Search arXiv⌕ Search

SEARCH · Search arXiv

Search Search arXiv

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,495 records · Page 83Linked to original sources
arXiv2026

A Catalog of Spectral Properties for DESI DR1 Quasars

The Dark Energy Spectroscopic Instrument (DESI) is carrying out a five-year spectroscopic survey designed to map the large-scale structure of the Universe and obtain spectra for millions of galaxies and quasars over a wide redshift range. We present a value-added catalog of spectral properties for 1,717,774 unique quasars (1,774,723 spectra) in the DESI Data Release 1. We perform a uniform spectral decomposition of the quasar continuum, host galaxy emission (at $z<1$), and \FeII\ emission, and measure the line properties of major broad and narrow emission lines from Ly$α$ to H$α$ covered by DESI spectra. We also provide single-epoch black hole mass estimates (from H$β$, \MgII, and \CIV), bolometric luminosities, and Eddington ratios for these quasars. We provide systemic redshifts that account for the intrinsic velocity shifts of individual lines; they agree with the DESI pipeline redshifts to within 500~km s$^{-1}$ for 88.6\% of the sample, and are most useful at $z>1$ where the pipeline redshift is heavily constrained by blueshifted high-ionization lines. We also provide local fits to several optical coronal lines, including \NeV~$\lambda3426$, [\ion{Fe}{7}]~$\lambda5722$, [\ion{Fe}{7}]~$\lambda6088$, and [\ion{Fe}{11}]~$\lambda7893$. We show that several emission-line ratios, including those involving coronal lines, vary systematically with Eddington ratio. The catalog is intended as a community resource for studies of quasar demographics, black hole accretion, emission-line physics, and the connection between the broad-, coronal-, and narrow-line regions.

astro-ph.GA↗
arXiv2026

Tandem jets in crossflow: the enclosed-recirculation momentum exchange and the unsteady shear-layer dynamics

Twin tandem jets in crossflow are examined by 40\ kHz time-resolved particle image velocimetry (PIV) over nine spacing--velocity-ratio conditions ($L=2d$--$4d$, $r\approx3$--7; 68 twin cases, 11 single-jet records). The front jet is diluted: its trajectory lies below the single-jet reference, with a near-field segment up to 40\% shorter, and its velocity ratio falls to an effective value $r_{\mathrm{eff}}<r$, while the shielded rear jet rises above the reference. This paper identifies the carrier of that dilution and the unsteady dynamics that sustains it. The rear-jet blockage closes a gap recirculation whose reversed flow puts the two gap-side layers in the counter-current shear-layer configuration, the classical route to enhanced entrainment. The measured Reynolds stress and Prandtl mixing length show that the absolute exchange is nearly independent of the velocity ratio and exceeds the single-jet leeward zone like-for-like by up to $1.5\times$ (zone peak, across conditions) and $2.8\times$ (area-integrated) at $r\approx7$. The gain comes from an enlarged mixing scale rather than from a steeper shear, and the front-jet near-field shortening follows the absolute exchange transport rather than the twin/single enhancement ratio. The turbulent term exceeds the pressure term in eight of the nine conditions, and the flux budget reads the dilution as a limited mass addition plus a momentum loss to the crossflow; the evidence remains observational. The gap-side layers remain convectively unstable---DMD and SPOD find no growing mode within the resolvable band and record length---so the exchange is carried by convecting vortex trains rather than by a self-excited global mode.

physics.flu-dyn↗
arXiv2026

A Resource Allocation Tournament

We introduce and analyze a round-robin simultaneous resource allocation contest, motivated by sports leagues and other competitive environments where agents must distribute limited resources across multiple encounters. Players allocate resources in advance to matches against all opponents, and match outcomes are determined by relative allocations. We study two payoff schemes: total points, where players accumulate rewards for each win, and winner-takes-all, where only the tournament winner is rewarded. For three players, we provide complete equilibrium characterizations under both schemes. For four or more players, we show that the total points payoff admits Blotto-type equilibria, while the winner-takes-all payoff requires numerical and algorithmic analysis. To address this, we develop a best-response dynamics algorithm with stochastic inertia and use it to approximate equilibria in games with four and five players.

math.OC↗
arXiv2026

A hybrid finite-volume scheme with learned rational reconstructions and exact nonlocal fluxes for the space-fractional Burgers equation

On coarse grids, fronts of fractional conservation laws can be narrower than one cell, while the flux through each face depends on the whole domain. For the forced space-fractional Burgers equation, we write the finite-volume defect exactly, with the Riesz flux acting as a nonlocal interface flux. Our proposed closure, NRC, is decoder-only, since the encoder is the exact cell-average projection and a small network only maps cell averages to a sub-cell reconstruction. The decoder adds rational atoms with learned complex poles to a trigonometric histopolant, preserves every cell average, and has closed-form advective and Riesz fluxes. We prove that the scheme conserves mass for every value of the weights and that the learned sub-grid flux has a dipole tail. On 64 cells, NRC reaches a median test error of 0.0308% over three seeds, against 0.431% for WENO5-Z and 0.153% for a black-box flux correction. Ablations show that the exact global operator, the projection and the sub-grid Riesz flux carry most of this gain, and most sub-cell fronts carry a deep learned pole within a quarter cell. NRC keeps its advantage over WENO5-Z under shifts of the order and the forcing and, when trained on several diffusivities, under the tested diffusivity shifts by factors from 0.6 to 2.5 in the median over seeds. From 32 to 128 cells, NRC stays an order of magnitude more accurate than WENO5-Z on the same grid. This gain lies in accuracy per degree of freedom and in an interpretable closure.

math.AP↗
arXiv2026

How Depth Growth and Looping Shape In-Context Learning in Language Models

Looped and depth-grown language models reuse computations across depth and have attracted interest for their reasoning capabilities. Building on prior suggestive evidence of improved in-context-learning (ICL), we investigate how such reuse favors ICL, including how it shapes the acquisition, retention, and mechanisms of this ability. Our analysis spans pretrained language models up to 3B parameters and controlled tasks. The experiments provide evidence for a shared inductive bias toward learning from context: grown and looped models rely more on information in context and, compared with a full-depth dense reference, acquire ICL earlier in controlled tasks and retain it more strongly when learning fixed associations in the weights is also possible. Tracing attention heads through training reveals that depth growth propagates early induction heads across layers. Ablations show that grown and looped models rely more strongly on induction-ranked heads for both few-shot performance and later-token prediction. Together, these findings provide evidence for a shared inductive bias toward learning from context and suggest that (soft) parameter reuse shapes how this ability develops and which attention mechanisms support it.

cs.CL↗
arXiv2026

How Well Do Decision Models Transfer to Vietnamese? An Empirical Study of Text Classification

Recent advances in specialized decision models have expanded the possibilities for zero-shot classification, yet their effectiveness on Vietnamese remains underexplored. This study evaluates four specialized decision models across 13 classification settings spanning six task groups, examining zero-shot performance, sensitivity to label-description language, and robustness to Vietnamese linguistic variations. The evaluation considers English, Vietnamese, and bilingual label descriptions, alongside common linguistic perturbations involving teencode, typographical errors, and diacritic removal. Under English label descriptions, JEV achieves the strongest overall performance, ranking first in 11 classification settings. The effects of label-description language vary across models and datasets, with bilingual descriptions providing no consistent improvement. Robustness experiments reveal that diacritic removal causes the most substantial performance degradation, reducing mean macro-$F_1$ by 9.51--15.47 points across model configurations, considerably more than teencode or typographical errors. These findings highlight the importance of label-description choices and Vietnamese orthographic variation when evaluating specialized decision models for zero-shot text classification.

cs.CL↗
arXiv2026

FocusVLA: Future-Intent Guided Dynamic View Routing for Vision-Language-Action Models

Vision-language-action (VLA) models have become an important paradigm for robotic manipulation by unifying visual perception, language understanding, and action generation. Since reliable manipulation requires sufficient observation of objects, end-effectors, and their spatial relationships, multi-view inputs are widely used to provide complementary spatial information. However, our experiments show that existing VLA models do not fully exploit such multi-view complementarity. The policy tends to persistently rely on the wrist view, lacking the ability to dynamically adjust global-local view dependency across manipulation stages, which limits stage adaptability and robustness. To address this issue, we propose FocusVLA, a stage-aware multi-view routing framework. FocusVLA predicts future interaction intent from historical actions and uses it to guide dynamic view routing during action generation. By adaptively coordinating global and local observations across different task stages, FocusVLA improves multi-view utilization. On RoboTwin 2.0, FocusVLA improves the average success rate over the strongest baseline $π_{0.5}$ by 10.10 pp. In real-world evaluation, FocusVLA improves the success rate from 71.7\% to 85.0\%, achieving a 13.3\% gain over $π_{0.5}$. Moreover, the intent prediction and dynamic routing modules of FocusVLA can be integrated into existing VLA frameworks without modifying their core architectures.

cs.RO↗
arXiv2026

Analytical Modeling of Modulated Heating for Three-Dimensional Anisotropic Optical Thermometry of Multilayered Cylinders

Thermally relevant cylindrical geometries are used in a plethora of applications with thermal property differences depending significantly on material complexity. Manu- facturing processes also induce substantial anisotropy ratios resulting in directionally dependent thermal transport. However, for cylindrical specimens, current thermal metrology resolves only isotropic thermal properties, assumes perfect thermal inter- faces, and offers no formulation for temperature spatial dependence. In this work, we present an analytical framework that models the frequency dependent thermal transport of orthotropic arbitrary layered cylinders to be used as a forward model for existing optical thermometry techniques. This formulation captures the effects of non-negligible thermal boundary conductance and offers the first closed-form solution of a projected Gaussian laser, a common form of heating in thermal metrology, hav- ing a finite elliptic spot size on a curved surface. The novel anisotropic cylindrical model was compared with literature derivations involving two-dimensional isotropic transport for cylinders by applying simplifying assumptions. For single- and two- layerisotropicsamples, thethree-dimensionalmodelagreedwithestablishedsolutions with negligible error, while predictions for orthotropic multilayer samples agreed with finite-element simulations within 1.5% across a range of material properties and laser geometries. Comparison with anisotropic planar models identified conditions under which a planar approximation can reduce computation time. These results estab- lish the first forward model for noncontact characterization of anisotropic thermal properties in arbitrarily layered cylinders.

physics.app-ph↗
arXiv2026

Modulation theory of Y- and X-wave Kadomtsev--Petviashvili solitons

The evolution of long wavelength perturbations of traveling wave 2-soliton solutions of the Kadomtsev-Petviashvili II (KP) equation are investigated using modulation theory. An initial-boundary value problem for a system of hyperbolic conservation laws is formulated for amplitude and slope perturbations to soliton legs of the resonant (Y-shaped) and X-shaped KP traveling wave solutions. The vertex where individuated soliton legs coalesce is modeled as a moving free boundary where information between the soliton legs can pass. The linearized hyperbolic system is proven to be well-posed under suitable initial conditions and is solved exactly. In the case of the resonant Y-shaped soliton, all admissible simple (rarefaction) waves of the nonlinear system are characterized. Specific perturbations of Y-waves can lead to the dimensionality of the modulated solution changing, corresponding to a physical geometric change to the multi-soliton pattern. Spontaneous leg deletion and leg creation can occur. The conditions under which the Y-wave modulation system can experience gradient catastrophe are given. Gradient catastrophe is regularized by the formation of a new leg, with an augmented modulation system proposed to describe the growth of the new leg. On the other hand, the X-wave system can be reduced to two independent line soliton modulation systems. These findings reinforce the prevalence and generality of resonant Y-shaped vertices in the KP II system and applications.The theoretical modulation predictions are supported by direct numerical simulations of KP II, which show good agreement with the analysis.

nlin.PS↗
arXiv2026

Unifying Time-Independent and Time-Dependent Franck--Condon Pictures in Quantum Mechanochemistry via the Linear Coupling Approximation

Quantum chemistry has achieved impressive results in modeling and interpreting mechanochemical transformations at the molecular level. An essential question, the time profile of force application, has however attracted less attention. Here, the Franck--Condon theory of quantum mechanochemistry, previously formulated for the sudden-force limit, is generalized to arbitrary time-dependent force profiles using the linear-coupling approximation. Solving the driven harmonic oscillator yields a closed-form, time-dependent effective Huang--Rhys factor: a natural measure of vibrational excitations upon force application. It is governed by the Fourier transform of the force derivative and recovers the sudden- and fixed-force limits. At the conceptual level, this linear coupling picture links mechanochemistry to vibronic spectroscopy and Marcus electron theory. For a shock-front force ramp, vibrational excitation is shown to be suppressed in frequency, only low-frequency modes being excited under realistic laser-shock-peening conditions.

physics.chem-ph↗
arXiv2026

FastHarness: Experience-Driven Review Scheduling for Vision-Language-Action Policies

Agentic robot systems combine language-model reasoning with robot policies to monitor execution and recover from failures. However, reviewing every action chunk incurs substantial overhead, even in familiar situations. We introduce FastHarness, an experience-driven execution harness that adaptively schedules GPT-6 Astra supervision of a π0.5 policy. A fast path uses stage memory and reviewed outcomes to reduce review frequency as experience accumulates, while a complementary recovery path focuses supervision on correcting errors. Once recovery is verified, routine execution resumes, with failure experience retained to guide future interventions. The policy remains frozen and generates fresh actions from current observations. Across four simulated manipulation tasks, reviewer token use decreases by an average of 57.2% per task from the first evaluated episode to the mean of the final five episodes. Mean episode time decreases from 85.5 to 52.0 minutes over the same interval. On the same tasks, mean success increases from 80% for GPT-as-Policy (π0.5 + GPT-6 Astra) to 97.19% with FastHarness. These results highlight adaptive supervision as a practical route to more efficient agentic robotics without additional model training.

cs.RO↗
arXiv2026

Predictable performance is suboptimal, and what it means for model routing

Large language models (LLMs) differ in capabilities and costs, motivating routing methods that adaptively choose a model for each query. These methods rely on performance prediction: given a query, estimate whether a model will answer it correctly. Most routing methods train separate predictors of candidate models' performance to guide model selection. Our work shows that performance prediction only beats a simple baseline when the candidate model is suboptimal on the task or the performance predictor adds significant capacity relative to the candidate model. Specifically, we take a model's self-confidence, the probability it assigns to its chosen answer, as a baseline and study how much a performance predictor can improve on it. First, we show that any performance predictor that outperforms a model's self-confidence can be used to improve the model's loss: the predictor's advantage transfers exactly in log loss through recalibration. Second, accounting for the predictor's additional parameters, we bound its advantage by the model's suboptimality plus the gain from those parameters. Third, we extend our theoretical results to two-model cascades, which route queries to a more capable model when the first model is predicted to fail. Experiments with 16 language models on four benchmarks show that predictors gain a larger advantage over self-confidence on more suboptimal models. This advantage shrinks as models improve through finetuning. Cascading exhibits the same two patterns: predictor-based routing offers larger gains over confidence-based routing for more suboptimal models, and these gains diminish after finetuning.

cs.LG↗
arXiv2026

PhysTrace: Interpretable Temporal Evaluation of Physical Consistency in Video Generative Models

World action models increasingly rely on pretrained video models assumed to capture physical dynamics for prediction, planning, and action generation. Existing evaluations assess plausibility, probe representations, or measure predictive surprise, but rarely identify which history-conditioned physical expectations conflict with observations. We introduce PhysTrace, a training-free framework aligning MLLM-generated, history-conditioned expectations with observed video windows to temporally localize physical violations and provide interpretable descriptions without task-specific probes or classifiers. We also introduce IntPhys2-Timing, adding violation intervals to IntPhys2. PhysTrace achieves a Hit Rate of up to 70.7% across physical principles on IntPhys2-Timing and outperforms the compared methods in overall possible-impossible discrimination accuracy on IntPhys2. Applied to reconstructed videos along the trajectories of pretrained diffusion/flow-based video models, PhysTrace reveals attribute- and noise-dependent patterns in physical violation localization, enabling interpretable physical consistency analysis across timesteps.

cs.AI↗
arXiv2026

A superconducting LC resonator in a multi-tesla magnetic field for tabletop kHz gravitational-wave detection

We present a concept for a superconducting inductance--capacitance (LC) resonator immersed in a magnetic field for gravitational-wave detection in the kHz band. In the long-wavelength limit, a gravitational wave in a static magnetic field induces an oscillating magnetic-field perturbation proportional to the static magnetic field applied to the resonator and to the gravitational-wave strain. We show that a superconducting pickup circuit can convert this perturbation into a flux signal read out by a superconducting quantum interference device (SQUID)-based or quantum-limited amplifier chain. Inspired by resonant-mass detectors, we develop a three-loop coupled-circuit model and derive the equivalent strain-noise spectrum, including thermal noise, SQUID backaction, and measurement imprecision. Prototype measurements demonstrate a quality factor of $Q\simeq 5\times 10^5$, while $Q=10^7$ is considered as a future development target. Under the assumed flux coupling and an idealized quantum-limited SQUID noise model, optimized circuits yield projected strain sensitivities of order $10^{-21}\,\mathrm{Hz}^{-1/2}$ over $100\,\mathrm{Hz}$ bands in the $1$--$20\,\mathrm{kHz}$ range. Increasing the active high-field volume and improving quantum-limited readout could enable further sensitivity gains. This compact, tunable detector concept could complement interferometric gravitational-wave searches and motivate investigations of high-frequency astrophysical sources.

hep-ex↗
arXiv2026

Dimension-free regimes for unadjusted Langevin in ridge-separable models: A subspace robust Wasserstein analysis

The Unadjusted Langevin Algorithm (ULA) is widely used for high-dimensional Bayesian inference. Classical analyses in Wasserstein-2 distance predict an asymptotic bias and a mixing time that grow with dimension, but this metric is stronger than necessary for many statistical tasks. We instead analyze the bias in subspace robust Wasserstein distance, which controls the mean and arbitrary low-dimensional marginals and is therefore well suited to observables typically estimated in practice. For ridge-separable potentials formed with i.i.d. data, we prove high-probability bounds showing a bell-shaped mixing profile as the number of data points $N$ varies relative to the dimension $d$. In the overparameterized regime, the effective dimension is the rank $N$ of the data-dependent perturbation, leading to complexity depending on $N$ rather than $d$. In the large-sample regime, Gaussian-perturbation estimates yield dimension-free complexity once $N$ is sufficiently large. In contrast, near the proportional regime $\asymp d$, we prove that dimension-free guarantees are impossible in general.

stat.ML↗
arXiv2026

A Sparse Topometric Graph for Scalable Cross-Session Indoor Navigation

A conventional embodied agent in an indoor environment builds a map by accumulating noisy real-world measurements and results from heavy inference to guide navigation. While the detailed measurements may differ due to scene changes or sensor settings, the unique topological structure of navigable passages or the connectivity between rooms may provide training-free guidance that complements local observations. We propose a sparse topometric graph as an efficient structural representation for robust indoor navigation, alleviating the need for fine-grained matching, which is computationally expensive, memory-intensive, and prone to real-world variations. The map consists of nodes that encode local 3D geometry and distance-weighted edges that capture traversable connectivity. A learned graph-matching network finds consistent subgraph structures across sessions by considering both node-level geometric descriptors and neighborhood connectivity. The holistic contextual matching is robust to local ambiguities caused by weak distinctiveness or repetitive structure, and establishes stable correspondences despite local scene changes, different sensor modalities, or limited spatial overlap. We analyze the practicality of the graph through cross-session point-goal navigation, where a goal specified in a prior reference map must be reached from an unknown initial pose in a later session. Experiments show that our method achieves the highest success rates (77.1% on Gibson and 40.0% under 3D-FRONT scene changes), while reducing memory by more than two orders of magnitude compared to point-level representations. The proposed approach further supports cross-session alignment and goal-directed navigation with strong scalability and robust real-world transfer.

cs.RO↗
arXiv2026

Drift Says When, Budget Says How Much: Rate-Controlled KV Re-selection for Long-Form Multimodal Generation

KV cache compression for multimodal LLMs usually selects the visual tokens to keep once, at prefill, and reuses that selection for the whole generation. We show that this one-shot selection is severely limited on long-form generation: on detailed image description, even an oracle that selects with the attention of the uncompressed answer reaches only 0.44-0.52 ROUGE-L agreement with the uncompressed output, because the evidence the model reads moves as the description unfolds. Re-selecting whenever attention drifts out of the kept set is the natural remedy, but we find that a fixed drift threshold fails systematically: drift magnitudes are not comparable across samples, so low-drift samples are almost never refreshed although they benefit the most. We propose Rate-Controlled Refresh (RCR), which lets drift decide when and where to re-select while a per-sample controller regulates how often, combined with a low-rank drift detector, candidate-restricted re-selection, and a decision that stays on the GPU. With one setting for all models and tasks, RCR outperforms the per-step retrieval method InfiniGen by 1.6-3.8 ROUGE-L points on five of six model-task settings with no significant difference on the sixth, while staying within 1.7 points of exact per-step re-selection, with comparable estimated data movement per step and a 4x smaller GPU-resident detector. On DetailCaps, drift-chosen timing beats random timing at the same refresh rate and periodic timing even at a higher rate, by 2.0-5.0 points. On pre-registered video description, where drift is homogeneous across samples, RCR is on par with InfiniGen on Qwen2-VL and exceeds it on Qwen3-VL but no longer beats a fixed threshold at the nearest tested rates, consistent with the diagnosis; object-level evaluation shows that RCR achieves higher object F1 against the uncompressed output than InfiniGen on two of three models.

cs.LG↗
arXiv2026

PerMix: Anomalous sound detection by predicting the perturbation ratio of mixed samples

In anomalous sound detection based on self-supervised learning, the performance of the detection model largely depends on the selection of the pretext task through which the characteristics of normal data are learned. A common pretext task is to mix samples from different classes and classify the mixed sample as its target class with an angular margin loss. However, this supervision becomes mismatched when the target sample is the minority component of the mixture and the margin remains fixed irrespective of the mixing ratio. To address these issues, we propose a pretext task that directly predicts the perturbation ratio of mixed samples by reformulating the logit so that the angular margin is scaled by the perturbation ratio. In addition, we adopt a supervised contrastive loss modified for mixed samples, in which positive pairs are formed according to the dominant class. Experiments show that our approach achieves state-of-the-art results on the DCASE 2020 Task 2 benchmark. Our code is available at https://github.com/kaistss/PerMix.

eess.AS↗