retrieval benchmarks, MV-IVF achieves strong latency–recall tradeoffs, the best batched throughput frontiers, and favorable build-time and index-size profiles. At $90\%$ Recall-$10$@$10$, MV-IVF achieves up to $8.03\times$ lower single-query latency and up to $173.5\times$ higher QPS than the baselines, demonstrating that native IVF-style indexing is practical and effective for scalable multi-vector retrieval.
Read the full abstract
Solar and Stellar Astrophysics
Dynamical formation of Gaia neutron star binaries in star clusters as a phase-dependent process
Gaia has discovered dormant compact-object binaries in which the compact object shows no observable interaction with its luminous companion. These systems, hereafter Gaia BHs and Gaia NSs for black hole (BH) and neutron star (NS) companions, provide a new probe of compact-object binary formation. These detached systems are difficult to produce through standard isolated binary evolution alone, and their formation pathways remain unclear. Previous $N$-body studies have shown that Gaia BH-like objects are formed efficiently in star clusters, while Gaia NS-like objects are not. This has been attributed to the BH subsystem dominating the cluster core. We performed direct $N$-body simulations of $10^4 \mathrm{M}_\odot$ star clusters in a Galactic tidal field and investigated how NS-main-sequence (NS-MS) binaries form during the late BH-depleted phase. We found that NS-MS binaries satisfying Gaia-sensitive criteria form only after BH depletion. NSs are present in the core throughout the evolution, but the strong encounters there are monopolized by the BH subsystem until it is depleted. Our results show that the dynamical formation of Gaia-sensitive NS-MS systems is not prevented in star clusters, but delayed relative to Gaia BH-like systems.
Read the full abstract
Machine Learning
Function-Space Certification for Neural Operators
Certifying the robustness and physical consistency of scientific machine-learning models ensures that their predictions remain reliable beyond the test set. For neural operators, which learn PDE solution operators by mapping infinite-dimensional function spaces, certification means establishing rigorous, domain-wide guarantees for their output functions. However, standard neural network (NN) verification methods are fundamentally designed for vector-to-vector mappings, and they can only certify neural operators at fixed grid resolutions and provide no guarantees over the entire domain or across another resolution. We close this gap by introducing a function-space certification framework that verifies neural operators across the entire continuous domain. We formulate interval bound propagation (IBP) and CROWN in function space, propagating upper and lower bound functions through each layer using rigorous numerical integration. In addition, we extend bound propagation to higher-order derivatives to certify PDE residuals and solution error bounds. Experiments across several neural operator architectures and scientific problems, including the 572.5-million-parameter spherical Fourier neural operator for global weather forecasting, show that our method produces tight, domain-wide guarantees efficiently, providing a foundation for trustworthy operator learning.
Read the full abstract
High Energy Physics (Phenomenology)
Probing flavor-dependent neutrino-dark matter interactions with the flavor composition of high-energy cosmic neutrinos
High-energy astrophysical neutrinos probe non-standard neutrino interactions over cosmological distances. We study flavor-dependent interactions between neutrinos and dark matter (DM) for neutrinos produced inside the dense DM spike surrounding the supermassive black hole of the blazar TXS~0506+056. A flavor-diagonal potential $V_\alpha=A_\alpha\,\rho_\chi(r)$ induced by coherent forward scattering modifies the mixing in the medium, and the flavor-dependent scattering cross section attenuates the neutrino flux along the line of sight. We show that the propagation is adiabatic throughout the spike and the halo, so that the matter eigenstates produced at the source are conserved and only re-projected onto flavors at Earth. Two regimes emerge. For $A\ll 10^{-38}\,{\rm cm^3}$ the mixing is unaffected and the IceCube three-flavor composition limits the cross section to $\sigma_{\nu\chi}/m_\chi\lesssim 10^{-28}\,{\rm cm^2/GeV}$. For larger $A$ the interacting flavor is isolated into a single matter eigenstate at the source; the two remaining eigenstates escape without attenuation, producing a survival plateau that relaxes the limits by up to four orders of magnitude. The sign of the potential can differ between neutrinos and antineutrinos, so that the isolated state connects to different vacuum mass eigenstates for $\nu$ and $\bar\nu$, which leaves a characteristic imprint on the flavor composition even without attenuation. The requirement that the potential induced by the dark matter density around Earth do not affect the observed PMNS matrix, so that the flavor projection at Earth is performed with the local matter mixing, gives A$\lesssim 10^{-30}\text{cm} ^3$. The combined constraints are presented in the $(A,\sigma_{\nu\chi}/m_\chi)$ plane and compared with a vector-mediator benchmark.
Read the full abstract
Robotics
ToolDex: An Ego-Exo Human Demonstration Dataset for Creative Physical Tool Use
Creative tool use is a fundamental human capability and an important step toward general-purpose embodied intelligence. It requires agents to reason about object affordances, physical constraints, and manipulation strategies, especially when standard tools are unavailable. However, progress in this area is limited by the lack of real-world datasets that capture human tool-use behavior with sufficient scale, fidelity, and diversity. Existing benchmarks are often simulation-based, focus on simplified interactions, or lack detailed observations and annotations. We introduce ToolDex, a large-scale egocentric-exocentric (ego-exo) human demonstration dataset for creative physical tool use. ToolDex captures scenarios in which humans select, adapt, and repurpose available objects as tools. Using synchronized egocentric cameras and eight exocentric cameras, the dataset provides high-fidelity multi-view observations and rich annotations for hands, tools, target objects, and their deformable interactions. Beyond data collection, we establish a benchmark suite characterizing three core capabilities: physical tool-use reasoning, sparse-view 3D and 4D reconstruction, and robot policy learning. ToolDex provides a foundation for studying human tool-use behavior and developing embodied agents that can adaptively use tools in complex real-world environments. Dataset visualizations and benchmarking results are available at https://tooluse-dataset.github.io/.